The Amp Hour Electronics Podcast

A weekly show about the trends in the electronic industry.

You are here: Home / Radio Show / #20 – Military Electronics and The Free Eagle (Freagle) Campaign

#20 – Military Electronics and The Free Eagle (Freagle) Campaign

Play

Podcast: Play in new window | Download

Subscribe: Apple Podcasts | RSS

That’s all for now. Please submit your (audio or text) questions/suggestions at the suggestions page and your pictures at the WOTW tab if you’re interested in being featured!

Comments

  1. LOL “What? you want to be a patent lawyer? Why not go intern for Beelzebub while you’re at it?”.
    Bodypart tools – i’ll admit i use my top lip as a radiated heat temperature sensor 🙂
    I’m sympathetic to Dave’s views on home chipping but the pie-in-the-sky smells too delicious for me not to encourage those who think they could, to have a go and try to miniaturise and kit this.
    What would satisfy Dave’s benchmark of something practical?

    • For those that missed it, I fully encourage people to go out and build their own chips too. Whilst it might be great fun, just remember that you can buy an FPGA or micro, or hundreds of thousands of other chips for only a few cents to a few bucks :->

      My benchmark I guess would be a fully formed and characterised chip in a usable package that you can be used in a non-niche serious prototype product.

      But even if that did happen, let’s be realistic. It’s not just about the practical manufacturing aspects and price either, it’s also about how it’s going to be “game changing” What universal benefit does it offer designers?

      Lets say someone comes up with a magic MakerChip machine that churns out custom prototype chips easily (bucking the trend of basically every chip manufacturing methodology, and assuming it can’t churn out production quality chips of course).
      Chris’s latest argument is that would be great so components don’t go obsolete. Yeah, maybe, fair enough.
      Same thing goes for other arguments like doing something unique, or quick time to market, or some other argument – IN THEORY. But will it become popular and change the game?

      Let’s take the Makerbot example again. It is useful and game changing because a good majority of useful mechanical items it produces NEED to be custom made in the first place. You can’t make your sexy looking custom watch case of lego bricks or existing parts from the hardware store for example.

      Electronic products that use chips on the other hand are a different beast, opposite in fact. The vast majority of products are made using existing parts, lego-style. That is why the parts market has evolved to produce millions of different parts, and why methods have been developed to stick them together (PCB’s, Pick’n’place etc)
      So what universal good is a machine that can churn out a custom prototype chip?, when you already have hundreds of thousands of proven ones to chose from at low cost?

      That leaves nothing but an incredibly niche market segment for such a machine.
      So I argue there is simply no place or market for a MakerChip machine, even if it was possible to produce one in the first place.

      And that’s leaving out other practicalities like what you do after you design and build your prototype chip in your MakerChip machine. Oops, you likely have to do a lot of work, and spend a lot of time and money in some way to get someone else to make it in volume production for you – that sucks. Not to mention retesting and re-characterisation etc.
      It more than sucks, it’s a reason to AVOID using such a machine if one even did exist! You are much better off designing your product from established parts, just like has been the progression of electronics design for the last 50 years. And the universality of chips like FPGA’s and analog arrays etc makes the need for custom chips even more redundant as they get better and better. Ultimately to the (theoretical) goal of there being no need for custom chips any more, thus rendering the MakerChip machine redundant before you even get started!

      One of the golden aims of electronics design is trying to avoid custom, or single source, or obscure parts etc in the first place. And if you don’t know that, you haven’t been in the game long.
      Yet here some people are advocating that a machine that spits out such custom parts would be revolutionary?! (yes, it would from a purely technical achievement sense of course, but not much more)

      Mechanical parts on the MakerBot on the other hand are easily and painlessly transferable to full scale production, and indeed such custom tooling is the normal desired goal of that industry.

      But the electronics industry ain’t like that, it’s the opposite, so I simply can’t see the MakerChip revolution ever happening!

      So get over it people and put your energies and dreams into something more practical.
      Unless you want to give it a go fun of course, in which case knock yourself out.

      • Imagine the following:

        You are knee-deep in your project, and miss a few chips for building your first prototype. You have a MakeChip machine. Instead of ordering the chips and awaiting delivery via mail you go to a semiconductor “iChip” store, order the blueprints for the missing chips (DRM protected, meh…), pay a few bucks per blueprint via paypal and download the blueprints into your MakeChip machine.

        The machine rattles for a few hours, complete with blikenlights, and steampipe, and out falls your chip.

