Many of the early analog electronic greats were inveterate pranksters. Sometimes the prank made the papers, as when Bob Widlar bought a sheep and staked him on the lawn at National Semiconductor (Fig. 1). National had announced it was cutting back on lawn maintenance to save money, and Widlar thought he should make a statement regarding excessive cost cutting. National was founded by Charlie Spork, among others. Charlie was the manufacturing guy at Fairchild, so he tended to have a factory sensibility. You don’t need a fancy lawn to manufacture semiconductors.

1. Bob Widlar poses with the sheep he staked to the lawn at National Semiconductor. His Mercedes is in the background. He transported the sheep in the back seat the 30 or so miles from Morgan Hill. (Courtesy of Fran Hoffart)

Pierre Lamond, a manager at National, also had that manufacturing mentality. He was notorious for expecting punctuality of his engineers, as if they should punch a clock in the morning. He used to stand at the entrance to the campus and take names of anyone who dared be even a few minutes late. A friend told me that the IC designers were so upset at this treatment that they all threatened to walk out en masse if Lamond kept up his strict tardiness rules.

It was around this time that analog geniuses Jim Williams and Bob Widlar figured out how to make a wall socket that converted the 60-Hz line frequency to something less. They got Widlar’s brother, Jim (Fig. 2), to install the socket where Lamond had his desk clock plugged in. Since the clock ran slow, it would make Lamond late to appointments and such. Williams told me the whole floor could hear Lamond’s shouts when he discovered the subterfuge.

2. Bob Widlar’s brother Jim, at the 2018 Analog Aficionados party in Silicon Valley.

April Fool’s

With the prankster nature of analog folks, its no wonder that April 1^st is a high-holy day. I have a pdf copy of what looks suspiciously similar to a Linear Technology design newsletter issue. The logo is similar, but the company name is “Unreal Technology.” Dated April 1, 2003, VOLUME XIII NUMBER 1.5, the issue has topics such as “SMBus/I²C Clarifier Improves Quality of Serial Bus Data.” The part supposedly can take text with gibberish and convert it to a Joyce Kilmer poem. The article is written by Dave “Smoke and Mirrors” Dwelley.

Jim Williams has an article, too, “A Coprossessor-Based Fast Logarithmic Computer.” The logarithms are created by the well-known trick of putting a diode in the feedback of an op amp. The coprocessor is a Z-80 microcontroller hooked up by its power and ground pins to reverse-bias them, so the whole 8,500-transistor chip serves as the diode. The newsletter had other great spoofs, like a quad temperature sensor where all four sensors were in the chip die, a pretty useless feature.

Bob Pease had his own joys on April 1^st. Back in October of 2010, Pease wrote me about one of his April Fools’ pranks. Pease remembered.

“Here at NSC (National Semiconductor), we took a few hundred dead TO-3 can regulators, and I cut the leads off, and had them silkscreened as BD-1. Bob Dobkin invented the BD-1, but I got them screened up. The BD-1 got some publicity in EEE magazine. It's a Battery Discharger, BD-1. If you bolt onto each end, it can carry about 900 amps. Maybe not 1000. Good for discharging large batteries. After the publicity came out, our marketing guy for regulators, Roy Essex, was furious that he could not find any info on this new IC. Some dumb customer had been calling him, and asking why he was so ignorant that he could not get him more info on this new NSC circuit. He was highly annoyed.

“I left a dozen BD-1's on Roy's desk, explaining that any press release dated 1 April, and published in a magazine on April 1, should be taken very lightly!! His furious comments increased, for a while. I bet he is still pissed off at us. Beast [sic] regards. / rap / It's past noon. I need a drink!”

This discussion was part of an email where he described silkscreening some parts that had an obscene acronym aimed at a competitor. He sent those parts to engineers at the competitor, and let them puzzle over the part until one of them pronounced it out loud, when its profane intent became very clear. Pease remarked, “The guys making silkscreens were usually agreeable to my insane requests.”

WOM-Batty

Creating absurd and phony parts were another trait of electrical engineers. They include the Signetics WOM, a Write-Only Memory. Pease noted that it “was professionally presented in full color in a four-page foldout section of electronics magazines.” You can still find the 1972 datasheet online. John G. (Jack) Curtis, the perpetrator of the prank, has a website describing the WOM in exquisite detail. He takes Pease to task for an unreliable account of the creation story of the WOM, but all Pease had to go on was industry hearsay, there were no websites when Pease would describe the WOM.

What is delightful is that the WOM was partially inspired by the Umac 606 (Fig. 3). Well-known to Ham enthusiasts, this was a spoof datasheet for a vacuum tube that the EIMAC division of Varian published in 1950. The datasheet begins, “The Umac 606 is an infernal anode, helical beamed phantasatron having a dissipation rating. The unique vacuum in the phantasatron is of the double-pumped type, permitting a clear view of the non-emitting triple-processed plunger-type plate. Because of its unique self-flushing construction, this tube will remain usable throughout its useful life.”

3. The Umac 606 was a spoof vacuum tube with an anode shaped like a toilet plunger. (Courtesy of EIMAC Division of Varian)

You can tell that engineers loved to parody marketing jargon, even 67 years ago. As a former application engineer, I love the “application notes” section of the Umac 606, which contains the phrases, “Because of its unique construction, the 606 can serve as an oscillator, modulator, or amplifier. It usually serves as all three simultaneously. No specific operating conditions are available. Ratings will be available however, as soon as our competitors issue their catalogue. Exhaustive tests in our advertising department have shown that the 606 will give 50 percent more output than you will obtain.”

The comment about the datasheet having better specs than you will obtain rings true for me. It took a lot of patience to explain to customers that specs in the datasheet were optimum specs at very specific operation conditions. Many customers thought it was somehow unfair or fraudulent that an amplifier could not have the same distortion or power-supply rejection at 3 V as it had at 15 V. Of course, they wanted to use it at 3 V and were quite miffed that the part did not measure up to their expectations.

The DEAD, the SCROM, and Pre-Earthquake Motion Sensors

Another great prank part was dreamed up by former Linear Tech CTO Bob Dobkin. Now part of Analog Devices, Dobkin dreamed up the DEAD in 1975, a Darkness Emitting Arsenide Diode. Pease noted, “This had many features— light output never fell off with time, and they were cheap due to high yield. Just check the acronym.”

4. Jerry Lawson at the eFlea breakfast in March 2011, a month before his death. Jerry was instrumental in applications and the cartridge video game at Fairchild Semiconductor.

Some spoof ICs have a basis in fact. While at Signetics, Jerry Lawson(Fig. 4) came up with the term SCROM, the “Scratchable Read-Only-Memory.” It was a joke reference to the way that they could prototype a ROM (read-only memory) by scratching the metallization on a die to give the memory values desired. This was when a 64-bit memory was big, back in the 1970s. I have a video interview I did with Lawson that I will get up soon. He was a great guy, yet another friend I made at the Silicon Valley Electronics Flea Market.

The pranksters at National Semiconductor (now Texas Instruments) came up with another spoof part in 1989. It was the LM80/LM280/LM580/LM680/LM880 Precision Pre-Earthquake Motion Sensors (Fig. 5). The datasheet date, October 17, 1989 5:04 PM, was the exact moment of the Loma Prieta earthquake in San Francisco. The LM part numbers are all interstate freeway designations around the Bay Area.

5. A spoof datasheet for a pre-earthquake sensor that the engineers at National Semiconductor came up with shortly after the 1989 Loma Prieta earthquake.

The datasheet starts, “The LM80 series are precision integrated-circuit pre-earthquake motion sensors, whose output voltage is linearly proportional to the realtime logarithmic Richter Scale reading. The LM80 thus has an advantage over logarithmic pre-earthquake motion sensors calibrated in momentum newtons, as the user is not required to calculate the log (base 10) of the measured motion from the device output to obtain the convenient Richter scaling.”

The datasheet features include wildly implausible accuracy, temperature, current, and self-heating claims. I am not sure where I got this datasheet image, but it was most likely from the prankster-in-chief, Bob Pease himself. Perhaps he should have been called the Czar of Pranks, not the Czar of Bandgaps.

With the wires inside chips growing smaller and closer together, miniature shower heads installed in their packaging could help keep them cool, spraying coolant directly onto the surface of the integrated circuit. The rinse could prevent chips from burning themselves up as semiconductor companies push performance boundaries.

That technology was recently introduced by semiconductor researcher Imec to help meet the growing demands of high performance electronic devices, including three-dimensional chips. These integrated circuits wire together memory, processing and other blocks to get around the limitations of existing chips, which have trouble handling artificial intelligence and other tasks.

“Our new impingement chip cooler is actually a 3D printed ‘showerhead’ that sprays the cooling liquid directly onto the bare chip,” said Imec’s Herman Oprins, senior research engineer focused on thermal management, in a statement. “3D prototyping has improved in resolution, making it available for realizing microfluidic systems such as our chip cooler.”

Imec's system is designed to be integrated into packaging wrapped around these integrated chipsets, which carry with them persistent thermal issues. The Leuven, Belgium-based Imec was able to manufacture the system out of polymer all at once, cutting production cost and time. The researcher would not say when the cooling technology could be generally available.

Imec says that its system evenly distributes coolant over the surface of the integrated circuits. The tiny spouts inside the system, which measure 300μm – or roughly twice the width of a human hair – can be custom manufactured to match specific heat maps, lowering the chances of hot spots forming. The nozzles can also be made to fit inside the chip’s packaging.

The spouts shoot liquid into a small space in between the cooling system and the surface of the semiconductor before heated coolant is expelled through separate plumbing. Traditional chip cooling systems use heat spreaders attached to the bottom of the substrate, but the thermal materials bonding the heatsinks in place have predetermined heat resistances, which is a major limitation, Imec said.

