October 2006

Microwave Career Killers

Mel visits Moonshadows on the Pacific Coast Highway

This page will help guide your career path in the field of microwave engineering. What technologies should you stay away from if you want to avoid precipitated upon by a future pink-slip blizzard? Without a single doubt, this page covers a topic that will NEVER be covered in any trade journal, IEEE paper or government solicitation. What can we say, we're there for you!

Career Figure of Merit

If you are looking for career enhancement, stick to the principal that cost reduction is just as important as performance improvement when you are developing technologies. We define the following Figure of Merit:

Career FOM=(Improvement in Performance) / (Relative Cost)

You can express this in decibels: using the example of RF CMOS, you might achieve 1/10 the price but at 1/10 performance, and your career will advance 0 dB. But if you were to perfect gallium nitride on silicon and process it on 8 inch wafers, the performance might degrade 1 dB compared to SiC but the cost will drop substantially and you may well achieve a 3 dB career FOM.

After we review these dead or dying technologies, your homework assignment is as follows: practice in front of a mirror repeating the response that is needed during meetings when a particular career-killing topic comes up... instead of just saying "no way Jose", just say "good one!" but try to minimize the smirk. Then if you are called on the carpet later by Mayor McCheese, you can play your comments either way, either "good" in the sense that Michael Jackson used the word "bad", or good as in, "not altogether crazy". Remember never to say "not bad" in front of any Limeys, or they will surely misinterpret this as the highest possible praise. Can you dig all that? For this excellent use of the quirks of the English language, we dedicate this page to Heidi and Eric who are masters of many two syllable responses.

It is only fair to warn you, the information provided here was surely fueled by one of "Mr. Gordon's excellent products" (to quote Hemingway). If you disagree, or have some other suggestions, chime in mate, or sponsor a page on your particular sacred cow and maybe we'll change our opinion. Our apologies if we step on someone's sacred cow and they can't take a joke, but better you hear about what people really think here at Microwaves101 than you hear from Mayor McCheese later when he figures out he's been wearing the Emperors New Clothes. There's an image we didn't need!

Like Charles Dickens' best work, we'll divide the topic into three tenses: past, present and future. Perhaps another way to look at these technologies is in the three chronological phases of technology: unbridled enthusiasm, backpedaling, despair, and blame. Hey, that's four!

The ghost of career killers past

These are technologies that you might not even have heard of if you are young.

IMPATT diodes (so called "solid-state transmitters")

IMPATT diodes have some interesting properties. Under certain conditions they can amplify a signal by about one dB. So to make a transmitter, you need to efficiently combine a chain of one zillion IMPATT diodes, using a low-loss technology such as machined waveguide. Forgedaboudit. Figure out another way. You are out of your mind.

Josephson Junctions

Does anyone even remember this stuff? Or was it just a nightmare like cold fusion?

Update, July 2011. Here's the answer, according to Colin (thanks!)

So you asked whatever happened to Josephson Junctions...... here's the latest:

Yep, quantum supercomputers use them. And LockMart (home of everyday low prices for all your defense needs) just bought one.

Aluminum silicon carbide (AlSiC)

AlSiC was received with great fanfare in the 1990s. Net-shaped housing with matched CTE to GaAs. Where are the purveyors and early adopters of this technology now? Haven't seen them!

The problems that have persisted include the inability to maintain a minimum aluminum skin thickness. Wherever the silicon carbide pokes through, plating won't stick. Poor plating means no chance of hermeticity, because you will have to braze in some feedthroughs because you sure can't fire in glass above the melt temperature of the housing. Also, the reality of net-shaped eventually meant "near-net-shaped", which means you might as well go to a machine shop, but you should pick something to machine that won't ruin your tools like silicon carbide will. Recall that silicon carbide is the main ingredient in sandpaper!

Integrated features on alumina

Over the years several thin-film vendors have tried to move up the food chain, offering filled "hermetic" via holes, airbridges, multilayer interconnects and even MIM capacitors. Guess what? This technology is low yielding and tres expensive.

Automotive radar (a.k.a. adaptive cruise control or "car radar")

New for September 2009! It is long past time to request a fork and spoon to eat our words. Not that long ago, automotive radar looked like a dead end. Then along came RF silicon. Now with 90 nanometer (or less) CMOS and 130 nm silicon germanium, it's now possible to create a radar cheap enough for the car market. We had all but forgotten about the topic, until Ben sent us an email. These are his words:

I would like to doubt that Automotive Radar is a microwave career killer. You can you buy a fully integrated automotive radar chipset in silicon from Infineon, Freescale trying to catch up, Fujitsu trying to do it in CMOS. Third generation automotive radar is on the market and is delivered in the Porsche Panamera designed by Bosch and including the Infineon Chipset. However, I think this is not the kind of mass selling car, but this system will be delivered in new Audis too.

