Retrotechtacular: How Not To Design With Transistors

Consider the plight of a mid-career or even freshly minted electrical engineer in 1960. He or she was perched precariously between two worlds – the proven, practical, and well-supported world of vacuum tube electronics, and the exciting, new but as yet unproven world of the transistor. The solid-state devices had only started making inroads into electronic products relatively recently, and mass production techniques were starting to drive the cost per unit down enough to start including them in your designs. But, your company has a long history with hot glass and no experience with flecks of silicon. What to do?

To answer that question, you might have turned to this helpful guide, “Tubes and Transistors: A Comparative Guide” (PDF link). The fancy booklet, with a great graphic design that our own [Joe Kim] would absolutely love, was the product of the Electron Tube Information Council, an apparently defunct group representing the interests of the vacuum tube manufacturers. Just reading the introduction of this propaganda piece reveals just how worried companies like RCA, General Electric, and Westinghouse must have been as the 1950s turned into the 1960s. The booklet was clearly aimed directly at engineers and sought to persuade them of the vacuum tube’s continued relevance and long-term viability. They helpfully explain that tubes are a reliable, proven technology that had powered decades of designs, and that innovations such as heaterless cathodes and miniaturization were just around the corner. Transistors, we’re told, suffer from “spread of characteristics” that correctly describes the state of materials engineering of silicon and germanium at the time, a thornier problem than dealing with glass and wires but that they had to know would be solved within a few years.

With cherry-picked facts and figures, the booklet makes what was probably in 1960 a persuasive case for sticking with tubes. But the Electron Tube Information Council was fighting a losing battle, and within a decade of swamping engineers with this book, the industry had largely shifted to the transistor. Careers were disrupted, jobs disappeared, and fortunes were lost, but the industry pressed forward as it always does. Still, it’s understandable why they tried so hard to stem the tide with a book like this. The whole PDF is worth a look, and we’d love to have a hard copy just for nostalgia’s sake.

Thanks to [David Gustafik] for the tip.

76 thoughts on “Retrotechtacular: How Not To Design With Transistors

    1. The whole tube vs. transistor war of audiophiles boils down to two things:
      1. Audiophiles actually don’t listen to music with their ears. They use wallets for that.
      2. Audiophiles (and not only them) like the way tubes distort the sound.

      Ad. 1.
      Typical audiophile expects that spending X amount of money will improve the quality of sound by factor of Y, which is calculated as the difference in cost from his previous setup. So equipment that costs 20kUSD should be twice as good as equipment that costs only 10kUSD. Audiophiles for this very reason hate double blind tests and objective measurements.

      Ad. 2.
      Typical vacuum tube amps introduce lots of distortions, in terms of THD+N it could be more than 10%. But they do it in softer, more pleasant manner than transistors or op-amps. Engineers assume that amplifier is a wire with a gain – something that will not distort the sound. Audiophiles ignore that preferring distorted, but pleasant (and expensive) lie over undistorted truth. Again, in double blind test they wouldn’t be able to distinguish between true tube amp and one faked with transistors, op-amps or just DSP.

      1. I recently saw an audio recorder that records 24 bit samples at 4 MSPS. The device is used to make audio CDs. They even got a prize for that innovative idea. When I asked, they admitted that they had never done a double blind test to compare it to common recording equipment.

      2. I’ve done live sound with class D and thrift store speakers, and I don’t even own any analog processing gear except my mixer. Nobody can hear the difference in a boomy room with unpleasant echoes, and a bunch of conversations going. My dynamic EQ and digital feedback elimination, and my off brand but all digital wireless mic make a difference though.

        I also take pictures with an eBay CCTV lens on a mirrorless camera. I’m not even an experienced photographer or anything. All my power tools are Ryobi and my phone cost $200.

        Digital stuff just isn’t all that expensive to make, and it’s pretty hard to get a lot of it wrong. Code generally does what you code it to.

        It’s the mechanical bits, the stuff that has high current going through it, the time saving automation features, etc, that I actually bother going name brand on.

        And the best known version of something is IMO almost never actually the best, aside from simple mechanical stuff that you really want to be tried and true.

          1. Yes, they can. Some years ago, radar transmitter engineers at Westinghouse tried to get the pulse to pulse cancellation ratio of the AN/TPS-75 radar down to the 70 dB range on a hot-rodded test radar. It was damping the vibration in an RF drive cable that finally got there. It put a slight phase modulation in the signal due to the coax center wire moving in the cable. Of course this was a machine with 400 Hz motors, pumps, and transformers buzzing away and the whole S-280 shelter rattles. And -70 dB is 100 parts per billion. Phase modulation with flexure is a problem in cable, noticeable and testable.

            As for the effects of floor vibration on your audio cables, if you think it “removes haze” and “makes music sound more like music”, then who am I to judge, but at those frequencies and vibration levels, there are not going to be any measurable effects. It’s complete bullshit and snake oil.