Tiny irrigation channels can also be etched onto the back of the substrate and flooded with coolant that absorbs heat as it flows over hot spots. But that technology can leave some parts of the integrated circuit hotter than others. Cooling systems like Imec’s can also be manufactured with silicon, but it can expensive to customize them based on the chip’s packaging – a shortcoming that Imec said that it can avoid.

Celebrating Bob Pease
It’s hard to believe so many years have passed since the analog industry lost one of its most highly respected gurus, Bob Pease.

To celebrate Bob’s memory, Electronic Design released two eBooks featuring collections of reader favorites. These articles are timeless and showcase why Bob Pease will always hold a revered place in the analog industry. We miss your wit and unabashed style of writing, Bob.

Complete the form below, click Submit and watch your inbox for an email with a link to access your personal copy.

In Volume I Bob Discusses:

Analog Engineering
Technical Reading
Transimpedance Amplifiers
Frequency to Voltage Converters
Capacitor Leakage
Noise Gain
Current Limiter
Output Impedance

Celebrating Bob Pease

We all remember Bob Pease fondly—for his analog design expertise as well as his sense of humor. We were lucky he shared his wit and knowledge with us for years with his special column for Electronic Design, “Pease Porridge.” Earlier we released Vol. 1 in a series of two eBooks to commemorate this legendary engineer. Vol. 2 of Electronic Design’s special collection of articles written by Bob is now ready for download.

Complete the form below, click Submit and watch your inbox for an email with a link to access your personal copy.

In Volume 2 Bob Discusses:

Battery-Power
Battery-Charger
Ripple Rejection and "Soakage"
Best Amplifiers
Bridge Amplifiers
Thoughts on Energy

Shaving costs from the supply chain is the priority for almost one-third of respondents to SourceToday’s 2018 Distribution Study survey. But the results also suggest that more companies are focused on boosting supply chain transparency as new threats, such as widespread parts shortages and trade uncertainties, collide with more familiar concerns.

The study uses almost 1,100 responses from SourceToday and Electronic Design readers in procurement, engineering and executive roles. The results demonstrated what has remained the same about the electronics industry’s priorities: Roughly 31 percent said that they are focused primarily on trimming supply chain costs, down from 32 percent in 2017. The number of respondents focused on creating new supply and distribution relationships fell marginally from 20 to 19 percent.

But the number of respondents primarily trying to increase transparency and collaboration within the supply chain ramped up slightly to 17 percent, from 15 percent in 2017. The survey showed that this is the priority of one-fifth of procurement professionals, versus 16 percent for engineering staff and 18 percent for executive management – up from 18 percent, 14 percent and 16 percent from 2017, respectively.

The transparency push comes amid mounting challenges in the supply chain, including the growing threat of counterfeits and recent parts shortages that have created a feeding frenzy for inventory. Facing shortages of everything from capacitors to semiconductors, many companies are digging into the supply chain, putting in requests for the same part with several distributors and then cancelling one of the orders after getting it – also known as double booking.

The Trump administration could also create supply chain disturbances. On Tuesday, the White House said that it would proceed with its proposed 25-percent tariffs on $50 billion worth of Chinese products. The original list, compiled by the U.S. Trade Representative, included parts in short supply globally like transistors, batteries, diodes and capacitors. But a final list of affected products is supposed to be released on June 15.

Counterfeits remain the worst nightmare for the survey respondents. Over the last year, the share of respondents that pointed to counterfeits as either very important or somewhat important supply chain risks jumped from 72 to 81 percent. The executive branch was generally less concerned about counterfeits (73 percent) than those with engineering (81 percent) and procurement jobs (88 percent).

Running afoul of regulations is another major source of apprehension. In 2018, almost three-quarters of respondents pointed to environmental compliance as a serious supply chain concern, up from two-thirds in 2017. Last year, complying with import and export laws was at least somewhat important to 58 percent. But that percentage has grown to 64 percent as the United States continues to stoke fears of a trade war with China.

Conflict mineral disclosure rules are also causing hand-wringing in the electronics industry. These regulations are at least somewhat important to 49 percent of survey respondents, up from 44 percent over last year. But if American officials follow through on plans to repeal or weaken rules that require the disclosure of conflict mineral use – mainly tin, tungsten, tantalum and gold – that percentage could fall.

The survey showed that compliance concerns were proportional to how close the respondent was to purchasing. Four-fifths of procurement staffers said that environmental regulations were at least somewhat important to them, as opposed to 73 percent of electrical engineers and 61 percent of executives. Conflict mineral rules were flagged by 58 percent of those with procurement jobs, but only 45 percent in executive roles.

That heightened sensitivity may have something to do with the 22 percent of procurement professionals that are focused on adding new technology to their supply chain processes. Last year, that number was at 18 percent. Industry analysts agree that businesses could be better at using software to keep track of pricing, suppliers, as well as contract conditions and penalties, giving them a clearer view of their supply chains.

Apple is again throwing its weight around in the semiconductor industry. Dialog Semiconductor, the company's primary source for power management chips, said in a statement that Apple would introduce a second source for the main PMIC components used in its smartphones. The result is Apple ordering fewer chips from the company this year than Dialog originally expected.

Dialog projects orders of power management chips embedded in Apple’s tablets, wearables and personal computers to remain unchanged. In addition, the company expects to continue supplying another power management chip used inside Apple’s smartphones, the sub-PMIC, in the same volumes. Some analysts estimate that the company owes more than 50 percent of its revenue to Apple’s patronage.

Jalal Bagherli, Dialog’s chief executive officer, said on a conference call that Apple was probably trying to reduce the risk of relying upon a single manufacturer for chips managing battery power inside its smartphones. Bagherli also said it was likely that Apple would design power management chips in-house, a possibility rumored over the last year that has hammered Dialog’s share price.

“Dialog understands its continued role as Apple’s main PMIC supplier is contingent on Dialog meeting Apple’s technology, quality, price and volume expectations, as well as continuing to develop advanced technology to meet Apple’s requirements,” the company said in the Friday statement, which sparked an overwhelming negative response from investors. Dialog’s shares plunged more than 17 percent and have hardly recovered.

The company said that its customer’s change of heart would only shave around five percent from its annual revenue and that its business would still report growth in 2018. Dialog has been trying to expand into new businesses in recent years, most recently through its $330-million acquisition of mixed-signal chip supplier Silego Technology. On Friday, the company said it would look into mixed-signal opportunities inside future Apple products.

Whenever the first fully autonomous cars start driving passengers, every vehicle will probably be equipped with three primary sensors to monitor road conditions, each one covering each other’s blind spots. Radar systems have low resolution but operate in bad weather conditions fumbled by cameras and lidar sensors, which themselves compensate for radar’s resolution issues.

On Semiconductor is already a major manufacturer of image sensors, but in recent years, the Phoenix, Arizona-based company has been acquiring automotive radar and other sensors to build more complete solutions around its core business. It is also tapping into the trend of sensor fusion, which involves combining raw data from different sensors into more insightful information, enhancing automotive safety or other applications.

The company recently made its move into lidar with the acquisition of Cork, Ireland-based SensL Technologies, which specializes in single photon avalanche diodes (SPAD) and silicon photomultipliers (SiPM) that can be used in lidar sensors that maps objects in three dimensions by bouncing laser beams off its surroundings. The terms of the deal were not disclosed.

The SiPM chips are designed to sense the physical shape of objects and how far away they are. On Semiconductor is betting that they could be used as the building blocks of solid-state lidar solutions because they can be manufactured with complementary metal-oxide semiconductor (CMOS) technology. The devices could be much more robust, smaller and cheaper than currently available sensors, because they will not have moving parts.

Taner Ozcelik, senior vice president of On Semiconductor’s image sensor group, said in a statement that “automotive sensor fusion demand is growing at an accelerated pace with a need for additional sensor technologies that are provided by the SensL team.” The company’s sensors can also be designed into industrial robotics, inspection drones and consumer devices, he said.

Last year, On Semiconductor agreed to acquire an IBM research group focused on millimeter wave automotive radar. The acquisition included 15 researchers based in Haifa, Israel, working on increasing the output power for fixed-beam silicon chips operating in frequencies between 60 and 90 gigahertz. Before the end of the year, On Semiconductor plans to introduce samples of the chips.

Radar systems function in all weather conditions, watching for other vehicles on the road when cameras are obscured by rain or thrown off by the sunlight reflecting off a snowy street. But they lack the resolution to distinguish objects, which is why lidar sensors are widely considered vital for autonomous driving. On the other hand, lidar sensors cannot see things like lane markings or the color in taillights—a shortcoming that cameras can compensate for.

On Semiconductor is facing off against numerous competitors that have raised hundreds of millions in venture capital to build better lidar. One stumbling block could be the significant expertise ranging from optics that spray lasers out into the world and the software that processes the reflections required to manufacture lidar for cars. On Semiconductor could always sell the sensors to other companies with those supporting skills.

On Semiconductor’s bet remains to be tested. Not yet clear is how much money the company plans to invest in its new SensL business unit, which will operate as part of On Semiconductor’s image sensor group. Last year, the company spent $594.5 million on research and development on annual revenue of $5.54 billion. SiPM sensors have more traditional uses in medical imaging and radiation detection, Ozcelik said.

When Vesper Technologies raised $23 million in funding last month, the MEMS microphone maker’s investors included many of the major players in voice-controlled devices. Amazon’s Alexa Fund, Bose Ventures, Synaptics and Baidu poured money into the funding round, which was led by American Family Ventures, the venture capital division of insurance giant AmFam.

As voice assistants like Amazon Alexa and Google Home turn talking to smart speakers into an everyday occurrence, Boston, Massachusetts-based Vesper is trying to capitalize on the push to control everything from televisions and wearables to refrigerators and headphones with a few simple spoken commands. The startup makes microphones that consume much less power than traditional ones.