This is what we originally said... Maybe car radar should already move to the "past tense". Sure would be nice to be able to kick back and let some Gunn diode apply the brakes when you're creeping up on the car in front of you. Better yet, why don't you put down the cell phone and pay attention to what you are doing?

The automotive radar idea has been around for decades. As near as we can tell, no one in North America ever made a dime off of this. One problem is that radar companies are used to dealing with juicy defense contracts which pay cost-plus for development. If you ask Ford for such a contract, they will respond "good one!"

Anyone else care to join in on this discussion? Here's Uncle Jam's take on this:

The career killers page is pretty spot on, but I have to disagree with you on Automotive Radar... first, building a whole radar system on your own is a pretty impressive point on your resume, and extends to other job sectors. I think you'll agree when I say that radar is generally used on things other than cars! This is what I mean when I say that an Automotive Radar is a system, it isn't locked in to any technology choice. I can assure you, that I've seen and done evaluations myself on forcing in no less than 7 of the technologies you list on the career killers page into this car radar! Based on this experience, I can say you're pretty spot on with everything else on that list.

However, these days alternate ways of doing things are really starting to transform the radar from something of luxury and long haul trucking to everyday driving. SiGe and CMOS development are starting to catch up with our frequency band (and as already pointed out on your site, SiGe is already in some current generations of radar). It's sort of like Automotive Radar is the St. Peter of microwave technology, deciding what sinks or swims.

As examples of end product from these advances, radar is now available on Ford Taurus and Toyota Prius cars. These cars aren't 'cheap', but they're not Mercedes either. VW and Audi have their own ones now. Also, the radar is not a single application of Automated Cruise Control. They are also used for Blindspot detection, and collision preparation. These applications are much cheaper than ACC because a lot of the supporting electronics used in ACC (like the car controlling the brakes and reading many other necessary sensors for the algorithm) are not needed. Collision preparation looks like it will become very cheap soon. This helps out by readying the car before the accident happens, instead of passively detecting the collision and taking advantage of the speed of light. One advantage I'll point out is that the airbag can deploy slower and avoids slapping people in the head. Another is emergency braking assist. Don't be surprised if in the future there is a push to have the government standardize sensors like these. Although with congressional hearings on EMI and car electronics, who knows!?

After watching the Toyota sudden acceleration debacle of 2010 unfold, I really don't want any electronics involved in braking decisions... also, with ROHS eliminating lead from electronics without a suitable substitute, the chances of any electronics working perfectly as a car ages grows less and less. Cheapskates like me have been known to drive cars for decades. And in reality, a good number of careers suffered trying to make a car radar out of discrete components prior to the latest silicon revolution. - UE

High voltage GaAs

For a few million bucks you can always tease another volt of operation out of GaAs pHEMT power amps. Today's GaN transistors will operate at 50 volts, and high-voltage GaAs pHEMT has effectively ended, suffering the same fate MESFET technology. Ask the former employees of Cobham's Blacksburg GaAs foundry....

GaAs for anything other than RF

Because GaAs is so "fast" you'd think it would be used to create the world's fastest computer. Some companies spent a lot of research money on this. As it turns out, they didn't follow an important corollary of Moores Law: anything that can be made of silicon, should be made of silicon. There's no way that a compound semi is going to displace silicon for computing power.

Today's career killers

These are technologies that still persist in the industry, but have lost some of their original luster. Sort of like an aging stripper that keeps on keeping on because she hasn't bothered to develop any other marketable talents...

Ultra Wideband (UWB)

This tip also came from Ben, the same engineer who debated our car-radar is a career-killer hypothesis.

I think UWB (ultra wide band) is such a career killer. It was a big buzz word last years, we did not hear much about the topic. Of course, there are some applications where you can buy it. They all have one thing in common: they don't have much to do with the idea of a pulse based system. Systems on the market only can achieve the high datarates with OFDM modulation, which is not the basic idea of UWB.

Low temperature co-fired ceramic (LTCC)

Note added April 2017: LTCC has come a long way since this mess was written, many of the original problems such as shrinkage and cost of gold have been solved (replace gold with silver...)

LTCC offered to provide us all with a low-cost, three-dimensional interconnects for microwave modules. No doubt about it, some of the coolest hardware to ever be demonstrated uses LTCC. So what is the problem?

Shrinkage is one issue. Or rather, shrinkage tolerance. If every circuit shrank the same way, every time, maybe you could talk about high yield. Or you could just put up with it and accept low performance.

Post-plated metal is another problem. The cofired metal doesn't always have the properties that one would want, for example, for soldering or wirebonding.

