An LM386 Oscillator Thanks To Tungsten Under Glass

Once ubiquitous, the incandescent light bulb has become something of a lucerna non grata lately. Banned from home lighting, long gone from flashlights, and laughed out of existence by automotive engineers, you have to go a long way these days to find something that still uses a tungsten filament.

Strangely enough, this lamp-stabilized LM386 Wien bridge oscillator is one place where an incandescent bulb makes an appearance. The Wien bridge itself goes back to the 1890s when it was developed for impedance measurements, and its use in the feedback circuits of vacuum tube oscillators dates back to the 1930s. The incandescent bulb is used in the negative feedback path as an automatic gain control; the tungsten filament’s initial low resistance makes for high gain to kick off oscillation, after which it heats up and lowers the resistance to stabilize the oscillation.

For [Grug Huler], this was one of those “just for funsies” projects stemming from a data sheet example circuit showing a bulb-stabilized LM386 audio oscillator. He actually found it difficult to source the specified lamp — there’s that anti-tungsten bias again — but still managed to cobble together a working audio oscillator. The first pass actually came in pretty close to spec — 1.18 kHz compared to the predicted 1.07 kHz — and the scope showed a very nice-looking sine wave. We were honestly a bit surprised that the FFT analysis showed as many harmonics as it did, but all things considered, the oscillator performed pretty well, especially after a little more tweaking. And no, the light bulb never actually lights up.

Thanks to [Grug] for going down this particular rabbit hole and sharing what he learned. We love builds like this that unearth seemingly obsolete circuits and bring them back to life with modern components. OK, calling the LM386 a modern component might be stretching things a bit, but it is [Elliot]’s favorite chip for a reason.

41 thoughts on “An LM386 Oscillator Thanks To Tungsten Under Glass

  1. “We were honestly a bit surprised that the FFT analysis showed as many harmonics as it did” well, they are at around -60dB compared to F0… that’s definitely nice for a simple negative feedback oscillator…. and yes, we have better energy efficient techniques rather than heating metal now :)

  2. Obvious politics aside, very interesting. An illustration of how some elementary things might still come back from the grave because of the basic interaction of the elements, combined with their ease of production and sourcing—for now.

  3. Incandescent lamps make wonderful charger circuits for Lead-Gel accus, too.
    A 12v halogen lamp is all it needs to safely charge such an rechargeable battery. It acts as a fuse, too. And a visual indicator. It’s such a wonderful piece of technology that shines through sheer simplicity. Too bad it’s too bright for the few dim users who favor LEDs over everything (little pun intended). Sigh. 😔

  4. I actually think using bulbs for their PTC attributes is charming. But if we’ve got to bring politics into it… Technically when you use them that way, are they slightly less efficient than an a parts-list PTC resistor, since they radiate heat faster than is actually required? :P :P

    Anyway, the last times they “took away incandescents” I was still able to get filament bulbs of one kind or another, just not the regular sizes for home lighting. If you’re using 60 watt bulbs for lighting when you can get 4 watt replacements for a few dollars that are the same color and high CRI, you must have very strange requirements.

    1. They’re actually more efficient than the PTC resistor, since at those low temperatures where the filament doesn’t yet glow, the gas filled bulb is an insulator.

      >4 watt replacements for a few dollars that are the same color and high CRI

      The trouble is, they aren’t. First of all, the cooling of the bulbs is pretty terrible so the lm/Watt brightness rating doesn’t hold for long, the “60 Watt equivalent” is calculated using some dubious assumptions in the first place giving you about a hundred lumens less than it should, and the color temperature/CRI is usually low and off by a lot, and keeps changing as the bulb heats up and ages. Then there’s the strobe effect problem with some cheap bulbs, and the fact that they don’t last nearly as long as advertised. There are very few LED bulbs out there that don’t have at least one or two problems.

      1. I guess I can buy that about the PTC for some situations. Was mostly cheeky about that.

        I wrote some more detail for you, but the comment glitched. ^F for 65465213 if you like. Summary, I don’t think that’s the case anymore mainly because I found some new ones a month or two ago that are impressive.

    2. It’s a bit harder here in EU, though.
      Light bulbs are almost being treated as if they were the tools used by criminals. Some cartoonist even made fun about that, I remember. In simple words, you can keep using up your stock, but dare you for trying to get hold of new ones (esp. importing).