The company expects the first products using its new microphones to reach the market before the end of the year. And to support customer volumes, Vesper will use the new funding to ramp up production from the several hundred thousand microphones it shipped in the first half of the year to several million per month, said Vesper’s chief executive Matt Crowley in an interview.

Crowley claims that current capacitive MEMS microphones consume too much power, lack durability, and have low dynamic range. These problems are connected to the fact that they are manufactured with two parts: a diaphragm and a back plate. Not only can water and dust get stuck inside the space between the back plate and the diaphragm, hurting reliability, but air pressure can accumulate inside the cavity, dampening the sensor’s sensitivity to voices.

Conversely, Vesper's microphones do not have a back plate. Instead, they use a single diaphragm based on piezoelectric materials that generate a small voltage when pummeled with sound waves, cutting power consumption and even enabling the microphone to wake up the system. The simple structure improves durability as well as signal-to-noise performance, which allows the sensor to isolate voices from farther away and sense a wider dynamic range of voices clearly, said Crowley.

Additionally, the microphone lets systems stay powered down and wake up when they hear a voice command, something that Crowley calls quiescent sensing. “We’re replacing buttons that you press with your finger with buttons you press with your voice,” Crowley told Electronic Design. “The power difference is between having the whole system up and running, versus only having the low-power microphone running,” he said.

And the company’s business is budding. Last year, the company shipped more than a million microphone wafers to partners including ACC Technologies. Vesper has since started to handle testing and packaging so that it can supply customers with complete microphones, including the wake-on-sound VM1010 microphone, which accounts for around 80 percent of the sensors it sells. The company’s other products are the analog VM1000 and digital VM2000.

“Vesper has the technology to disrupt established MEMS microphone players and some sweet spots to run after,” said Guillaume Girardin of semiconductor research firm Yole Développement, in a statement. “People are now expecting a ramp up in the production to fulfill market needs, especially in the smart speakers or wearable markets.”

Vesper has friends in high places. Amazon’s Alexa Fund, which assisted with the company’s $17 million funding round in December 2016, pushes Vesper’s microphones to potential customers when it considers the technology a good fit. Vesper has also partnered with chip companies including Ambiq Micro and Synaptics on development kits that help integrate voice assistants like Alexa into embedded devices. Crowley declined to comment on whether its microphones are used in any Amazon products.

The company, which was founded in 2009, is trying to close the massive gap with market leader Knowles, which ships more than a billion capacitive MEMS mics every year. While Vesper has pushed into new materials to boost the performance of voice assistants, Knowles is working to pair its microphones with audio processing software that can amplify faint sounds and suppress loud chatter in noisy spaces.

Paris, France-based Yole Développement projects shipments of MEMS microphones to reach 5.6 billion units by the end of the year, generating revenues of $1.1 billion. Device manufacturers are adding more and more microphones to voice-controlled devices in an attempt to improve the clearness of different voices and identify the direction that they came from. And that plays into Vesper’s strengths.

“As the number of microphones increases, the need for the reliability of the individual components increases,” explained Crowley. “If you have an array of a dozen microphones, and you have drift in one microphone, the whole array suddenly starts pointing in the wrong direction.” He added: “The array just stops working properly.”

Crowley told Electronic Design that it would release a new microphone line by the end of the year focused on ultra-wideband dynamic range so that it can scrub speaker distortion and suppress wind noise. He said that the company is also considering building modules that package its microphones together with sensors that measure things like temperature and humidity.

Because of the power consumption differences, the company’s microphones are not drop-in replacements for capacitive sensors. But Crowley said Vesper will use the funding to expand implementation support for customers and hire an additional 15 employees to fill out its engineering and sales teams. Vesper, which currently employs 27 people, will also expand its sales operations in South Korea and China.

Medical ultrasound imaging is a non-invasive method of viewing internal organs and structures of the human body using high-frequency acoustic waves. Sending an acoustic wave into the human body and listening to the acoustic echoes can create an ultrasound image. Given the acoustic properties of the human body, the optimum frequency range for general medical ultrasound imaging is 1.0 to 12.0 MHz.

An acoustic wave is created with a piezoelectric transducer (PZT). The PZT physically expands and contracts when a voltage is applied to it, thereby converting electrical energy into acoustic energy. High-voltage pulses are used to excite the PZT. This is done with a pulse generator, sometimes referred to as the transmitter, with amplitudes of up to ±90 V and current capability of ±2.0 A. The same PZT then is used to convert acoustic echoes back into electrical signals, which are referred to as the receive signals. The received signals are processed and analyzed to construct an ultrasound image of the transmitted path.

To better understand the principle of operation, consider a single-channel PZT (Fig. 1). During the transmit cycle, a single 5-MHz cycle of ±90 V is generated from the transmitter and applied across the PZT. The PZT creates a 5-MHz acoustic wave that travels into the body, and the receiver starts listening for echoes. Acoustic waves travel on the average of 1.54 mm/μs within the human body.

1. Shown is a typical single-channel piezoelectric transducer (PZT).

To create an image of an object that is 10 mm away from the PZT, a time duration of 10 mm/(1.54 mm/μs) = 6.49 μs is required for the acoustic transmit wave to reach the object. Another 6.49 μs of time is needed for the acoustic echo to reach the PZT from the object. The signal received at 12.98 μs corresponds to an object 10 mm away from the PZT. The time duration for the received cycle is therefore longer when imaging objects further away from the PZT. Once all echoes are received, one line of image can be constructed.

The receiver is a high-performance, low-noise, low-voltage device that can be easily damaged if the high-voltage pulses from the transmitter are applied to its input. Fig. 1 shows a T/R SW block—a transmit/receive switch—that’s used to block the high-voltage transmit pulse, but allows the low-voltage receive signal to pass into the receiver’s input. The receive voltage is typically less than ±500 mV.

To create a two-dimensional image, an array of PZTs is needed. An ultrasound probe is used to house the array of PZTs. There are many different types of ultrasound probes, such as abdominal, cardiac, and pediatric probes, among others, and all are specifically designed for various applications. The number of PZTs can vary depending on the probe type, ranging from 128 PZTs to 512 PZTs. An array of 192 PZTs will be used as an example for this article.

Use of High-Voltage Analog Switches

Figure 2 shows a basic medical ultrasound system driving an ultrasound probe with 192 PZTs. High-voltage analog switches are used to multiplex the transmitter, receiver, and T/R switch set to different PZTs.

2. High-voltage analog switches are used in this medical ultrasound system that’s driving an ultrasound probe with 192 PZTs.  

The example in Fig. 2 has 64 sets of transmitters, receivers, and T/R switches. Each set drives three PZT elements. Its 192 high-voltage analog switches are arranged whereby groups of 64 PZT elements are being driven. The analog switches route the 64 sets of transmitters, receivers, and T/R switches to drive PZT 1 to 64. The 64 transmitters send high voltage pulses into PZT 1 to 64. The system waits until it receives all echoes from PZT 1 to 64.

For the next cycle, the analog switches reroute the 64 sets of transmitters, receivers, and T/R switches to drive PZT 2 to 65. The 64 transmitters send high-voltage pulses into PZT 2 to 65, and the system waits to receive all echoes from PZT 2 to 65. This routine of transmitting, receiving, and rerouting repeats with the analog switches rerouting the transmitters, receivers, and T/R switches by one PZT increment at a time. Once all 192 cycles are completed, one frame of image can be constructed in about 50 ms.

Without the use of high-voltage analog switches, 192 transmitters, receivers, and T/R switches are required to drive the 192 PZTs. With the use of high-voltage analog switches, the number of transmitters, receivers, and T/R switches is reduced by a factor of three, saving cost, power, and size.

Advantages and Benefits of MPS High-Voltage Analog Switches

Figure 3 shows the basic differences of using conventional analog switches versus analog switches developed by Monolithic Power Systems (MPS). Conventional high-voltage analog-switch ICs require two high-voltage power supplies, +100 V and −100 V, for proper operation. The high-voltage pulses going through the analog switch must be within 10 V of the high-voltage supplies. For ±100-V supplies, the maximum transmit voltage is ±90 V. Additional circuits are needed to generate these two high-voltage supplies, since they’re not required elsewhere in the system. These are dedicated power supplies for the conventional high-voltage analog switch ICs.

3. A conventional analog switch is compared with an analog switch developed by MPS.  

Safety concerns come with such high-voltage supplies. Protection circuitry is needed to eliminate the risk of shock under various fault conditions, such as damaged insulation. Power-up and power-down sequencers should also be considered for safe operation. The voltage level on the high-voltage supplies must be monitored to inhibit the transmitter if the voltage levels are too low.

On the other hand, the MPS MP4816A IC is a 16-channel, high-voltage, analog switch that uses a 10-V supply instead of two high-voltage power supplies. This removes the need for the two high-voltage power supplies, which reduces the complexity of the power-supply design and lowers power dissipation. The support circuitries associated with the two high-voltage supplies, such as power-up and power-down sequencers and voltage monitors are also eliminated. The end result is a system with lower cost, reduced size, and increased reliability.

High-Voltage Analog Switches in the Probe Head

Using analog switches configured as a 1-to-3 multiplexer inside the ultrasound probe head with 192 PZTs reduces the number of coaxial cables by a factor of three. Instead of 192 coaxial cables, only 64 coaxial cables are needed for the PZTs, plus 10 or fewer additional coaxial cables for the logic supply and I/O interface.

Reducing the number of coaxial cables results significantly lowers cost for the ultrasound probe head, as coaxial cables are very expensive, as is the labor cost to connect them. An added user benefit for the sonographer is that the probe head becomes more maneuverable, creating less fatigue.

The probe head is generally severely limited spatially and thermally. The housing is waterproof, as it must be able to withstand being submersed in alcohol for sterilization after use, making it difficult to remove any heat being generated inside the probe head. Eliminating the high-voltage supplies reduces power dissipation and wipes away any safety concerns of high-voltage dc lines on the coaxial cables.