        You don’t want to pay for the expensive blueprints in the iChip shop? Then you go to http://open-chip-blueprints.example.org, look through the heap of junk people have uploaded there, if there happens to be an open chip blueprint you can use. You download it into your MakeChip machine …

        Maybe not in my lifetime, maybe the thing won’t be called MakeChip, but a replicator 🙂

  2. Regarding the military industrial jobs, I have had this argument with my father. My uncle has been working for Northrup Grumman for decades and my father suggested that I should strive for a position within the company.

    In my opinion, I would not work for any company in the military industry simply because I could not watch the news and live with myself at the same time.

    • It is extremely difficult to completely stay away from military products in the electronic industry. Many things can have a civil and a military use.

      For example, communication equipment. There was a reason many countries treated their normal telephone network as an asset for military use, too. If you were a phone network technician, how sure could you be that your phone switch wouldn’t once be used for military purposes in a state of war?

      Or you are an IC designer. You have really no control over what your IC will be used in.

      What you can try is to distance yourself from military use as fare as possible. But it is very difficult to stay completely in the clear. You will stay in a gray area, and have just a little control if it is a dark or a light gray.

  3. Sorry Dave, we will all be printing our own chips in the future. There are start-ups already getting into this,
    http://www.kovio.com/index.html
    Eventually we will be able to download an 8051 core, plop it into a design and print out the chip. The IP will be pre-characterized, or done through software, much like FPGA design. Making chips with current technology is not realistic, but the future won’t use current technology.

    Also, I’ve never worked for the “military industrial complex” but think many need to get off their high horses on these types of issues. If you design weapons for Kim Jong-il, that’s evil, for a peaceful democracy, no way. New technology saves lives, both of the soldiers and of innocent civilians. Politicians decide when wars will happen, and they will happen one way or the other. Taking the ideologically “pure” high road means greater death tolls. Being able to laser in on a terrorist camp is superior than dropping dumb bombs out of the belly of a plane flying hundreds of miles an hour over a broad area, or even sending thousands of people running at each other with swords… thank God for engineers! Defense technology is also a deterrent, making some politicians less likely to start wars.

    Also, in defense of the US, our defense spending as a percentage of the GDP is not especially high. Our absolute spending on just about anything is huge because our GDP is huge. And many of our foreign friends don’t need defense spending because… they’re friends with the US.

    • No high horses here about war, just personal choices. Your point that wars “happen one way or the other” is an interesting stance. In that regard it’s a preventative measure.

      Thanks for that link though. I really think Dave likes being the contrarian on this one though, so I indulge him. And fight him. But really, no one wants to back Dave up?? I’m starting to feel bad for the guy.

      • I like to think I’m being sensible and have weighted up the evidence fairly reasonably. Once again, it seems no one wants to argue my points, they are just off in fairly land.
        I have better fairly land dreams, like a hover board for instance.

  4. I’m in the people-kill-people camp so have no reservations about working on mil tech. I actually lament that I’ll probably never get the chance. It is my dream to work on cutting edge mill tech and it will probably never happen.

    There are no major military engineering jobs in New Zealand. Now I’m in Aus (just finished ME) staring at all those posting for grads to join on SEEK, I can’t apply since I’m not Aus citizen so have no chance of getting the needed security clearances.

  6. Freeagle:
    That would be great, but personally I could not care less. I have made one board with eagle and only thing good about the software is that so many people use it and it has a free limited version. The part drawing and adding is really made so hard to do that i just gave up on it. I believe most of the people that use it just use parts that are already in library. But for real work you need to either draw parts for every project. I use PADS at work and DipTrace 4 layers licence on personal projects.

    Rambus:
    I have mixed feelings about the Rambus case. I was really disappointed in 2002 when DDR came and all motherboards suddently started using 266MHz ram. What a downgrade from 400MHz… It took long before DDR reached higher frequency that was a step down in tehcnology. But of course Rambus had too high prices for the memory so the lost the market. I still have a motherboard that has 1GB of RD ram attached in storage room. I paid a lot of money for that ram so just don’t want to throw it away.

  7. I’m totally 100% siding with Dave on this. The QC required for making an IC is totally out of the realm of possibility for small scale. Vibration management would be a nightmare. Airborne particle management would be a nightmare. Static control would be a nightmare. Wirebonding would be a nightmare.

    And even if you get past all of that, the consumables cost will bankrupt you. I never heard Dave mention that element, but getting a good, usable wafer is not something you can do with play sand from the hardware store…

    • Yay, that’s one!
      Yes, there are many practical considerations I have not mentioned, and I’m sure there are many more if you get down to the nuts and bolts of it. But seeing as that I do not have a background in chip manufacturing I’ll leave that to the experts. Suffice it to say I know it’s complex, that is why companies like Kovio are trying to find new and different ways to do it. Because the existing methods are not practical which is what I’ve been saying all along, yet Mike is the first to recognise this.
      The MakerChip will have to use new and probably as yet uninvented manufacturing technology. But even then, you have the other problems to overcome as I mentioned above.

