Build the Simplest Bipolar Power Supply

How many integrated circuits do you need to build up a power supply that’ll convert mains AC into a stable DC voltage? Would you believe, none? We just watched this video by [The Current Source] (embedded below), where he builds exactly that. If you’re in the mood for a very well done review of diode bridges as well as half- and full-wave rectifiers, you should check it out.

First off, [TCS] goes through the basics of rectification, and demonstrates very nicely on the oscilloscope how increasing capacitance on the output smooths out the ripple. (Hint: more is better.) And then it’s off to build. The end result is a very simple unregulated power supply — just a diode bridge with some capacitors on the output — but by using really big capacitors he gets down into the few-millivolt range for ripple into a constant load.

The output voltage of this circuit will depend on the average current drawn, but for basically static loads this circuit should work well enough, and the simplicity of just tossing gigantic capacitors at the problem is alluring. (We would toss in a linear regulator somewhere.)

Quibbling over circuit designs isn’t why you’re watching this video, though. It’s because you want to learn something. Check out the rest of his videos as well. [TCS] has only been at it a little while, but it looks like this is going to be a channel to watch.

45 thoughts on “Build the Simplest Bipolar Power Supply

    1. I’m sorry that you can’t believe there aren’t hate comments yet. The video is considering the basics of unregulated linear power supplies. The main power supply build is fairly common for HiFi applications and there are good reasons for doing it this way; High power transients require massive surge current, which large capacitance can provide. Good luck with a linear regulator. This is not intended as a power supply for digital applications. This is a power supply for a power hungry Class A amplifier as designed by Nelson Pass. I believe this is all stated in the video.
      To prevent any ambiguity, I have changed the title of my video to “Unregulated Power Supply Design” in your honor.
      Thank you for watching.

      1. Yeah, that was a comment on Hackaday growing up, not the quality of your video. Your video is fine, as was every other video or blog that Hackaday links to.

        I was just surprised that no one had made up a bunch of things to be angry about is all. Couldn’t believe my eyes.

    2. There’s nothing to hate on. It’s about as simple as it gets — a unregulated power supply. It’s hard to mess this up, much like a flashlight. It’s about balancing cost/size/inrush current vs acceptable ripple mainly…

    1. I think audio amplifier power supplies are often still built this way. Sure, switch mode power supplies are lighter and probably more efficient too, but this is simple and works.

      1. Can confirm. Nicer amps are powered by a bigass toroidal transformer. While, realistically, you could build a switching supply that would be indistinguishable from a regulated linear one without much expense, they’re scoffed upon by elitists who insist they can detect the 1% difference in THD, even after it’s actually gone through the amplifier stage itself. So, manufacturers, who fear appearing “cheap”, tend to avoid digital ANYTHING like the plague in their premium amps

        I actually happen to have a nicer stereo amp sitting on my bench at the moment. There isn’t a single piece of digital electronics in it. Hell, there’s only TWO integrated circuits in the whole thing, and they’re both op amps. Everything else is discrete analogue.

        1. The main difference between a switcher and a passive power supply isn’t the THD from transient currents, it’s the character of the undesired AC components on the DC output. With a passive supply, most of this is 60 Hz ripple, which a) you can always put a bigger-ass capacitor on, and b) is consistent and less noticeable even when above the threshold of perception in human ears. By contrast, the noise introduced on the DC output varies greatly in frequency components across changes in both load and input voltage. Just putting a bigger capacitor doesn’t usually help because of the parasitic inductance of big-ass capacitors, but much worse is that this noise is much more noticeable and objectionable. Most audio mixers also use linear power supplies because they have to amplify microvolt signal components without introducing perceptable artifacts.

          1. Just to be a pedant, the ripple is actually 120Hz. The frequency is effectively doubled when run through a bridge rectifier, as the negative and positive peaks become identical.

          2. I knew I was doing that wrong. You are right – 120 Hz. Unless you’re full-wave rectifying a three-phase source, in which case it’s 360 Hz, but I haven’t seen many three-phase bench supplies. Or audio amps, for that matter, and they wouldn’t be a good idea anyway, since 360 Hz hum would be a lot more objectionable than 120 Hz.