Conclusion

High-voltage analog switches are commonly used in medical ultrasound systems, both in the system and in the ultrasound probe head. The design of MPS’s high-voltage analog switches eliminate the need for high-voltage supplies, simplifying the power-supply design, lowering overall system cost and size, and increasing reliability.

As nearly all audio amplifiers have transitioned from linear or near-linear operating modes (Class A, AB, and B) to digital (Class D), it may seem that the only areas for improvement are in power rating, efficiency, enhanced filtering, and output protection. Certainly, that’s where the bulk of the new product development has occurred, but the world of audio amplifiers for consumer and home applications is changing, too.

To address this situation, Texas Instruments has introduced thee digital-input, Class-D audio amplifiers with distinct, non-overlapping features and functions. They target the evolving needs of audio installations, such as smart speakers, sound bars, TVs, notebook computers, and projectors.

TAS2770

The TAS2770 15-W monophonic audio amplifier increases loudness while supporting easier voice-input interfacing (Fig. 1). This IC can deliver high-output-level peak power into small loudspeakers, and includes loudspeaker protection via built-in monitoring of speaker current and voltage (I/V sense). It also includes a digital-microphone input channel and captures voice and ambient acoustic information for echo cancellation or noise reduction in voice-enabled applications; TI claims that the combination of digital microphone input with a I/V sense amplifier is an industry first.

1. The TAS2770 audio amplifier from Texas Instruments includes integrated speaker voltage and current sense functions for real-time monitoring of loudspeaker condition and behavior.

To avoid the distortion of audio clipping, often due to, but not limited to, supply “brown out” (when the supply rail can’t deliver adequate dc power for the desired output power), it automatically decreases gain when audio signals exceed a set threshold. Up to eight devices can share a common bus via either I²S/TDM or I²C interfaces. Pricing for the 26-pin VQFN starts at $1.49 (1,000 pieces).

TAS5825M

The TAS5825M audio amplifier delivers high-resolution audio due to the device’s high input-sampling frequency (192 kHz) along with flexible, integrated processing flows. In addition, the TAS5825M provides bass enhancement and thermal protection for the speaker. To simplify echo cancellation, its dedicated data output via the I²C audio interface provides ambient-sound information to the applications processor.

An integrated sample-rate converter (SRC) detects the input sample rate and auto-converts it to the target sample rate. A proprietary hybrid-mode modulation scheme reduces idle-power losses and thermal dissipation without degrading sound quality. The 32-pin VQFN starts at $2.64 (1,000 pieces).

TAS3251

According to TI, the TAS3251 audio amplifier is the first integrated, dual-channel, digital-input device to support high output power and performance, delivering 175 W per channel into 4-Ω loads or 220 W per channel (3 Ω loads) in a single package (Fig. 2). In bridge-tied load (BTL) mode, the channels can be combined to double the power to the single channel.

2. The TAS3251 175-W stereo/350-W mono, ultra-high-definition audio amplifier from Texas Instruments has a built-in DSP core that supports advanced signal processing, filtering, and other functions, and can compensate for loudspeaker and room characteristics.

The 90% efficient design (4 Ω) minimizes heat. However, at these power levels, “excesses” can be a problem, so the device guards against and includes error reporting of factors such as undervoltage, cycle-by-cycle current limiting, output short circuit, clipping detection, overtemperature warning and shutdown, and dc speaker protection. It operates from a 12- to 36-V supply rail and includes an I²C interface. The 56-pin HSSOP device is priced at $5.95 (1000 pieces).

Audio ICs such as these pack numerous features and functions into their design, so design-in may not seem easy, at first. Easing the design-in effort begins with detailed datasheets (96, 71, and 123 pages, respectively), each replete with performance charts and tables across a wide range of operating conditions; register configurations for setup and operating modes; and suggested PCB layout options. In addition, design-in is bolstered by reference designs and evaluation boards, as well as TI’s PurePath Console software, which eases configuration of these amplifiers.

If you’re designing a linear circuit of any kind today, it most likely involves an op amp and even an analog-to-digital converter (ADC). The op amp is your go-to device simply because it can be configured to perform almost any linear function. And with most applications being digital today, your design probably includes an ADC, either as an individual IC or integrated into your MCU.

If you have experience in linear design with these devices, your end product will be a success.
Such an outcome can be achieved by enhancing your knowledge with an input of fresh new knowledge about these ubiquitous devices. There are some excellent online ebooks that can provide you with all of the latest hints and tricks to optimize and bullet-proof your design.

Op Amps Are the Answer

Op amps have been around for over 70 years. I first encountered them in an old Heathkit vacuum-tube analog computer we used in college. Then later in one of my first jobs, I was encouraged to use other vacuum-tube op amps that were on hand. These were the infamous K2Ws made by Philbrick. They worked great, but those ±300-V power supplies were just not appropriate for the truck-mounted mobile equipment I was making. I then introduced the company to the early solid-state op amps developed by Philbrick and Burr-Brown.

In the early 1970s, the first IC op amps came along. I used the popular 301 and 709. And later on, who didn’t use the 741? It’s still around today. Now, of course, we have superior CMOS op amps with mega-bandwidth. Furthermore, the specifications are so good that we often don’t have to worry so much about how to compensate for input offset voltages, bias currents, and other limitations.

Virtually all EEs learn about op amps in college. It’s one of those basic things that they do teach you in school. Most of us learned about the most popular circuits, such as the follower, inverter, the non-inverting amplifier, summer, and integrator—all still widely used.

But, as you know, hundreds or thousands more special circuits can be built with an op amp. In fact, at this point, most common basic linear circuit needs and applications have been discovered. And it’s likely that you do NOT know them all. The solution to that knowledge gap is a compendium of useful circuits that you can reference when designing your next linear project.

Download this article in PDF format.

If you’re designing a linear circuit of any kind today, it most likely involves an op amp and even an analog-to-digital converter (ADC). The op amp is your go-to device simply because it can be configured to perform almost any linear function. And with most applications being digital today, your design probably includes an ADC, either as an individual IC or integrated into your MCU.

If you have experience in linear design with these devices, your end product will be a success.
Such an outcome can be achieved by enhancing your knowledge with an input of fresh new knowledge about these ubiquitous devices. There are some excellent online ebooks that can provide you with all of the latest hints and tricks to optimize and bullet-proof your design.

 Sponsored Resources: 

• Analog Engineer's Circuit Cookbook: Op Amps
• Analog Engineer's Circuit Cookbook: ADCs

Op Amps Are the Answer

Op amps have been around for over 70 years. I first encountered them in an old Heathkit vacuum-tube analog computer we used in college. Then later in one of my first jobs, I was encouraged to use other vacuum-tube op amps that were on hand. These were the infamous K2Ws made by Philbrick. They worked great, but those ±300-V power supplies were just not appropriate for the truck-mounted mobile equipment I was making. I then introduced the company to the early solid-state op amps developed by Philbrick and Burr-Brown.

In the early 1970s, the first IC op amps came along. I used the popular 301 and 709. And later on, who didn’t use the 741? It’s still around today. Now, of course, we have superior CMOS op amps with mega-bandwidth. Furthermore, the specifications are so good that we often don’t have to worry so much about how to compensate for input offset voltages, bias currents, and other limitations.

Virtually all EEs learn about op amps in college. It’s one of those basic things that they do teach you in school. Most of us learned about the most popular circuits, such as the follower, inverter, the non-inverting amplifier, summer, and integrator—all still widely used.

But, as you know, hundreds or thousands more special circuits can be built with an op amp. In fact, at this point, most common basic linear circuit needs and applications have been discovered. And it’s likely that you do NOT know them all. The solution to that knowledge gap is a compendium of useful circuits that you can reference when designing your next linear project.

Signal-Chain Circuits Simplified

The op amp has been massively documented over the years. As mentioned, it’s covered in college texts and many other trade books for engineers. No doubt you have several, as do I. However, there’s probably no better source of information than that from an IC manufacturer, whose devices show up in a huge number of designs. Chip companies are the experts, thanks to their field engineers, datasheets, and application notes. Now they offer e-books.

The Analog Engineer’s Circuit Cookbook: Op Amps is an example. It compiles more than 25 popular op-amp circuit designs like current sensing, attenuating, and full-wave rectifying into a single e-book. Download the e-book for op-amp circuit ideas that you can easily adapt to your design needs.

Op Amp Quiz

1. Schematic for op-amp quiz.

Here’s a simple test of your basic op-amp knowledge. Refer to Figure 1, and answer these questions:

1. What is the gain of this circuit?

2. What is the input impedance of this circuit?

3. Given the sine-wave input signal shown, what is the output?

You should probably know the answers, which can be found at the end of this article.

Welcome to the ADC

Like early op amps, ADCs were fussy devices. They suffered from low speed, noise, and linearity issues. Today these glitches have mostly been worked out, making ADCs almost as easy to use as an op amp. But you still need to know the rules and procedures to produce a clean, workable design. ADCs are also well-documented. Again, though, seeking the experience of ADC suppliers is the secret to creating a workable design the first time around.

A good starting point is The Analog Engineer’s Circuit Cookbook: ADCs, which compiles more than 15 popular ADC circuit designs like level translation, input drive circuits, and commonly used analog front-end (AFE) circuits. You can download the e-book for ADC circuit ideas that you can easily apply to your system needs.

ADC Quiz

2. Schematic for ADC quiz.

Check your knowledge of ADCs with this simple quiz. Refer to Figure 2, and answer these questions:

1. What is the minimum voltage that the ADC can resolve?

2. What is the minimum sampling frequency for proper conversion?

If you passed both quizzes, congratulations. However, if you did not, then the e-books recommended here will give you a refresher.