  8. Hi guys,

    Dave was correct it is a fluke 19 and yes it still works. It did fail early on, but I sent it away for repair and and it has not failed since. The Scope is a Tektronix TDS 2014B Mains powered not Battery. The Esr meter is our favorite tool, we had 3 of them that is the last survivor.

    Mike.

  9. I have to go with Dave on this one as well. Theoretically you could pull off the litho work, sputtering, passivation, and the rest in a small shop. Even die bonding isn’t all THAT complicated (did you know that in the 60’s it was all done by hand?) provided you could source the gold wire. However there are two killers; First is sourcing raw silicon wafers of sufficient purity to give a decent yield, and second is the EPA (or other health/safety folks) coming down on you like the fist of thor for the stunningly hazardous assortment of chemicals you just ordered on the internet.

    I really hope some tries it. I really really hope they live quite a distance away from me. 😎

  10. The current problem in my eyes are substrate supply. Where can you easily get your hands on sillicon to work with?

    De-ionized water is available.

    Alcahol is available.

    Doping agents are available.

    Masking agents are available.

    Home made temperature controlled ovens are available.

    Conductive epoxies, and welding can be done at home.

    My big question is “why.” The sizes we could do this at would make very large chips. Why wouldn’t you use a FPGA or a micro-controller to simulate the functionality.

    I think it can be done. I don’t think the results will be very sophisticated. I think it can even be done fairly cheaply.

    • My big question is “why.” The sizes we could do this at would make very large chips. Why wouldn’t you use a FPGA or a micro-controller to simulate the functionality.

      And therein lies the rub.
      MakerBot = Simple to do, and has a massive and obvious “why” factor = A popular game changer
      MakerChip = Ridiculously complex, and little upside in the “why” factor (that diminishes by the year) = A lost cause

      Anyone still want to join Chris riding his mythical dead donkey? 😛

      • Dead donkey is … graphic? Also a new phrase to me.

        No need for the prodding, no need. I’ll be sure to let you guys know when I come up with the brilliant new silicon printer. In the mean time, you guys can go look up all the people that couldn’t possibly recognize the need for cellphones, PCs and lots of other inventions in the past. We’ll see!

  11. I agree with Dave. Fabricating your own chips using traditional silicon based processing is not going to be practical, even if it is based on process technology from the 70s. I’ve studied micro-fabrication in grad school and the facility we worked in was about as small as they come and I still couldn’t imagine it being crammed in your garage. You’re going to need equipment for deposition, etching, annealing, diffusion/oxidation, lithography (spinner, aligner, etc…). The list goes on and I haven’t even mentioned inspection/test or packaging equipment. None of this equipment is what I would consider cheap. It is also very difficult to keep it running and customer service is essentially non-existent. Just the lithography supplies alone are expensive. If I remember right, one of the bottles photoresist I used approached 1000USD. The lithography photomask sets, for 100mm wafers, that I designed cost about 700USD per layer to have made, which was from Bandwidth Foundry in Sydney. Then of course the chemicals you need, the consumables for the equipment, and the infrastructure needed for the equipment and to provide a safe, clean environment to work in. The point is that in grad school I had access to pretty much the bare minimum to produce a circa 1970s IC and it would still be extremely difficult to reliably and repeatedly produce. I just can’t imagine doing it at home.

    With all that said, I don’t believe it would be impossible to one day produce your own chip at home, it just won’t be on silicon. Printable circuit technology is being research which uses modified inkjet printers. If the technology could get scaled down to chip level, could you produce a chip at home? Ignoring the fact that it would still need to be packaged, sure. Could you afford to purchase the printer and the necessary cartridges? To be determined.

  12. No one has yet commented on what I said about what happens if (perhaps when) such a machine ever becomes possible.
    If the MakerChip machine was available for $500 today, would I buy one? Probably not. because I can’t see why I would want to produce my own chips (apart from the fun aspect). I’d much rather design my projects around fully designed, tested and characterised chips available off-the-shelf with production and sourcing options, just like I always have. I only see downsides in using my own chips (not the least of which is the extra design time and effort required). Especially in terms of other people being able to build the project as well (open hardware).

    Unless of course you start talking a utopian future where every chip design is open and can be downloaded and instantly printed on your universally available MakerChip machine.

    Unicorns anyone?