      Which is sad, I think, because an incandescent lamp can be used to replace much more complex and error prone circuits. It’s not just about the warmer, more natural light spectrum that it has.

      For example, you can put it in series with another, power hungry device. If something is broken inside there, the light bulb will prevent further damage. It’s like a smart fuse, and so forgiving in its application.

      And last, but not least, It’s also a big part of human history (invention of electric light) and culture. It’s not just “a thing”. It’s an achievement. Something humanity should be proud of, not ashamed.

      But all the lawyers and lobbyists merely see it a hot-heated lamp that wastes energy.
      They don’t understand that electricians and hobbyists may have other, special uses for incandescent lamps. Or they just don’t care, which might be even worse.

      In the way EU develops at moment, it looks as if it is taking away autonomy of its countries and their citizen’s rights. Everything is so top-down, so strict. And this angers and annoys people.

      Even if there’s just the slightest sign that something could be bad or used for a bad application, it will be quickly restricted (except taking actions against light pollution, eye damaging LED light (city lights) and toxic/dangerous Chinese products etc).

      All in all it’s the reverse what we young Europeans tried to accomplish in the past years, I think. That’s not what we hoped for in year 2000, when we saw an united Europe of the future. We were looking for a democratic, free community with commonsense. Not prohibitions at every corner and even more bureaucracy.

      Hm. Maybe it’s also a bit because the EU (European Union) has taken over the duties of the EC (European Community), which ceased to exist in 2009? Hm. I don’t know for sure. I wished politics were made more transparent to everyone, being better communicated. It’s not as if we weren’t interested in it.

      But it feels as if the bad guys get the bad stuff more easily than ever, since CE certificate and other lame inventions replaced national, more detailed/strict-if-really-needed regulations of the member countries. Back in the 90s, cheap Chinese products as we have now were being unheard of.

      Anyway, these are just my two cents. Maybe I’m wrong, also, not sure. I just try to put the pieces together in a way it makes sense. So please don’t take everything at face value, please double check rather what I said. Thanks.

      1. Sure! I like them, truthfully, for the neat things you can do. Can’t say I know what’s banned in EU, but I agree that filament bulbs in a lot of formats should still be around for things. I just think the short-life high-brightness 60 and 40 watt regular bulbs really shouldn’t still be a very common option for normal everyday lighting. The heat lamps, automotive stuff, lots of appliance bulbs, etc seem like they have more reasons to stick around and are still filament bulbs. And I do think as a major historic object, they had better not be so hard to find that no-one knows what they’re like anymore, at least for a few generations.

        1. The first round of efficiency regulations pushed the manufacturers to use halogen bulbs inside the regular ones, which improved the efficiency about 30% and made them last twice as long – and three times as expensive.

          1. And the halogen cycle that preserves the filament only works above a certain temperature, which isn’t the best for dimming. But I always liked the light from the high-power halogen light I used to have pointed at the ceiling, back before LEDs were an option. Wasn’t really worried about the higher cost to buy, because the running costs made up for it easily. It was even dimmable to an extent, come to think of it.

      2. >Everything is so top-down, so strict. And this angers and annoys people.

        That’s the fault of local governments. The regulations are “suggestions” that need to be implemented into local law, and the nanny-statists use the EU as an excuse to implement very strict versions, while the more “relaxed” countries simply ignore them.

        For example, Sweden should be using the Euro already – they’ve just kept using the Krona because they want to, and nobody can do anything about it.

      3. I dont know what part of EU you recide in, but over here in the north eastern part i can find 60W incandescent lamps for sale at a price of 2€ a pop online. If you dont need massive wattage, electric kitcen oven lamps are abundant with wattages between 7-20 watts.

  5. Really? Well, I am happy to see it come around again, as it is a _wonderful_ sine wave oscillator in the audible range, and is was many makers’ first bespoke circuit as recently as the aughts. Much better performing incandescents are not hard to find in personal quantities, even now (for this use case).

    I’ll spare everyone explanations and measurements (for < 10^3Hz, you need a good 12-bit digital synthesizer to match it), just search for Jeri Ellis or look at TAOEE (we still remember this, right?).