Answers to Quiz Questions

Op Amp

1. A = R_f/R_i = 15k/4.7k = 3.19

2. Z_IN = R_i = 4.7k

3. The output is a ±5-V rectangular wave. The input signal and gain are too high, and the output saturates at the limits of the supply voltages.

ADC

1. Minimum resolution = 5/1024 = 4.88 mV

2. 40 kHz; no less than twice the maximum input frequency (e.g., Nyquist)

 Sponsored Resources: 

• Analog Engineer's Circuit Cookbook: Op Amps
• Analog Engineer's Circuit Cookbook: ADCs

When Professors Sir Andre Geim and Sir Kostya Novoselov of the University of Manchester (UK) discovered and isolated a single atomic layer of carbon for the first time—now known as graphene—there was both praise and concern. Some comments were similar to those which, decades before, accompanied the first demonstration of the optical laser, heralding it as “a solution looking for problems to solve.”  (Note that the professors received the Nobel Prize in Physics in 2010, in recognition of their work.)

Of course, we know how the story of the laser’s applications turned out with its truly countless uses, many of which were literally unimaginable at the time. Graphene is tracking along a similar trajectory with applications in high-performance batteries and supercapacitors, advanced high-strength and lightweight materials, materials having extremely high thermal conductivity, plus many more.

Among the latest developments is the use of graphene as the basis for washable electronics, from a research team led by Iowa State University. By combining graphene and sophisticated laser-based processing, they have developed circuits that are low-cost, flexible, highly conductive, and water-repellent.

“We’re taking low-cost, inkjet-printed graphene and tuning it with a laser to make functional materials,” said Jonathan Claussen, an assistant professor of mechanical engineering and an associate at the U.S. Department of Energy’s Ames Laboratory, as well as the corresponding author of the paper published in the journal Nanoscale“Superhydrophobic inkjet printed flexible graphene circuits via direct-pulsed laser writing” (with additional modeling, simulation, and supplemental information here). The material and resultant circuitry could be useful for a wide range of applications such as self-cleaning, wearable, or even washable electronics, they claim.

The process begins with an inkjet printer (another mass-market product finding all sorts of fascinating, unforeseen applications such as 3D printing), where the “ink” consists of flakes of graphene, an excellent electrical conductor that’s also strong, stable, and biocompatible. Unfortunately, the printed flakes aren’t highly conductive and have to be processed to remove non-conductive binders and weld the flakes together, thus boosting conductivity and making them useful for electronics or sensors. By applying a technique called direct-pulsed laser writing (DPLW), the team was able to adjust and fine-tune the water-retaining/repelling characteristics and electrical conductivity of the inkjet-printed graphene (IPG).

Wash and No Wear

The graphene surface pattern (left) and wettability contact angle (right), shown before (top pair) and after (bottom pair) the direct-pulsed laser writing procedure. (Source: Iowa State University)

The DPLW rapid-pulse laser process treats the graphene without damaging the printing surface—even if paper or ultrathin polymers—and transforms graphene-printed circuits that can hold water droplets (hydrophilic) into circuits that repel water (superhydrophobic). Their experimental results indicate that the DPLW process can convert the IPG surface wettability from one that’s initially hydrophilic (contact angle CA of about 7.7°) and electrically resistive (sheet resistance ∼21 MΩ per square) to one that’s superhydrophobic (CA ∼157.2°) and electrically conductive (sheet resistance ∼1.1 kΩ per square) (see figure).

“We’re micro-patterning the surface of the inkjet-printed graphene,” said Claussen. “The laser aligns the graphene flakes vertically—like little pyramids stacking up. And that’s what induces the hydrophobicity.”

The energy density of the laser processing can be adjusted to tune the degree of hydrophobicity and conductivity of the printed graphene circuits. (Note that DPLW is another one of those fascinating technologies having widespread, disparate applications including welding tiny materials, “drilling” precision holes without burrs or workpiece distortion, microsmoothing and super-polishing finished surfaces, and printing identification legends on tiny devices.)

The studies are supported by the National Science Foundation, the U.S. Department of Agriculture’s National Institute of Food and Agriculture, the Roy J. Carver Charitable Trust, and Iowa State’s College of Engineering and department of mechanical engineering. The Iowa State University Research Foundation is hoping to patent the technology, and has also optioned it to NanoSpy Inc. ( an Ames-based startup) for possible commercialization.

Reference

https://graphene-flagship.eu/Pages/default.aspx

Dialog Semiconductor is looking to diversify its business after one of its largest customers, Apple, slashed orders for the power management chips used in the iPhone. The company, which industry analysts estimate owes more than 50 percent of its business to Apple, announced on Tuesday that it was discussing an acquisition of Synaptics, which supplies chips for touch and voice interface applications.

In a statement, Dialog disclosed that it’s performing due diligence on Synaptics but cautioned that nothing has been decided yet. The company, which holds a market value of more than $1.4 billion, will have to overcome its recent troubles trying to diversify its product offerings. Two years ago, the company fumbled its acquisition of microcontroller supplier Atmel, which agreed to Microchip Technology’s rival bid of nearly $3.6 billion.

The discussions with San Jose, California-based Synaptics were first reported by Bloomberg.

Dialog, which has long been focused on power management chips for smartphones and other gadgets, is trying to reduce its dependence on Apple. When Apple reined in orders for Dialog’s chips this month, the company said it expected to lose five percent of its annual revenue, though it would still report growth in 2018. Dialog is currently looking into selling Apple on mixed-signal chips it acquired in its recent $330-million acquisition of Silego Technology

Synaptics has also been expanding its business beyond display drivers and touch sensors used in smartphones. Last year, the company completed its $300-million acquisition of Conexant Systems, which supplies voice and audio processing solutions for smart speakers and a growing number of voice-controlled devices. Synaptics, now with a market value of $1.7 billion, used another $95 million to buy Marvell’s multimedia solutions business, which also targets smart home applications.

Component obsolescence was flagged as a supply chain threat by roughly three-quarters of respondents to an Electronic Design and Source Today survey, representing an almost seven percent jump over the last year. The results of the 2018 Distribution Study reflect a growing concern within procurement, engineering and executive roles toward unexpected changes to the global electronics supply chain.

Managing obsolete and end-of-life electronics is a persistent concern in manufacturing and other industries that generally support products for long periods of time. When components are discontinued, companies often have to put money into redesigning products around other parts. Most respondents say they are looking to avoid that at all costs, with roughly a third trying to cut supply chain expenses over the next year.

Those sweating over obsolescence are also likely among the 54 percent of respondents pointing to supplier consolidation as a supply chain risk, up from 43 percent in 2017. Mergers and acquisitions are worrying because the combined companies often shut down overlapping product lines, and unless drop-in replacements are available, procurement and engineering staff will be forced into costly redesigns.

The threat of consolidation has loomed over the semiconductor industry in recent years, as development cost swell and growth flags. Market research firm IC Insights says that the sector generated $107.3 billion of deals in 2015. The total was roughly equal at $99.8 billion the following year. Last year, the pace slowed to $27.7 billion as the pool of available companies evaporated, according to the Phoenix-based IC Insights.

Not everyone shares the same sense of apprehension, according to the 1,100 survey responses. While 77 percent of both engineering and procurement professionals say that the threat of obsolescence is at least somewhat important to their supply chain, only 62 percent of executives agreed. Only 46 percent of executives flagged consolidation as a hazard, as opposed to 55 percent of engineering and 57 percent of procurement staffs.

Many respondents are also wary of potential new sources of supply. Only 14 percent say China’s investments in semiconductor production and design pose an important threat to the supply chain, while another 33 percent say it’s somewhat important to them. The total number of respondents skittish about China’s push into semiconductors from 42 to 47 percent over the last year, the survey shows.

China reportedly intends to announce a new $19 billion fund as part of the nation’s larger push to boost its self-sufficiency in chips. The country’s national strategy has drawn the ire of the Trump administration, which has ramped up criticism of the country’s alleged theft of intellectual property. Last week, the White House imposed tariffs on $34 billion of Chinese goods partly as retribution, while weighing another $16 billion of levies.

To boost manufacturing capacity and close the gap with American technology, the country has slashed taxes for many local companies and funneled funding to them through provincial governments. Roughly one-third of the money is being used to build new factories in China. This year, the Chinese are projected to spend nearly as much on fab equipment – $5.8 billion – as major multinationals – $6.7 billion – operating there, according to SEMI.

The country has also redoubled its efforts in fundamental chip design. The number of Chinese companies that outsource the production of chips they design has grown from 500 to over a thousand over the last five years. And more multinationals are transferring technology to Chinese partners to help muscle into the country’s massive market. These pay-for-play tactics could indirectly erode the competitive of American companies, analysts say.

Survey respondents clearly seem more concerned about unpredictability than they were last year. New companies in China could start splashing around in the supply chain at the same time that the electronics industry is struggling through part shortages, which could worsen as semiconductor suppliers continue to buy each other – all of which creates uncertainty for everyone from engineers to executives.

When I was doing the National Semiconductor Analog Seminar with Bob Pease back in 2003 (Fig. 1), I gave a talk about rail-to-rail amplifiers. I stated that there were two ways to make a rail-to-rail amplifier. One is to use dual differential pairs in the input. The other way is to use a charge pump inside the IC so that the input stage operates at a higher voltage than the power supply to the amp. Pease heard this and corrected me. “There is a third way to make a rail-to-rail amplifier. Check out the LMC6482.”

1. The author and Bob Pease mug for the camera during the 2003 Analog Seminar held in America and Europe.

It turns out the LMC6482 and its quad version LMC6484 are magical parts still widely used. Its front-end stage uses MOSFETs (metal oxide field-effect transistors) hooked up in a very clever way. As the input pins swing from the positive rail to the negative rail, the FETs change type. They transmogrify from being depletion devices to being enhancement devices.