    • In all fairness, I think your approaching this from a professional PCB designer stance, which is great. Now assuming the MakerChip could exist, a professional chip designer whose dabbling with the hobby or getting the entrepreneurial itch could test and characterize their chip just as you would do the same with one of your boards. But they are going to be catering to a very small niche market.
      I think this brings up a couple of questions. Would anyone here use a homemade chip for there personal projects and would here anyone use a homemade chip for a project that they would end up selling?
      My answers would be yes and no respectively.
      As for the open IC design utopia full of reputable IC designs, Unicorns indeed.

      • Size is what would drive people to use a machine such as this, especially in industry. We’re already seeing vendors combine multiple pieces of silicon in the same bonded packaged in order to save space, I think the making of an SOC by printing might be the next step after that (or few steps, since Dave doesn’t think this will happen in his lifetime).

        • If you’re referring to die stacking then yes. But I don’t know about printed SoCs, to me that seems pretty far down the pipeline. A SoC needs many more component types built-in and right now printed circuits don’t seem to have that kind of flexibility yet. Although Kovio has claimed to produced RFID tags, so maybe it’s closer than I think. Oh well, time will make fools of us all.

    • Why is it unreasonable to expect free and open chip designs to use with our MakerChip? There is already similar things with FPGAs and ASICs – http://opencores.org/projects. Assuming the process itself is characterized, there would be no need to do anything other than software characterization on the final design. There could also be commerical IP, you could buy a ARM core, drop in an AD ADC and a TI opamp. With the future going to smaller and more efficient designs, it seems like it is just a matter of time before this happens. You could add just enough memory, just the needed peripherals, etc., to your design.

      If this was a printed silicon solution, I could easily see how a proto design could be moved to production runs. Once the design is completed and verified on your personal MakerChip, send the files off to a silicon printer and get your design mass produced, or put it on-line for the open community to download and use on their MakerChips.

    • Having worked on a few chip designs, and also been frustrated in my hobby projects when I can’t buy chips with the functions that I want from existing IC vendors (though they’d be simple to design), I very much look forward to chip manufacture being more available, like PCB manufacture is now. Pretty much any circuit that you could make with transistors on a breadboard will work about the same on a chip, but roughly 20x faster for say a 0.5um process, which is about the best a serious hobbyist could afford at present from somewhere like MOSIS. Also, imagine what you could design if adding transistors is basically free, and all your resistors and capacitors match each other within 0.1%, and the VBE of transistors matches so you can make good current mirrors etc. The downsides are that capacitors become very expensive, and the absolute tolerance of resistors and capacitors might be 20%. There are some things that you can do on a chip, that you personally might want to do, and that the existing semiconductor companies don’t want to do because their marketing departments will only offer what they think you want. One example: there is a very poor offering Operational Transconductance Amplifiers (OTAs) on the market, because marketing departments think you only want more and more variations of the op-amp, even though chip designers probably use as many OTAs as opamps as parts of their bigger chips, buried deep inside where marketing has never noticed. If you could design your own chips, you could bypass having the description of your requirements filtered by a few levels of distributors, sales reps and marketers, then averaged with a bunch of other random customers before it gets turned into a product specification. So – there is the “why”.

  13. Designs and “home versus production” use aside. To make chips that are anywhere near current production quality requires very, very thin coatings, several layers of masking, etching, and doping, and ideally exotic light sources to get good mask resolution.

    Big stuff, done at home, makes some sense. Like the FET’s that Jeri made. Now think of how many of those switches are in a 555 timer. Using Jeri’s method, a 555 would be the size of a business card.

    Lets say you even refine it, so you can make transistors 1/10 the size jeri can. You’re still looking at a massive chip for anyhing but some simple integrated logic.

    A “microprocessor” would be the size of a dinner plate.

    I think the discussion is at far to high of a level right now. Like talking about the ramifications of a flux capacitor. We need to talk about practical sizes and production methods, and who here knows enough to get into the nitty gritty of it?

    And did anyone note how lossy her FET was?

    And why would this be better than a FPGA or Microcontroller? Or if you need a device not served by current parts, why would an antiquated production method be better?

  14. Regarding working for military companies.

    I work for the biggest weapon manufacturer here in Norway, producing high-end missiles and weapon control systems (among other things). The way I “live with myself” (as Dave would put it) is due to the fact that we only deliver our products to nations within NATO. All the terroristinfected countries can bugger off.

    I actually got a question of what I think of working with designing weapons that can potentionally kill people when I was at my job interview. Don’t have a problem with it really, as long as they fall in the right hands.

  15. I would have to agree 100% with Dave here on his point on making IC’s @ home. Completely not practical.
    Nevermind the countless $$’s you’ll have to invest in chemicals, machinery, equipment, etc. The yield would be terrible… ANd there’s no way you could get a density good enough to make it worthwhile.