      1. Sorry, lol. But the point I was making was that it is a bit of a trope, the “HP 200 like” generator, as mentioned by someone else in this thread. And there are many comments on this story that are better than mine! (Btw, iirc, the original used big variable capacitors instead, which is a very uneconomical choice these days.)

        It is a great circuit for a jumping-off point; from The Art of Electronics to the beginnings of the “maker movement”, to the early days of HP… I’ve even seen Jim Williams mentioned, which is a sign of a virtuous conversation.

        Tldr there’s a lot of informative comments here, now.

      1. You’re right. And the properties of the lamps vary widely, and are not specified. Still, with a lucky or informed choice, it has a fundamental advantage of being continuous, as opposed to discrete with enough points to be good enough for purpose.

        And I was in error, sorry. The usable frequency range is about 10Hz to somewhere under 100kHz; ie great for audio. When I made these as semi-disposable gens for audio work, years ago, I included a range switch section with different R. It seemed to me, at least, that this was required for it to be useful.

  6. The brightness dropping to 80% is what defines the rating for the bulb lifetime. If they do this, it is a violation of their rating.

    800 or 810 lumens is reasonable incandescent bulbs have always had a tradeoff between life and brightness. The typical bulbs started around 840, I believe, for non-frosted bulbs which haven’t yet gotten a layer of dust on one side or tungsten on the other. The long life ones were 500 by comparison I believe. Halogens were better.

    The CRI is testable. So are the remaining color differences not covered by CRI and color temperature alone. There are good and bad options on the market. Photographers use the best LEDs for light panels, and it’s been years since they all put a high temperature stable color filter in front of a 500W halogen trying to get the color right for a scene instead of bi-color LEDs.

    The color should not change enough to notice when operated at a steady brightness for a period of time. If it heats up and turns blue, the bulb is overheating due to being overdriven, is a piece of trash hazard, and shouldn’t have been bought. It’s also not efficient, which means it’s not one of the ones I was referring to or is defective.

    Vimes “boots” theory in action right there. The worst bulbs cost more to own than the decent ones while doing a worse job.

    Sure, there’s a lot of bad ones, but there’s also a lot more variety to choose from. With incandescent the top two for efficiency are regular short-life bulbs (if you need continuous dimming) and halogen (if you don’t mind that dimming too low makes them burn out faster due to not having enough temperature to cycle tungsten back to the filament).

    I don’t want to sound like an advert, but I mean… The bulbs I’m thinking of are from Philips. Their ‘ultra-efficient’ family of bulbs has a few that claim 50,000 hours with 90cri, 3000k like halogen, and yet remarkable efficiency. I consider them to be widely available since it’s at Walmart and on Amazon, even though there’s not a bunch of different brands with equivalent options. They stock it in a couple of versions – 4.5 watts and 810 lumens for $7.50 per bulb (2-pack) or 8.5w and 1600lm for $15 in whichever color you prefer. Those don’t have dimming and are not frosted, but that’s impressive compared to what we bought the first time they wanted people to move away from incandescent. I didn’t notice any flicker or color issues with the one I got, although I haven’t used it much because I prefer the other color typically and I didn’t feel like replacing the bulbs too early. If you want heat rather than just light, heat lamps last longer. If you just want a light bulb, we’ve got it easy nowadays, especially since we’ve had a lot of time to improve the alternatives to incandescent since the first time they “took them away”. If you want another format, they’re going to make them for automotive and appliances still. Hopefully for a long time.

    1. 80%, 70%, even 50% – bulb lifetime is not standardized in the EU. The manufacturer can choose which rating system they use. The other problem is that the brightness is measured immediately at startup, but LEDs fade as they heat up. The lm/W rating of a diode is measured at a junction temperature of 25C by convention, and they can sell bulbs by that rating directly as it’s what the manufacturer of the diode claims. Same goes for the CRI etc.

      The third problem is external factors: power quality issues kill LEDs quickly, and these faults are not covered by any consumer warranties or regulations. They don’t need to make them robust against voltage transients or brownouts etc. and this is the major killer.

      And consumer grade bulbs available at the regular stores have poor CRI as per course. 70-80 is normal because people don’t know to demand any better and the stores won’t stock bulbs that cost $25+ a piece for 90+ CRI. Nobody would buy them, which means you have to go out of your way to find any. What you find instead is these stupid IoT bulbs that change color by bluetooth.