2. Don Archer, designer of the LMC6482, later a technologist for the Silicon Valley Analog group at Texas Instruments.

The LMC6482 was designed by Don Archer (Fig. 2), who recently left Texas Instruments. When the LMC6482 came out, Electronic Design did an article about it, currently unavailable online, on April 16, 1992 (Fig. 3). Back in 2003, I saw Archer in the cafeteria and asked about the part. A supremely modest fellow, he said it was no great accomplishment. “I just threaded the needle” when it came to biasing the depletion mode P-channel input-stage transistors so they would properly change from enhancement to depletion type as in the input voltage changed. Collaboration with the fab led to changes made on the process, which allowed for an enhanced body effect of the p-channel differential pair.

3. This figure from the 1992 Electronic Design article states: By using depletion-mode p-channel MOSFETs for the differential input transistor pair M1 and M2 in the LMC6482 CMOS op amp, the op amp can handle input common-mode voltages that exceed both supply rails.

The Inside Scoop

I had a chance to sit down with Archer in September of 2017, so I asked about the part. Still modest, he said the biasing scheme was not purely his effort. He based it on a suggestion from Dennis Monticelli, who later became CTO of National Semiconductor, now retired from Texas Instruments.

I emailed Monticelli for this article and he recalls, “The predecessor part was the LMC660, where rail-to-rail output was the prime feature along with input precision and gain that rivaled Bipolar or BiFet.  I wanted to implement rail-to-rail on the input also, but the body effect was not great enough in our first-generation Si-gate process to pull it off.  When Don took on the next-gen project, he convinced our Salt Lake fab to make a specific process change that provided the body effect he needed to thread that needle on the parallel input architecture.”

The trick Archer used to get rail-to rail inputs was to tie the body connection of the input FETs to the positive power-supply rail (Fig. 4). As the input pins swing down to the negative or ground rail, the bias across the body will change the FETs so that they become enhancement-type devices. This happens about 3 V below the positive rail.

4. Depletion-mode devices are normally on. You apply voltage to the gate to pinch off the channel. The LMC6482 ties the body connection to the positive rail, so the transistors change from depletion to enhancement as the input pins sweep the common-mode input range. (Courtesy of Linear Systems)

Who is Don Archer?

Archer was one of those wunderkinds that played with radios as a kid, as well as having a paper route delivering newspapers. He repaired typewriters and copy machines after graduating high school. A later job had him climbing smokestacks to repair and calibrate gas spectral absorption pollution monitors made by Thermo Electron Corporation. That gave him the impetus to go to California State University, Sacramento, where he earned a BSEE with a minor in physics. This would serve him well as an IC designer. In 1986, he was hired by National Semiconductor (now Texas Instruments), working in the hybrid group. While there, he obtained a Master’s degree from San Jose State University.

Archer told me, “You don’t survive without being flexible.” That and curiosity are hallmarks of an analog engineer. He also told me a good policy is to seek out knowledgeable people and to understand the boundaries and capabilities of devices so that you can get exemplary performance without “falling off the cliff.”

He noted that designing with circuit blocks invented by others is fine, but you must understand the blocks fully, or they can be dangerous. He said an analog IC designer must deal with thermal considerations, package stress problems, and noise issues. Archer reflects that you not only have to have reliable circuits, but the circuits have to maintain reliability under bias conditions as they operate.

Analog IC design appealed to Archer since the design is more focused on the physics of the transistor than is the case with digital circuits and “one designer gets to do everything.” He pointed out that for very low input bias current, electrostatic-discharge (ESD) protection is a big part of amplifier design. There’s also the problem that any design must have good yields in production, or it won’t be economical to produce.

Archer reminded me that working for a large company like National Semi or Texas Instruments creates a great responsibility to get things right. Back when he designed the LMC6482/84, the mainframe Spice level 2 model simulations did not give an adequate picture of the operation, so he did a piece-wise linear model of the transistors to make sure the simulation results reflected real-world part behavior. Sure, an analog IC designer is an electrical engineer, but also part physicist, to understand the silicon, and part computer scientist, to understand the design and simulation tools. It’s a tough job, but a rewarding one.

Like many clever IC designs, the LMC6482 design is based on a thorough understanding of the analog IC process used to make the die. An example of process and design working together was when the process development group at National Semi figured out how to put JFETs (junction field-effect transistors) into the analog process. This gave rise to parts like the LF411, one of Pease’s favorite parts.

Another example was when the audio group at National Semi (now Texas Instruments) figured out how to use a subset of devices to make the LME49720 audio amplifier that they characterized at 44 V to give a 36-V absolute maximum part rating. That’s better than the typical 30-V power-supply range of many high-voltage amps. This is important to audio folks, since a wide power-rail voltage tends to minimize distortion and improve specs, as well as provide a wide dynamic range for audio signals. Not all of the devices in the process would stand this voltage, but the IC designers worked with the process development group to figure out what transistors and structures could take this higher voltage, and then only used those devices in the part.

Other Methods Go Off the (Rail-to-)Rails

To understand the coolness of the LMC6482 and Archer’s achievement, consider the drawback of the other two methods of making a rail-to-rail operational amplifier. You can use a dual-input stage with both P- and N-type differential input pairs (Fig. 5). This works well if you have a circuit that operates with its inputs near either power-supply rail.

5. One way to make a rail-to-rail input amplifier is to have P-type and N-type differential pairs and gradually switch between the two as the input pins sweep the common-mode range. (Courtesy of Texas Instruments)

However, problems can occur when the input pins sweep across the transition zone. Since there are really two different differential pairs, the part will have two different offset voltages, depending on the location of the input pin voltage (Fig. 6). Worse yet, the particular offsets will be different for any given part, which may make it hard to achieve the accuracy you want. If the input stage is bipolar, the input bias current will also change from positive to negative as the inputs sweep across the common-mode range.

6. A problem with dual-input stage amplifiers is that the input offset voltage will change, depending on where the input pins are in the common-mode range. (Courtesy of Texas Instruments)

The other method, putting a charge pump inside the part, eliminates the problem of having two different offset voltages (Fig 7). Yet the charge pump has problems of its own. The charge pump needs on-die capacitors. This takes a lot of die area and will increase the cost of the part. The size of the capacitors will be related to the tail current of the input differential pair. So now the designer might want to use a low tail current to minimize the size and cost of the charge pump circuit.

7. Another way to make a rail-to-rail amp is to use just one input pair, but feed it with an internal charge pump so that the input pins work at both rails. (Courtesy of Texas Instruments)

The problem there is that the equivalent input voltage noise is inversely proportional to that tail current. The more you starve the tail current, the worse the noise performance of the part. In addition to the input noise is the noise of the charge pump itself. The switching frequency of the charge pump will invariably leak into the output of the amplifier. Usually this noise is at a greater frequency than the bandwidth of the amplifier, and you can filter it out. The filter circuit adds cost, complexity, and other issues, like stability and phase problems.

Another lecture for those analog seminars in 2003 was given by Martin Giles (Fig. 8). Like all of the presenters, Giles exhibited great intellectual honesty. While rail-to-rail parts were new and profitable, Giles would advise seminar attendees that they might not need one. If your circuit operates near the positive rail, you can look for amps with NPN or N-channel input pairs. If you want a circuit that operates near the negative rail (which might be ground), you can pick an old cheap amp that has a PNP or P-channel input.

8. Martin Giles was at the Analog Seminar in 2003. He reminded engineers they might not need a rail-to-rail part for many applications.

Rail-to-rail parts come in handy if your circuit must operate at both extremes. They can also make sense if you have two circuits that operate at each power rail, and you only want to buy and stock one part number. Just remember that each type of rail-to-rail part has its own quirks. You have to understand a little of what is inside the part to know if those quirks will bite you in prototyping or production. Bob Pease loved amplifiers, in part because they were so complex and, well, analog, despite only having a few pins.

At a power-on cycle, inrush current, if not limited, can peak to tens of even hundreds of amps when capacitive load is present, thereby increasing the probability of failure and decreasing usable lifetime of a device. With large capacitor banks, an inrush current limiter is a necessary component.

A MOSFET as a voltage-controlled current device is perfectly suitable to be part of an inrush-current limiter. Power supplies with pulsed current across a current passing MOSFET at increasing duty-cycle rate are very common inrush-current limiter designs with active components. However, such designs stress the MOSFET due to currents often order of magnitude higher than the maximum safe current in a dc mode.

This current-limiter circuit design idea is based on a feedback-controlled current, where current across a MOSFET rises with an increase of the actual voltage across the capacitor bank V_cap(Fig. 1). Based on that voltage, a controller sets the maximum charging current while current limit is controlled through a real-time feedback sense of the charging current.

1. The maximum charging current is established by a controller that measures the voltage across the capacitor bank during charging cycle.

Charging current is determined based on a MOSFET safe-operating dc curve; its maximal drain current is a function of drain-to-source voltage (V_DS). V_DS is the difference between rectified ac input voltage, and V_cap in this design idea. Current rating also depends significantly on MOSFET temperature. For example, the datasheet for the APT75M50L N-channel MOSFET used in this design idea shows that in dc mode, the maximal current is ≈1 A at V_DS = 300 V, and ≈4 A at V_DS = 100 V at a case temperature of 25°C and junction temperature of 150°C.

In the practical circuit design (Fig. 2), U1 through U5 are photovoltaic MOSFET drivers. For the circuit to operate, U1’s driver is first activated via the digital on/off control pin, which in turn drives power MOSFET Q1 (the control element). With U1 active and the capacitor bank at 0 V, the charging current is set by R3, while Q2 acts as current limiter controlling the gate-source voltage of Q1. At this stage, Q3 through Q6 are not conducting and given Q2’s V_BE = 0.48 V, the charging current is set and limited to 0.48 V/R3 ≈ 0.5 A. Resistors R16 through R23 form voltage dividers, providing a measure of the capacitor bank voltage for the quad comparator U6 (LM339), which uses a reference voltage of 2.5 V from the bandgap reference U7.