          2. Well, I hate to validate crap like this, but if you move a conductor through a magnetic field gradient, it induces a current — that’s how generators work. So yes, vibration CAN affect an electrical signal.

          3. We have routinely seen vibration (multiple G’s) feeding back into our RFID system antennas for rail applications along the cables – all minimized with more effective strain relief. So yes. Tie those RF cables down.

      3. “Typical audiophile expects that spending X amount of money will improve the quality of sound by factor of Y, which is calculated as the difference in cost from his previous setup. So equipment that costs 20kUSD should be twice as good as equipment that costs only 10kUSD. Audiophiles for this very reason hate double blind tests and objective measurements.”

        And so it started….
        But to be honest, I doubt that the relation is a linear one.

      4. What’s great, is the fact that by the time most men have enough free money to afford all this nonsense, their own hearing has deteriorated to mud anyway. So not only is the difference between X and Y imaginary, they couldn’t hear it even if it wasn’t.

        The whole thing mystifies me, it’s like there’s a shortage of things in the world to spend money on. They usually have atrocious taste in music anyway.

  1. There’s a famous QST article about transistors from about 1955, and they sum it up “transistors will never amount to much”. I’ve seen the article.

    The problem was, that was early. Power and frequency were needed, and that was in the future. People did take literally from tube design, class A audio output stages with a transformer. It was a big shift from high impedance tubes to low impedance transistors. Power supply articles in the sixties just kept paralleling large value capacitors as filters, and still hum, the shift coming with active voltage regulators. Suddenly electrolytics were much more common, there’d be a few in a tube receiver, but with transistors they were endless a lot of early transistor receivers were awful, I’m thinking they focused on the low power aspect rather than good signal handling.

    When people started doing things differently, that’s when things got better. So audio stagescwith big coupling capacitors disappeared, DC coupled amplifiers took over. Using extra transistors made design easier, while with tubes they added cost and space.

    The transistors had to get better, but so did design, and that never comes immediately.


    1. Reminds that of an article I read in one of the major computer magazines of the mid-80’s which reviewed the latest cutting-edge input technologies of the day (i.e. mouse, touch-screen and voice recognition) and concluded they were all fatally flawed and none would succeed.

      1. Kinda also reminds me of the early 1990s, when so many experts were predicting silicon’s days were numbered, soon to become as obsolete as germanium transistors. Much faster gallium arsenide chips were widely thought to be the future of semiconductors, soon to be powering all computers.

        Today gallium arsenide transistors are used for RF, but the GaAs future so many anticipated never quite materialized.

      2. I read an article in a computer magazine, in the ’80s, that claimed that we would never really faster computers since computers were I/O bound. They discussed the variety of I/O devices that were slower than the processor. They claimed that mice, NTSC video, printers, floppy drives, and keyboards would always hold more powerful computers back, to the point that they would be useless.

  2. I work at a semi conductor fab (flash memory) and frequently make comments along the lines of “Who will ever need THAT many kilo-bits?!?”. I think the graphic on page 10 of the .pdf will make a great talking piece for my cube wall

    1. Yeah, it’s like the “carrots are good for your eyesight” misinformation that the Brits invented during WW2. In that case, the objective was to cover up the fact that their advanced cryptanalysis was providing them with too much information about the times and locations of attacks, enough to be suspicious. In order to prevent the enemy from doubting the security of their communications, they came up with the alternative explanation that their observer corps just had fantastic vision due to the carrots in their diet. People STILL believe that carrots are good for your eyesight, and people STILL believe that vacuum tubes are better at some things.

  3. It may be industry propaganda, but man is that a nice looking book! It just wreaks of the 60’s in the best way – beautifully typeset, crazy pop-art colorized photos, spiral-bound. I think Don Draper must have had his creative team put this together.

    1. Exactly! I forwarded the pdf to our art director Joe Kim, and he was in love with the style. He said he was going to show it around the art department to give people design ideas.

      What I wouldn’t give for a hard copy of this book…

  4. “But the Electron Tube Information Council was fighting a losing battle”

    I don’t know. Maybe you are taking too much at face value. I doubt they expected or intended to keep the world on vacuum tubes. More likely they were only trying to buy a little time, bad talking the transistor with one hand while the other was furiously building up their own semiconductor fabrication capabilities.

    I doubt they were trying to save the vacuum tube from the transistor. Instead they were just saving their own empires from the new upstart semiconductor companies such as Fairchild that weren’t tied to large vacuum tube investments.

    1. Not weird, pretty common in fact, all people love to forget something, tube allows you to work with high voltage where transistor allows you to work with high current…. See the point, it’s often easier to work with high voltage than high current.

    2. I graduated EE in 1970, and of course never learned a thing about tubes. I spent the next 40 years working with them. TWTs, thyratrons, klystrons, ignitrons, at very high power levels. Most I needed a crane to lift.