Last we checked, the price of gold was over $1000 per ounce. Imagine wastefully screen printing thick-film gold interconnects, including solid-filled vias, whose conductivity is significantly reduced by organic binders that must be burned out, such that you can never compete with the loss characteristics of the cheap copper used on soft boards. It's no wonder that the LTCC scrap pile in your factory is always empty, the employees are stealing it, selling it to reclaimers, and spending the money on Las Vegas High-Roller vacations!

But the primary problem is: ceramic sure is expensive compared to soft substrates like PTFE... so what happens when you evolve to a multilayer ceramic? It gets more expensive! What happens when you evolve to multilayer soft-board? You kick ass and take names! If someone tells you LTCC is affordable, you can quote us here... "good one!" Sure you can make some hero results with LTCC, but it will never be cheap. Look for a soft-board solution.

Superconductive filters

Pretty much anything involving superconductors and microwaves belongs on this list. Superconducting filters sound attractive as a way to minimize system noise figure.  Just try to buy one. Then try to cool it. Note that as you move up in frequency, superconductors no longer exhibit superconductivity.... at 100 GHz you can get the same surface impedance with copper.

Ferroelectric tunable filters

Tunable filters are probably a good career path, they are the key to reconfigurable RF front-ends. There is more than one way to skin that cat, ferro-electrics were an early contender but try to find someone that mentions this skill on their business card.


It is fair to say that MEMS may yet prove to become a billion dollar per year microwave component, especially in the past "00" decade when many of the shortcomings seem to have been addressed. But so far we have been less than impressed with the notion that MEMS was going to revolutionize microwave industry. Wake us up when this happens! Check out the Riddle of MEMS.

Here's a musical tribute to stiction!

Future career killers

These technologies are still in the "enthusiasm stage" so they might represent sweet music to you for now, but musical chairs on the associated change number will commence sooner or later.

Carbon nanotubes

CNTs: the most unfortunate acronym of all, there to remind us about the joke about the difference between a pack of girl scouts and a tribe of pygmies... Here is a lab curiosity that is looking for a problem to solve... is it a transistor? or a heat sink? or a mixer?

From Uncle Jam:

I once saw a guy simming in HFSS a patch antenna made out of carbon nanotubes without any substrate. When I asked him about it he told me they had no idea how to grow them onto a substrate so this is what they were settling on!

Actually, there has been some real progress in CNTs in 2013. HRL Labs actually make a CNT transistor that pinches off. They must have cheated, as CNT has no bandgap. But wait, that's not true according to Gabriel:

I really appreciate all the work that has gone into Microwaves101 and it has helped me a lot the past couple of years (We wanted to use superconducting stub tuners to investigate single molecule devices at 4-8 GHz). I just happened to enjoy reading the career-killers page and saw you mention CNTs (which I'm working on now, we can actually impedance match a CNT device with a resistance of 200 kOhm), where you stated CNTs have no bandgap, which is not true, they come metallic (yes, no bandgap) and semiconducting (has a bandgap), when just growing them, 1/3 will be metallic, 2/3 semiconducting.

I know it's pretty close to nitpicking, but I thought I'd tell you anyway. And yes, I guess it will be another 5-10 years until CNTs will move from the realm of research (1 in 20 devices working, everything "handmade") to an area where one can actually do engineering.

Career FOM: -3 dB

An idea competing for nanotube attention is the space elevator. Nanotubes are so strong that they could form a tether from Earth out past geostationary orbit, if only you could make them 140,000 kilometers long rather than one micron. Climbing up the tether would save massive quantities of rocket fuel and pollution (did you think there was a catalytic convertor on a Proton rocket?) In the best scenario, this idea can be used in a future science fiction movie, it is at least plausible, unlike "warp drive". Here's the start of a screen play: a shipment of medical cannabis is needed desperately by the homeless shelter that was once the International Space Station, and the Space Elevator is called to action. Unfortunately, the Y10K calendar issue causes a mishap, sending cars up and down the same tether at the same time, similar to recent terrestrial train incidents. After a space hacker saves the day with his Google glasses, a secret plot to snatch the weed by aliens from the planet Bong goes awry when Space Lassie barks at them....

Anything that involves diamond

As Fritz recently pointed out, diamond thin film attenuators and resistors are a great application of CVD diamond.  If you are designing a solid-state power amp for graceful degradation, the isolation resistors get pounded with power when one amp fails.  Without diamond resistors, combining 20W or higher power GaN power amps with wired combiners becomes a lot harder. In this case, diamond provides much better performance than beryllia (poisonous) and aluminum nitride, and you will have to pay whatever the diamond resistor vendor asks.

We've seen two ways to use diamond in microwave technology both of which are potential career killers. The first is as a wide bandgap semiconductors. But diamond doesn't offer a whole lot higher performance than silicon carbide, which is cheaper by a factor of Avogadro's Number.