2. Photovoltaic gate drivers control the MOSFETs, which, in turn, control and step the current flow.

The role of each comparator is to control conductivity of branches with resistors R4 through R7 being either conductive or cut off, therefore setting different current limits depending on the V_cap. With the given resistor values, comparators U6A to U6D will be triggered when the V_cap exceeds approximately 60 V, 120 V, 175 V and 215 V, and will make branches with R4 through R7 respectively conductive.

For example, when the V_cap exceeds 60 V, it will trigger comparator U6A, setting its output low. This turns on MOSFET driver U2, driving Q3 into a fully conductive mode, and the branch with R4 will start to conduct current in parallel with R3. The charging current passing through Q1 is now determined by a parallel combination of resistors R3 and R4 + R_DS (the on-resistance of Q3, maximum value 40 mΩ), while Q2 maintains the same current-limiting function. The new current limit is 0.48 V/(R3 || (R4 + R_DS) ≈ 0.9 A.

A printed-circuit board was built and tested extensively to prove the robustness and efficiency of this circuit. The input voltage is rectified 240 V ac from a 120/240-V ac transformer. The charging current was measured across a 1-Ω resistor with an isolated oscilloscope probe. The approximate maximum current levels were 0.5 A for V_cap of 0 to 60 V; 0.9 A for V_cap of 60 to 120 V; 1.5 A for V_cap of 120 to 180 V; 2.25 A for V_cap of 180 to 215 V; and 5.7 A for V_cap above 215 V.

3. The voltage versus time graph shows inflection points (knees) where the charging current changes.

In the graph of the V_cap during its charging cycle (Fig. 3), the different knee points illustrate different charging currents, quantified with a changing slope of the line. Resistors R1 and R8 through R11 are fail-safe measures that keep the MOSFETs shut off in a case of open-circuit failure of photovoltaic MOSFET drivers. V_CC can be either 3.3 V or 5 V with the given resistor values, or higher if resistor values R2, R12 to R15, and R24 are appropriately re-scaled.

Ilija Uzelac is a Research Scientist at Georgia Institute of Technology, with a PhD in Physics from Vanderbilt University, and a Master’s in Electrical Engineering from University of Belgrade, Serbia.

References

1) Ilija Uzelac, "Feedback-Controlled Constant-Current Limiter Includes Digital On/Off Control,"Electronic Design.

2) Ilija Uzelac and Ron Reiserer, “Series-connected MOSFETs increase voltage & power handling,” EDN.

Numerous test-and-measurement applications require a high-frequency magnetic field. Oftentimes, high field strength is needed. Examples of such applications include bio-medicine research on the effect of a magnetic field on living cells, scientific experiments, probe calibration, magnetic-field interference on electronic products, and much more.

One the most common methods to generate a magnetic field is a Helmholtz coil pair. It produces a highly uniform magnetic field over a large open area. Figure 1 shows a depiction of a Helmholtz coil pair driven by a function generator amplifier. Although most Helmholtz coil magnetic fields are static or dc, increasingly more tests and experiments are requiring ac magnetic fields over a wide frequency range. Obtaining a high ac magnetic field faces a number of challenges that aren’t present with dc fields.

1. The Helmholtz coil pair is driven by a function generator amplifier to produce an ac magnetic field.

Achieving high magnetic fields in coils requires high electrical current. At dc or low frequency, the coil impedance is low and high current is fairly easy to obtain. The coil impedance is usually dominated by the coil’s parasitic resistance, which usually is small. Common power supplies or current sources are available to drive the coil at moderate to high current.

At high frequency, however, the magnetic coil impedance is increased proportionally to frequency. The impedance can be very large, often many times greater than the resistance. The coil impedance, Z, is proportional to the frequency and inductance (see Equation 1). At higher frequency, the impedance can be tens, hundreds, even thousands of times greater than the resistance. It’s difficult to obtain high current with such high impedance.

To calculate the coil current, use Equation 2. The current though the coil is inversely proportional to frequency. For a given voltage amplitude, the coil current decreases with increases in frequency.

I is the coil current magnitude, V is the voltage amplitude, Z is the coil impedance, ω is the angular frequency (ω = 2πf), and L and R are the coil inductance and resistance, respectively. Equations 1 and 2 are for generic coils such as solenoids, Helmholtz coils, inductors, etc. For an ac Helmholtz coil pair, these two coils are connected in series, increasing the resistance by a factor of 2 and the inductance increasing slightly more than 2X (approximately 2.11X for most coil pairs).

In the case of low frequency or low inductance, or both, it’s straightforward to drive high ac current through the coil using a high-output current amplifier such as the TS250. The coil’s impedance is low enough whereby it can be driven by an amplifier directly (Fig. 2). The coil can be modeled (low-frequency model) as a parasitic resistor in series with an ideal inductor. The parasitic resistor resistance is generally small. In the case of Helmholtz coil, two coils connected in series are still modeled as a single coil, but 2X the inductance and resistance.

2. A high-current waveform amplifier is used to produce an ac magnetic field.

When the frequency is very high, though, the impedance of an electromagnet coil increases with frequency as discussed in Equation 1. When a high-frequency magnetic field is required, the coil impedance is very high. Thus, a high-voltage driver is needed to drive high current through the coil.

For example, at 100 kHz, the impedance of a 10-mH electromagnet coil will be 6283 Ω. To produce a high-enough magnetic field, high current is required. If 4 A is needed, the required voltage is more than 25 kV! It will be very difficult and not practical to design a driver than can produce 25 kV and 4 A with reactive power of 100 kW.

Resonant Technique

The direct-drive method shown in Fig. 1 is unable to drive high current into the magnetic coil at high frequency. Achieving a high-intensity and high-frequency magnetic field requires a resonance technique to reduce the impedance.

As illustrated in Figure 3, a capacitor is added in series with the coil. The coil and capacitor impedance are added together; their impedance is calculated in Equations 3 and 4. The capacitor impedance is negative and the coil impedance is positive. When the capacitance is correctly chosen, it acts like an impedance cancellation component. The capacitor therefore reduces the overall impedance.

3. High field strength at high frequency is obtained by using a resonance capacitor to cancel the coil impedance.

In fact, at resonant frequency, the impedance of the capacitance completely cancels the impedance of the inductance. In other words, the coil and capacitor impedances are equal in value but opposite in polarity. At resonance, the waveform amplifier driver only “sees” the coil’s resistance. With only a small level of resistance left in the system, the high-output current amplifier now can drive very high current through the Helmholtz coil or solenoid, even at high frequency. The resonant method allows the function generator amplifier to generate a high magnetic field.

Let’s use an example to further understand how the resonant capacitor can cancel impedance. The coil or solenoid in Figure 4 is 2 mH and the desired frequency is 200 kHz. If the frequency is at resonant, the voltage across the coil is +2.5 kV, and the voltage across the series capacitor is −2.5 kV. Consequently, the total net voltage is zero across the inductor and capacitor combination. The LC, therefore, is essentially a short-circuit at resonant frequency. 

4. Impedance is cancelled by the capacitor.

The TS250 Waveform Amplifier only “sees” the coil’s parasitic resistance as a load. Generally, the magnetic coil resistance is small, which enables the amplifier to drive high current through the solenoid coil with low voltage. The voltage across the coil is still very large. Interesting to note the voltage sum in a closed loop is 0 V governed by the Kirchhoff’s Voltage Law.

The resonant technique is the most practical way to generate a strong high-frequency magnetic field. The only drawback is that it operates over a narrow frequency range near the resonance. To be able to produce an electromagnetic field over wider frequency range, the user needs to change the capacitor multiple times. Usually, a perfect resonant isn’t needed—you just need the capacitor to cancel enough impedance to enable the driver to drive enough current. This allows for operation over a slightly wider frequency window.

Calculate Resonant Capacitance

The resonant condition is when the capacitor reactance is equal in magnitude to the inductor reactance, but opposite polarity as detailed above. Therefore calculate the series resonance capacitance such that the capacitor’s reactance is the same as coil reactance at a given resonant frequency.

Using the above example for 2-mH Helmholtz coils and 200-kHz operation, the series capacitance is calculated as 317 pF.

Choose a high-Q (low ESR) and low-ESL (electrostatic inductance) resonance capacitor to cancel the impedance. The capacitor needs to be rated for high voltage. The voltage rating is calculated by the following:

where I is the peak current.

Using the above example, the voltage rating must be at least 2.5 kV (V = 1 A * 2512 Ω = 2512 V). Add additional voltage-rating margin if higher current is used.

Maximum Frequency Practical Limitation

The resonant technique uses a series resonance capacitor to cancel out the coil reactance; theoretically it will reduce the impedance to just the parasitic resistance. In theory, the frequency and magnetic-field strength can be very high. However, there are some practical limitations.

The first limitation is the capacitor voltage rating. Equation 8 is used to calculate the capacitor voltage rating for a given coil current, inductance, and frequency. If the required voltage is less than 10 kV, there are generally plenty of capacitors to choose from. If the voltage is higher than 10 kV, fewer capacitors are available. As a rule of thumb, the maximum practical voltage is about 50 kV. If higher than 50 kV, other practical challenges such as electrical arc will arise.

The second practical limitation is the capacitance. At higher frequency, the capacitance value is reduced. Generally, a capacitance of 100 pF or larger is recommended. Capacitance down to 10 pF is possible, but parasitic capacitance from the connection wires and the coil itself start to take effect.

Coil Design

The magnetic field in solenoid coils is given in Equation-9 and Equation-10 for Helmholtz coil pair.

B is the magnetic field, µ is the permeability, N is the number of turns, L is the length, I is the current, and R is the coil radius.