          3. It depends on the design of SMPS. When your switching frequency is high enough, there will be no audible noise caused by it. And don’t forget that LC output filter you have there. Also there are Hi-Fi class D amplifiers (and derivatives from that class) that use switching frequencies in range of decent SMPS.
            Still linear supply is best choice when dealing with very small signals and high gain stages…

          4. If you have switching frequency harmonics audible in the output, you’re doing it wrong. If you have anything as a result of SMPS on the output measurable even with the volume turned up to max, you’re doing it wrong.
            The reasons SMPS aren’t used in audio gear is pure elitism. Kind of the same reason why they go for toroidal transformers (hint, they don’t sound different from others). And if you can hear a difference …. you’re doing it wrong.

          5. Also (c) typical audio amplifier designs are less affected by 120 Hz ripple than high frequency ripple (PSRR drops at higher frequencies). Oh, and toroidal transformers are used because standard EI-core transformers produce strong magnetic fields which couple 60 Hz hum into the surrounding audio paths, not because of audiophile elitism.

        2. Hmmm I wouldn’t have thought they’d actually favor the things. When the volume is up to 11 and the track got busy you’d think they’d pull the voltage down and clip the peaks.

          Though that kinda used to be my favorite thing about linears… you’d need 12V at 500mA, but you’d find a 16V at 400mA and it would pull it down to 12 and prrrrrobably not blow anything up… same in reverse, you’d need 18V at 1 or 200 mA and you’d find a 15 @ 800 or beefier, and it would work fine because it wouldn’t be enough load on it to pull it down to nominal.

  1. More capacitance is great until you consider the inrush or the time constant.
    Thermistors are clever until you turn it off and want to turn it on again.
    Connecting 0v to the mains ground is definitely application dependent.

  2. I like how in basic principle this is how I thought you’d build a power supply about a week after I learned what a full wave rectifier was. Simple principle, it just takes good execution.

    1. Yeah, it was almost a right of passage to build your own bench supply.

      Today, I think people have forgotten linear supplies as almost everything is switching now. Linears still have some advantages and are better for some applications.

      One advantage of a linear is that a well designed PSU will run faultless for several decades.

      I have been considering rewinding a Microwave transformer with many taps to make a variable linear, just don’t seem to get around to it though.

      1. In electronic technician apprenticeship type programs it wasn’t an almost, it was, they’d build own bench supply, and breadboard system and other equipment as part of lab/workshop.

      2. Linear supplies are still prevalent in the amateur radio world as they tend to be far more quiet, rf-wise. When I got my HF radio (shortly after I got my ticket) I purchased a switching power supply that was marketed for radio use. I could hear the darn thing quite well on HF and later found out that when I had it on, quite a bit of noise would end up on my tx signal on my vhf radio which is powered by a separate supply on the other side of the room.

        The effective fix was to improve station grounding by adding an additional ground rod (and also replacing the ground rod clamp on the original ground rod). I still hear some noise of HF but it’s much less noticeable.

        I’ve also learned in my adventure into HF is the large amount of switch mode wall warts in my house just love to generate noise!

    2. The very first thing I made was a power supply. A transformer with a smallish bridge rectifier from a microwave and a biggish capacitor on there, all soldered together dead-bug style and insulated with hot glue. (Don’t worry, the transformer already had the plug in.)

  3. My rules of thumb (sources long forgotten)
    – list out specs – desired voltage Vo @ expected max current Im
    – transformer: if I plan to use a regulator (linear or switching), I’d select a transformer whose secondary Vrms is pretty close to Vo, and current out is rated close to Im.If the PSU is going to be unregulated, then the choice of secondary is a bit harder, depending on how precise a voltage you need (but then, you should be regulating, then!)
    – diodes/bridge usually selected for a current rating of twice Im. I don’t bother stocking rectifiers with less than 50v peak reverse voltage, generally I have stuff that has a prv of 100v or more.
    – filter caps 2,000 uF per amp of Im. Maybe a 0.1 uF poly cap to keep the impedance low at RF
    – linear regulator – if used, 0.01 or 0.1 uf caps from in to ground and out to ground, right on the terminals of the linear reg
    – switching regulators – having great results with the 2596-based ones from the East

    When did I last build a new PSU… summer I think… but that’s mainly because I have a decent pile of complete linear power supplies ripped out of trashed gear. And a bag of wall-warts. Doesn’t everyone?