The second microwave application that diamond might be bandied about is as a low-cost heat sink. Hey, diamonds are called "ice" for a reason, this crystal has staggeringly high thermal conductivity. But when deposited as an amorphous film, the reduced thermal conductivity might not be that much better some cheap metal alloys.  Diamond has very low coefficient of thermal expansion, something to think about when you are mounting GaN power amplifiers to it.

Last, any industrial application of diamond will remind the public that diamond is just carbon, and should be cheap. Then you will have to deal with the DeBeers family, which wants us all to pretend that diamonds are worth multiple paychecks per carat. Some South Afrikaner with a crooked nose might come and break your knuckles for messing with diamond's artificially high price. Yikes!

Career FOM: -2 dB

On-chip MEMS tunable matching networks

We can't wait to buy that 2-18 GHz power amp with 50% efficiency! Too bad we'll have to package a computer on top of it to optimize every possible operating point! Heck, throw an onboard network analyzer on that as well. Maybe Agilent will help us out and give us each a free copy of Eagleware to run the optimizer!

Three-dimensional MMICs

The premise here is that you could drive down the cost of a MMIC power amp by using Group III-V semiconductor only for active devices, and stacking matching networks on multiple top layers. Good one!

You might notice the use of the word "three-dimensional" implies horizontal and well as vertical interconnects. And transitions thereof. The devil in these details is that vertical/horizontal transitions always stink. Keeping everything in the horizontal plane will give better performance, trust us.


It is hard to take seriously any technology that might enter a conversation thusly: Remember on Star Trek...? Some day metamaterials will enable us to wear clothes that make us invisible. Good one! Here's some meta-discussion from Uncle Jam:

About metamaterials: they're going to end up like MEMS. There will be a very small niche area where they work well in, but most of the things they were promised for won't materialize. For example, MEMS can be found in very small microphones, useful for laptops and phones and the like. There are a few bulk metamaterial ideas that do work and provide some advantage over existing technology, but they're going to go into the black box and you won't really hear of them unless you directly work on them. This paper pretty much made metamaterials a complete joke for me:


It appeared on the BBC in their science section, unfortunately. Basically these guys just made a conformal matrix transformation with electromagnetics. They keep publishing papers with it. This one is really bad. If you smash a woman down to half her size, encase her in some lossless metamaterial (lossless metamaterial, ha!), then a guy, who also has a clone smashed into the box, stands next to the box, he looks like a woman. Hong Kong has certainly heard of dressing in drag, hasn't it?

Wearable electronics

The whole topic of any wearable technology is pretty lame. Below is a picture of the evolution of wearable computers which we scarfed from Wikipedia. One thing that is incorrect about the picture, is that the "wearer" should be getting progressively fatter with each decade, ending up in the "10s" sprawled on a bed 24-7, unable to wipe himself. But he can watch 3D porn at the next department meeting, thanks to his magic glasses!

We wrote that a few years Google is involved. Maybe we spoke too soon.

Regarding wearable antennas (the portmanteau is mantenna), from Uncle Jam:

One technology that I keep seeing at conferences and such but never a real product are wearable antennas. I can't say I know much about the subject, but it seems to me that since the physical size of an antenna defines its operation characteristics, it seems counter productive to make one that is always changing in shape. Can't you just move to a higher frequency and make the antenna small enough to not matter? Especially since this one is military centric and they can move into different frequency bands much more easily...

A counterpoint from Jim (December 2017):

Uncle Jam says something to the effect that military radios have more opportunity to switch bands (than commercial radios, it is implied).

I have to disagree with that statement, after working on tactical radios for the last 10 months at XXX (after taking a detour into the commercial world for about 20 years). Nope, we are pretty much told what bands we can use. Previous generations of UHF tactical radios used 420-450 MHz, right in the 70 cm ham radio band and right where that massive Pave Paws radar system operates. Now we go 225-450 MHz (wideband!) I don’t think we have issues with Pave Paws, but we do have a big tower with some repeaters on top of it at the YYY site where I work. Gotta keep the local hams happy by being a good corporate citizen (in the current PC terminology).

But yeah, I agree that wearables are yet another technology that created a big stir but went nowhere. I was considering a wearable computer back a few years ago if the price would come down to something reasonable below a kilobuck, but the smartphone kind of supplanted that. Passe’!

Unknown Editor's wearable musings: now you can buy dog-tracking and kid tracking GPS wearable devices to help treat your paranoia.  Maybe you one of those people that back into a parking space every time you park?  Are you proud to "carry?" Now you will aways know where all your dependents are, just like the Department of Justice tracks Paul Manafort's movements. Hey, what happened to Google glasses?  The perfect gift for the perv in your life!

Please send us more ideas for this page, and feel free to disagree!

Check out the Unknown Editor's amazing archives when you are looking for a way to screw off for a couple of hours or more!