      Halogens were considerably better. At the time when the first efficiency regulations came around, a 43 Watt halogen bulb that replaced the normal type was equivalent to a 60 Watt regular tungsten, and a 73 Watt bulb was equivalent to a 100 Watt regular. The LED replacements kept referencing to the plain tungsten filament bulbs even though they went off the market, so the comparison was invalid.

      1. Most consumer led incandescent replacements sold the past 20 years are built as cheaply as possible. None I have used has surpassed the lifetime of an incandescent lamp. Lately the gap has decreased but the consumer stuff still isnt as reliable as the old incandescents were.

        1. Every Compact Florescent or LED I have purchased in the last twenty years have far outlived any incandescent bulb I have ever owned. The set of six BR30 LED bulbs I replaced this past winter were over ten years old. I have six CFL in my basement that are over fifteen years old. The five outdoor rated floods in my back yard and driveway are six or eight years old. Those run all night, and I got sick of climbing a ladder once a year to replace the halogen bulbs they replaced.

      2. I don’t know about the EU so I’ll assume you’re right. I do know that I’ve not observed massive differences in cold vs heat-soaked LED bulbs, and that the high-efficiency ones also typically don’t get very hot because there’s just not enough wasted heat to do it.
        Power issues are what normally kill the bulbs, yeah. The decent ones do seem to last a LONG time though. I’ve got some of the first cree a19’s still working, although not in my most-used fixtures.

        And I don’t really think it matters what the majority of options do, just what you can get if you read the labels and compare options. And if they meet their label, then the ones I’m thinking of aren’t $25 and they are in the same aisle as the crappy stuff. And some of the less-efficient high CRI options aren’t $25 either, for what it’s worth. Maybe $10-15 now.

    2. Oh, and color uniformity is yet another thing for cheap bulbs, especially those with the filament sticks.

      The “white” LED is a blue diode surrounded by a green-yellow phosphor. The diode casts light in one direction where the diode is pointed, while the phosphor shines all around, so different parts of your room get different color temperature.

      1. And the related problem is that the color temperature actually “pumps”. Using slower phosphors gets rid of the noticeable flickering issue – but not quite. The blue diode still blinks on and off, causing a ripple to the intensity and color of the light.

        1. This is hilariously evident when buying a led light imitating an old incandescent light. Those led strings, impressive as they are can flicker like crazy, depending on the manufacturers short cuts.

          1. Cheap ones are half wave, better ones are full wave. The difference is obvious.

            Non-string light engines make better light, but they don’t work as bare bulbs.

        2. The bulbs that don’t direct drive off of an AC wave are plenty smooth, of course. I’ve got an decoration that doesn’t, and it’s annoying even just how it flickers regardless of the slight variation from drive current vs output CCT or any possible lagging from phosphor choice.

      2. In the realm of flashlights and focused light sources, this unevenness shows up as a rainbow colored beam, and so they have enough buyer pressure to make leds that don’t have the issue. So all the manufacturers know about it and know how to avoid it. It’s a shame they like to skip that to squeeze out better numbers when they can get away with it.

  7. Back in the mid eighties I worked at Bell Labs, and I had a drawer full of compact incandescent bulbs purpose built for this. They were the diameter of a nickel, and about 3/4 of an inch tall, with a bi-pin base. I think they were 12v, but they could have been 48v (telecom gear). They were replacement bulbs for the oscillator in a portable test set. The filaments in them were optimized for oscillator feedback.

    In light of this article, I wish I had saved that box of bulbs when we cleared that lab. I had about twenty or thirty spare left, and about three same number of Nixie tubes that all got scrapped.

    Hind sight and all that.

  8. One trick with the 386. They’re prone to crossover distortion which might be a source of the distortion. Try putting about a 470 ohm resistor to ground from the output of the amplifier , directly, before the output coupling capacitor. This forces quiescent DC current flow through the output transistors and shifts both output transistors a little closer to class A than a really lightly biased AB1.
    Just guessing but I’m thinking you should be able to get really low distortion maybe in the range of .01 or less out of this with proper power filtering and circuit layout.
    Make sure you’re not operating anywhere close to the rails. Turn it all down (with feedback) maybe 3-6 dB below maximum output.

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.