    It would be far easier to simply re-engineer the device in question and use an FPGA or some similar to get the same result. That is, if someone hasn’t done all that work for you already (www.opencores.org)

  16. I have to agree with Dave about homebrew chip fabrication. I think it’s probably possible, and would no doubt be a fun thing to do and experiment with, so there’s mileage in it from that point of view, but as for using it in serious applications… nah, not feasible. You’re never going to be able to compete with the big players who have carefully refined processes and have already invested in expensive equipment that allows them to manufacture chips quickly and efficiently.

    It’s the same as homebrew PCBs, but on a much larger scale. If you’ve ever done your own PCBs at home, then you’ll know that it is a major pain in the arse. It takes bloody AGES to manufacture your own PCB, there’s nasty chemicals, there’s a high failure rate involved, it’s just awful. For that reason, homebrew PCBs are limited only to one off prototypes and stuff like that. Nobody ever does small production runs (even say, 20 boards) using homebrew PCBs because it’s just more hassle than it’s worth. You’re far better off paying a professional company to fab your boards, and the results you get are infinitely better than you could ever have accomplished at home anyway.
    Homebrew chip fabrication would be a little bit like that, except much much more intricate, it’d have a much higher failure rate involved, and it’d be a much bigger pain in the arse.

    So in summary, it’d be a cool project just for fun and to experiment with, but has no real practical use.

  17. Hey Dave and Chris!

    Freagle really should have two e’s. Freeagle. No?

    Also…you know if you guys want to bump this podcast up to twice a week, I have plenty of time to listen. Just sayin’!

    Love the show!

  18. Finally getting caught up with the show. As for military. This is always an interesting discussion. The portion of money dumped into military per death is so much more/better than money put into alcohol production and deaths caused by alcohol. So it cost less to kill someone with alcohol than a military tool and military apps have, in some instances reduced casualties. BTW, the average deaths per year around the world from alcohol is on the same order of magnitude as war (at least in the stats I was able to come up from war). Of course there is always the discussion on self inflicted death v. mass destruction etc. I am not really an advocate for either but I am not totally against them either. Just some ponderings as I listened.

    • Interesting point, I’ve never heard the cost of alcohol compared to the military. If you ever find numbers, please let us know, that’d be an interesting followup.

      • Hi Chris. Unfortunately that was a study I did a while ago b/c a large portion of our local population is anti military but pro “party” so I do not have the references on hand but a quick search shows the Canadians claim 1 in 25 deaths are alcohol related (best summary on this: http://www.medicalnewstoday.com/articles/155447.php) and several studies say between 75 and 100 thousand deaths in the US directly caused by alcohol (on OK summary here though they are very bias in reporting http://www.msnbc.msn.com/id/6089353/ns/health-addictions/). WOD Facts (http://drugwarfacts.org/cms/?q=node/30) seem to indicate 85,000 from consumption (though they too are probably bias and do not have good references on their chart) and not secondary death.

        40-54% (depending on site) of the US tax money goes to the military (please note that the US Consitution mandates a lot of resources to military related activity – but that is another “soap box” subject). The average US citizen pays just under 27% of their income in tax. So at least almost 10% of US income goes to military.

        Death by war is a hard one and I do not have time to research that well but wikipedia has a good list (http://en.wikipedia.org/wiki/List_of_wars_and_anthropogenic_disasters_by_death_toll). However, since I only have US stats, US related war casualties are well under 10,000/yr for the past 100 years or so (http://www.historyguy.com/american_war_casualties.html). “Ethnic Cleansing”, if included in war numbers will drastically change the stat.

        In summary, a US household spends about 1% (very liberal. See this great chart: http://www.visualeconomics.com/how-the-average-us-consumer-spends-their-paycheck/) of their income to “kill” 75,000 US people per year by alcohol and 10% of their income (or 10x what is spent on alcohol) to kill 10,000 (it is actually probably less) US people by war.

        Disclaimer: Third world countries are much more effective in their war spending than the US and casualties caused by the US “military machine” are higher than the US casualties. I am not against a good military or a good wine but I am not condoning many of the problems of both. Sorry there are only US stats since gathering world stats are a lot harder and I would probably loose a day of work and/or have a not-to-happy wife for taking her time to do this.

        Conclusion: The US would probably be more effective by giving the enemy alcohol (it has been a known strategy in several conflicts). Alcohol is a cheap killer when used in excess. Too much money may be spent in military, but is does not kill near as many people and most likely protecting/saving (tons of innovation is used in military for these apps.).

        I could go on but I hope that gives you some fodder.