A high magnetic field in an electromagnetic coil can be achieved in various ways: increase the number of turns, increase current, increase the permeability, and decrease the radius.

Increase the Number of Turns (N)

In electromagnetic coils such as solenoids, inductors, and Helmholtz coils, the magnetic field is proportional to the number of turns. Increasing the number of turns will result in a higher magnetic field. However, it also increases the inductance and parasitic capacitance. As discussed above, higher inductance isn’t desirable and will require higher capacitor voltage.

Generally, the inductance is proportional to the square (power of two) of the number of turns. For a high-frequency magnetic field, it’s recommended to reduce the number of turns, but increase the current. This way, you can obtain the same field intensity, but lower the inductance and lower the capacitor voltage rating.

Self-Resonant

Increasing the number of turns also raises the parasitic capacitance C_P(Fig. 5). Higher C_P lowers the coil’s self-resonant frequency. In general, the frequency of operation should be 2-5 times lower than the self-resonant frequency (see table below). Lower self-resonant frequency due to C_P will limit the maximum coil working frequency.

5. Inductor model with parasitic R and C_P.

Reduce Coil Radius

Typically, reducing the coil radius doesn’t change the magnetic field for long solenoids, but it will reduce the inductance and C_P. Reducing C_P will increase the self-resonant frequency. Therefore, when designing a coil, keep the radius as small as possible.

In the case of Helmholtz coil, reducing the radius will have three positive benefits. A smaller radius will increase the magnetic field, increase the self-resonant frequency, and reduce the inductance. Lower inductance is paramount, as discussed in the “Maximum Frequency Practical Limitation” section above. Again, keep the radius as small as possible.

Increase the Permeability

For scientific experiments other than an air core coil, a magnetic core can be inserted into the coil to increase the magnetic field. Not all core materials are equal. Some magnetic materials have high permeability, but in low-frequency and low-saturation applications. Choose a magnetic material for the rate of frequency of operation that doesn’t saturate at the desired magnetic-field strength. The magnetic core also increases the inductance.

In summary, use the follow criteria to design AC magnetic coils:

• The coil must be rated for the current and power (heating) handling capability.

              -Low resistance to reduce heating and allow for higher current.

              -Consider resistance increases at high frequency due to skin effect.

• Consider reducing the number of turns, but increasing the current to lower the inductance.
• Make sure the coil self-resonant frequency is 2-5X higher than the working frequency.
• Keep the coil radius as small as possible to reduce resistance, inductance, and parasitic capacitance.
• If desired, choose a magnetic core with high permeability but rated for the working frequency and high saturation field.
• Design the coil to handle high voltage (avoid electrical arc).

Simulation Results

Using the inductor model in Fig. 5, a coil is driven by a ±1V sine wave. In this example, L = 1 mH; C_P = 125 pF; R = 0.5 Ω; Cs = 470 pF; and the working frequency is same as the series resonant frequency of 206 kHz. The coil’s self-resonant frequency is 450 kHz.

6. The inductor is operating at a series resonant of 206 kHz. At about half the self-resonant frequency, the inductor current is reduced due to current “leak” in the parasitic capacitor C_P.

Figure 6 shows the inductor current. The peak inductor current is 1.56 A and the peak C_P current is 328 mA 180 deg. out-of-phase. Contrast that with the 2299-kHz self-resonant in the table—the peak inductor current is 1.96 A with only 20-mA C_P current. Therefore, when the working series-resonant frequency is close to the self-resonant frequency, it reduces the inductor current. Looking at the table’s simulation data, it’s acceptable to operate the coil up to about half of the self-resonant frequency. At this frequency, the coil current is reduced by about 25%. It’s not recommended that the working series resonant frequency be higher than half the self-resonant frequency.

Caution: Potential Electrical Shock

The high-current electromagnetic coil discussed above can store enough energy to become an electrical shock hazard. Make sure all electrical connections are insulated with high-voltage insulators. Wires must be rated for voltages discussed earlier. Always disable the amplifier output before connecting or disconnecting the coil and capacitor.

Conclusion

A high-current amplifier driver is needed to produce a high ac magnetic field. When a high-frequency magnetic field is required, the resonant technique will reduce the coil impedance and allow for high current to drive the coil with a low-voltage function generator amplifier.

The resonant technique is the most powerful way to generate a high-frequency ac field. At high frequency, the practical limitation is the availability of high-voltage capacitors. Another limitation is the magnetic coil self-resonant frequency. Furthermore, the self-resonant frequency should be 2-5 times higher than the working resonant frequency.

References:

Waveform Amplifier for Function Generator

Helmholtz Coil

High-Frequency Electromagnet Using Resonant Technique

For years, mobile carriers funded initiatives like distributed antenna systems (DAS), repeaters, and smalls cells for their enterprise customers to improve connectivity throughout buildings. The logic was simple: In exchange for absorbing the hardware costs, enterprises would agree to equip all of their employees with a particular carrier’s devices. A building’s telecom closet served as a de facto hub for all connections, but upgrades to equipment were largely driven by, and depended entirely on, carrier initiatives. Over time, these closets grew into odd-looking bundles of unlabeled wires that carry a slight resemblance to medieval dungeons.

Today, a number of factors, such as bring your own device (BYOD), Public Safety, 5G, and the innovation bands, are propelling shifts in the telecom value chain. To be prepared, enterprises must now look for ways to future-proof their connectivity initiatives, and incorporate designs built on flexibility and modularity.

Enterprise-level BYOD programs have dramatically shifted the onus of DAS away from individual carriers. Enterprise IT now needs to identify which cell providers are most common across an employee base, and then support the frequencies and technologies of that carrier, or more likely, more than one carrier. These costs aren’t static, but instead change regularly.

For instance, some carriers are building out their network on the 600-MHz spectrum owned by the company. Prior to BYOD, it would have provided connectivity upgrades to its enterprise clients as part of its rollout plans. Today, however, it’s up to IT departments to understand the changing landscape of connectivity, and adjust to it accordingly.

Onset of Transformative Technologies

As carriers begin to roll out true 5G solutions, the number of improvement and product rollouts will increase. Mobile carriers are expanding trials throughout the U.S. (and the world), and a U.S. rollout will require most buildings to revamp their connectivity solutions to meet their employee and client needs; 80% of mobile traffic is generated in-building. The success of 5G and high-speed wireless connectivity might ultimately rely on what’s in a telecom closet, an irony that can’t be understated when discussing ubiquitous wireless coverage.

5G isn’t the only transformative near-term technology that’s about to become mainstream. Public-safety communications, those that take place between first responders such as firefighters and police departments, are undergoing change as a result of a governmental decision, namely FirstNet.

Today, in order to obtain a certificate of occupancy, buildings are required to have certain levels of connectivity available for first responders. However, the actual regulations and requirements that must be met to accept tenants can vary wildly between cities and states.

That situation is changing, though, as the U.S. Government has awarded AT&T a contract to build FirstNet, a nationwide LTE network based on 700/800-MHz spectrum that will offer interoperability between states. In practical terms, this means Nevada firefighters who assist in a California wildfire will have access to the same communications band as the California responders, which will help both work together in times of emergency. 

It’s unclear as to how rapidly FirstNet will roll out, and how each regulator that currently oversees building connectivity will respond. However, the idea that jurisdictions will require blanket connectivity for first responders using AT&T’s spectrum seems likely to transpire soon. For the buildings and IT departments that handle telecom closets, this is a new wrinkle, and one that must be addressed.

There’s a third technology being utilized that also requires flexibility in design: The launch of CBRS and LAA bands for connectivity. The “innovation bands” are gaining traction in numerous ways. In many instances, enterprises utilize this unlicensed spectrum to build out their own, private wireless networks.

5G, FirstNet, and the innovation bands all point to the same thing: Support for new devices and bands will require new hardware, placing the onus on IT departments to find ways to future-proof today’s enterprise investments. The organizations that have been successful have done so by focusing their purchasing on modular solutions, those that can be expanded through simple plug and play.

The Modular Way

A modular design approach was incorporated into ADRF's ADX V Series DAS solution. (Courtesy of ADRF)

There are three guidelines that IT departments should follow when incorporating modular approaches.

• Interoperability is critical: It’s important to select hardware from vendors that offer interoperability among their peers. This ensures that telecom closets are equipped to meet an enterprise’s unique needs. For instance, not every DAS provider has support for CBRS or LAA bands, meaning that the only ways to support these bands is either to replace entire legacy systems, or find interoperable hardware. The latter offers tremendous cost savings.
• Size matters: There’s a limited amount of space in a typical telecom closet, meaning a finite number of cubic feet available for hardware. As such, the actual physical size of equipment must play a role in the decision-making process. In some instances, there are solutions available that combine various products, such as a DAS head end and repeater, into one smaller box. These products can save space and enable IT departments to maintain flexibility for the next upgrade.
• Remote control: As telecom closets become more packed, it’s important for IT and engineering teams to have oversight on how equipment is performing via some sort of remote mechanism, either web-based, or ideally, via mobile application. In addition to monitoring that includes real-time alarms, the mobile control should allow for simple and incremental adjustments, as well as additions to a network. For instance, if a company needs to install a new repeater, the mobile control should be able to recognize and activate the hardware once it’s installed.

Fortunately for IT departments, there are systems integrators who have extensive experience in the telecom markets, and can offer advice and implementation alike. These integrators should be able to share best practices on modular design, and explain how to future-proof connectivity solutions.

Connectivity is undergoing massive transition, as technologies such as 5G and private wireless networks built on LAA and CBRS take off. At the same time, enterprise IT is being asked to support a wider range of devices that all need a connection. The only way they can be successful is by incorporating a modular design into their approach.

William Wong is DAS Engineering Manager at Advanced RF Technologies Inc. (ADRF).