  4. In psu designs and even power amps used in high quality brand, i’ve seen timed inrush current limiters, can be an resistor in series with the ac lines … or even scr phase controlled bridge rectifiers.

    As ken n says i apply theses rules myself.

    Rob the problems i had in the past with switching power supplies, they did not support inductive loads, spikes etc … had one who could withstand those spikes, but stayed in protection mode when that happened.

    They got better with better designs and carefully crafted output compensation and or protection. I still prefer linear power supply designs, with switched secondary transformer outputs to limit the dissipation.

  5. Bench 101 stuff. But informative to those who didn’t know. Linear all the way for my audio amps, also my bench supplies. (If for no other reason than caps last longer!)

  6. Too big capacitors and you may kiss your diodes bye. Say you have the ripple so small that the diodes conduct only 1% of the cycle. Guess what? Then they have to carry peak currents of 100* your average current! Your 1N400X may not like 100 A.

    1. Notice the resistors between the first and second sets of capacitors in the schematic. This helps a bit with the inrush current, though I’m guessing they’re mainly there to provide some extra hum isolation by adding a pole to the low-pass filter.

      Also: http://www.diodes.com/_files/datasheets/ds28002.pdf shows their particular 1N400X diodes to be good for 1 A average, 30 A peak, although the peak value is “non-recurring”, meaning it’s just for the initial inrush current. But THEN, they cryptically say that these specs are for resistive or inductive loads only, and must be derated by 20% for capacitive loads!

    2. If your ripple is that small on 1n400x series diodes that the only conduct for that small period you’re either running a relatively low average current supply or have an entire room to dedicate to capacitors. Most designs where you end up with this style of design won’t reach this point because they either need regulation anyway in which case ripple is tolerable, or you’re driving something like a class-A power amp in which case ripple is tolerable.

      But if you really get to this point, as always component selection is the key and a 1n400x may not be suitable for you. For $5 you can get an off the shelf bridge rectifier which will happily take 400A peak repetitively, and if you’re pulling that much current $5 is not going to break the bank either. Don’t forget to bolt it to a heatsink.

    3. Having a 1A draw yet the diodes only conduct 1% of the time is a very unrealistic scenario.(and bad design), Besides, underrated diodes would have blown when you first turn on and have to charge that suitcase full of capacitors.

      For circuits where a ton of capacitance is a feature (eg esoteric power amps), there’s usually some sort of “soft-on” circuit that adds temporarily adds some resistance into the diode path, to limit current as the caps charge up. Once the capacitoras are at the nominal output voltage, that extra resistance is bypassed.

  7. I got a pair of image intensification starlight binoculars shortly after the fall of the USSR. I noticed the lightweight nature of the battery charger and couldn’t see a circuit board inside through the vents, intrigued I opened it. there were two components one as each conductor between the battery clip and the mains pins. A single diode to 1/2 wave rectify pulsed DC and a resistor on the other side to drop the voltage. Scary as hell to my western eyes but it worked just fine to charge the unknown chemistry battery.

  8. C or LC filtered power supply without the regulator is just fine for many purposes. I only started using regulators later with digital chips requiring 5V. I still often use additional simple RC filters in the + line (between amp stages) to prevent motorboating (positive feedback through the power supply).

  9. We’d refer to the relay or thermistor circuit switching some resistance in and out to soften the inrush current as a “soft start circuit”. We use them in big, 3 phase high voltage power supplies. Transformers will act like capacitors too being that you have two windings (metal) separated by space. A lot of big transformers will have a shield between the windings to ground to help with that too.

    And I love the audiophile deniers who conveniently forget things like harmonics and aliasing when spouting about how digital is somehow the same signal wise as analog when it comes to audio, just because they don’t know what to measure. But there again, they are trying to compare their ears to an o’scope or spectrum analyzer and don’t see the difference between flesh and blood and brains and electronic hardware….

  10. The title is misleading. It uses the term “bipolar” which implies a bipolar TRANSISTOR! There is NO transistor anywhere.

    While a diode is technically “bipolar,” most people just call them diodes. I have never actually seen a field-effect diode (although supposedly field-effect diodes do exist).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s