      The Air Force had a program to train Ph.D.s in tube physics, alarmed at the diminishing pool of experts. When you need megawatts out of one device, there’s one cheapest way to get it.

  5. I used to have a book, published in the mid ’50s.It was a textbook for consumer electronic technicians. It had a section called “The Crystal Triode.” It pretty much trash-talked transistors (never referred to as such, they stuck with Crystal Triode.) They covered, low gain, fragility, that transistors produced on the same line could have radically different gains, they couldn’t handle practical voltages (from the view of vacuum tubes could have hundreds of volts, and transistor couldn’t handle that.) they also stressed that transistors would never handle enough power to say, drive a speaker. It did it’s best to teach technicians that the future was nothing to worry about. Surprisingly, it was a decent book except for that section. I got a job with a TV shop when I was a teenager. (1977) There were guys there that had no knowledge of transistors and wouldn’t work on anything using them. All the solid-state devices were repaired by me and a friend who worked there.

  6. It wasn’t really such a strong choice between the two.

    I worked in communications. Valves had variations to. We had to test the characteristics of valves and design circuits specifically for those valves for noise and distortion meters.

    Valves were still the better choice for this for a long time as bipolar transistors are noisy.

    Valves until very recently were still used for the RF power output for low frequency broadcast transmitters AM FM etc.

    Until we had better quality FETs, valves were still better for the input stage of sensitive test equipment.
    Early transistors were noisy had a low HFE that varied from one to the next. If you had enough of them in a box then you could find one with the exact HFE you wanted.

    They really were crap but they were small and quickly became cheaper and hence the first products like the 3 or 5 transistor pocket radio receiver.

    Now I see that bipolar transistors have fallen out of favour with younger engineers. FETs are so much cheaper now and the quirks of bipolars are less understood.

    You can tell the age of a designer by the transistors they choose.

    Consider these choices: –

    OC 71

        1. My age too and I have used BC108 (BC107) but they were already too expensive (metal case) when I was a child and learned, that for most projects a BC238 (TO92) instead of a BC108 was a better choice. Today it is a BC847.

        2. Huh. I assume this list is going from oldest to youngest? That’s interesting because I am 8 years younger than Miroslav but the 2N2222 is the one I am most familiar with from that list.

          Maybe that’s because I started unusually young, pouring over magazines in the 90s like Popular Electronics, Radio Electronics, etc… and even reading old ones that I found in the cut-up corner of the school library from before I was even born.

          Today the internet has replaced those magazines but there isn’t much talk about individual transistors. It’s all Arduinos and Raspberry Pis so where would a younger person get their favorite go-to transistor from or where would an older person get their’s updated?

  7. I have to say that the spread of characteristics continues to be a challenge. I used to be a test engineer for a semiconductor manufacturer (about 15 years ago now), and I was somewhat alarmed by the range of characteristics I got from dice on the same wafer. In fact, curious about this, I converted certain characteristics to colors, and plotted these on a graphic that showed each die on the wafer in its actual position, in the color representing the value of a particular parameter. What I got was smears of color, which were different in shape for each parameter I measured. That is, there might be a gradient in leakage current that had a “hot spot” in the upper right area of the wafer, and a separate “hot spot” for gain toward the lower-right. There was literally NO part of the wafer that was particularly good for all parameters. This was on a cutting edge high-speed Si-Ge process, so that may have had something to do with it.

    1. Parameters can be complementary from each other so that it’s not possible to produce a perfect device. From what I remember of transistors, leakage and some other parameter, possibly gain, were the result of a tradeoff in manufacturing so that there was no such thing as a device with both low leakage and high gain. That could be what you saw.

      1. No. I’m specifically talking about variations across a wafer, which were clearly unintended. I mean, the specs for each parameter were well stated, but in reality were all over the map, so to speak.

  8. Just had to provide some feedback about the previous feedback about feedback (see the previous feedback).
    On design techniques, some audio amps used “feedforward” to minimize the limitations of feedback with early devices. As the previous comments point out, new design techniques allowed quality issues to be addressed by quantity when individual devices became cheap.

  9. Tubes are very linear devices; sadly many people seem to think the opposite. Our lead RF designer, recently retired, at the RFID company I work for had very positive things to say about their characteristics, when compared to transistors. The tone of this article is pretty negative for no real reason. Who knew technology progressed? Some of us grew up around CRTs and might dispute when tubes ‘died’ off. In fact many satellites still have ‘obsolete’ tubes. Even dying industries fight to stay alive. Shocking.

  10. I think the comments about the booklet are far harsher then deserved.
    The booklet claims that tubes would still be a good option for “several years” to come and they were right about that.
    In that day and age transistors were expensive and indeed unreliable.

    Also, engineers are not as easily mislead as “general public”, and the lobbies for tabacco and asbestos.
    Unhealthy effects of tabacco were well known before the 1880’s
    RIP Nellie

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.