DIY SYNTH PSU: How to design a simple dual power supply 

yes definitely! forgot to mention this. thanks for pointing it out!

What did you do? Simply isolate the heat sink from the 7912?

Supertyp there are some special parts available for isolating the heatsink from the tab of the regulator, but these also prevent some conducting of heat. Most of the time I don’t connect the heatsink of the negative regulator to anything on the PCB, if it is mounted there.

Hi, I’m a mechanical engineer working in the power electronics field and have a suggestion on how to electrically isolate the device from the heat sink while minimizing the thermal impact. The best way I have found is to use Kapton tape (polyimide film) which is an electrical insulator with a very high temperature rating. As long as a single tape layer is used the temperature increase will be negligible. I’d lay down a single strip of Kapton tape on the heat sink, then burnish it (rub it) really well. Rubbing it with a hard edge such as a credit card will make sure any air bubbles are removed and you get the best adhesion possible. Don’t use the corner of the credit card since it could apply uneven pressure and put indentations into the tape. You want as smooth a surface as possible, so start at the middle of the tape strip and rub outward towards one end, then repeat for the other end. Gradually apply pressure when starting in the middle each time. Start the final few strokes near one end and finish towards the other to make sure the middle section is evened out. (Who knew applying tape could be so exacting?! Lol.) Once the tape is applied just mount the device as usual, including using thermal grease between the device and the tape to fill any remaining air gaps. It’s really hard to screw this up, short of leaving a huge air bubble right under the device. Good luck! Edit: I forgot to mention that Kapton tape is semi transparent, which makes spotting air bubbles really easy. When you burnish the tape you’ll see a distinctive darkening underneath it as the adhesive wets the heat sink surface and air bubbles are driven out.

You can also directly mount to an anodized aluminum surface, which is not electrically conductive. The only caveat there is to be on the lookout for fresh scratches or abrasions exposing raw, conductive aluminum.

Great help. Thanks

Darn. I just explained the same thing before I read your comment. Note: C(I will buy) > C(I need) / (1 - percentage) the typical +80% -20$ cheap capacitor needs to be scaled up by 1/(1-20%) = 1/(1-0.2)

But when you use a bridge rectifier, you have double frequency, so not 50 but 100 Hz.

@@igorzherebiatev5751 Indeed. So you use 10ms instead of 20 in the formula.

I've seen some datasheets that specify a MAXIMUM capacitance that you can put on the output side, and this does vary by manufacturer. At least one manufacturer provides a formula for determining the size capacitor you should use. The same goes for the very similar LM317 regulators, which are used when you want to use the same part for a number of different voltages, or even a variable voltage. It just requires two additional resistors that determine the output voltage. That "maximum capacitance" can be a problem, because in many cases, the circuits you are operating from the power supply have additional capacitors added to their power inputs. But since that sawtooth you see is caused by the regulator going into oscillation, this is also affected by the inductance between the regulator and the capacitor, and in the case of modules being powered from a power supply, just the wire connecting them together usually has enough inductance to eliminate the oscillation.

@@BrightBlueJim This is an old post. If you build this I think there's actually too much reserve capacitance and it can kill your AC power supply due to the inrush. There are better options.

@@justovision Nonsense. AC wall warts are just transformers, and nothing else. You'd have to put a LOT of capacitance on it to kill the wall wart, and even then, the rectifiers would probably go first. A lot of commercial audio products use pretty much exactly this kind of supply to get bipolar power to run op amps.

​@@justovision- NO, because what is supplying the AC is a transformer. At high current, the voltage sags. So when it starts up, the capacitors form an almost short circuit so the transformer outputs it's rated Current and a very low voltage. As the capacitors charge (perhaps 1 second) the current decreases and the voltage increases. Once the capacitors are charged, the current drawn follows the load. You could add a low resistance high wattage resistor as a "current limiter" however the transformer's output windings and the cables from the wall wart to the PSU will do the same thing in real world terms. It might be a consideration for a PCB mounted transformer and 10's of Amps, but for simple low power stuff it is not an issue.

but what for? 15V isn't gonna hurt anyone and it's constant power wasted

probably a good idea!

came here to say that. i kept expecting him to switch the diode to full bridge

From what I understand, if you were to use a full bridge in this scenario, you'd end up with +12V and 0V, and no third voltage level

Medi from elektroboom would recommend this too

Normally, you would be correct. But in this case, the circuit that he is using is a voltage doubler. So, you can build a doubler with a bridge rectifier.

As an electrical engineer, I WISH I had this channel while I was learning This is better explained and visualized than most classes I had to power through hahaha

it's a beginner's video, but not a beginner's project!

can’t hide my german descent forever i guess!

@@MoritzKlein0 Your English skills are great - it's just our language that is inconsistent. Sorry about that.

In german it's spelled with a k.

Before he was knighted he was just a young ramic

@@MoritzKlein0 English was German at one time until it was tainted :). Seriously though, there is a lot of crossover between the languages. When I was in HS my science teacher had a German pen pal (this was about 35 years ago :P) and one day I was reading over his shoulder and I actually could understand a lot of the letter! My science teacher was impressed. I guess he thought I was some kind of language savant. I am a software engineer, so languages are near and dear to me, but no, it's just because there are a lot of similarities and I noticed the pattern.

Wouldn't that just require adding another "rail" with a voltage regulator with a 5V rating to the 12V input?

AC current is a square wave if you have a cheap inverter.

I don't understand your comment.

AC is a spinning generator, thats why it's sinusoidal. In your home you have three separate AC phases. L1, L2 and L3. If you look att all three waves at the same time you will see L1 start at 0 degress, L2 at 120 degress and L3 at 240 degress. Of a cirkle. And the wave will go over 360 degress. Thats a generator spinning with three outputs spread even over a cirkle. So a completed sinus cycle=A complete 360 degree rotation 😊. I'm sorry. My english is limiting my explanation.

Yes I came to the comments to ask the question, where is the ground rail.

​@@sine0 I think you arbitrarily choose one of the wires from the wall wart to be ground/0v and treat the other like it is the source of the AC current. I think it's more intuitive if you look at the circuit diagram at 24:00 and imagine the ground connections on the bottom half of the diagram being flipped up to meet the ground connections from the top half in the middle. It looks a lot more like a rectifier when you view it that way. It reminds me of how we think of current as going from positive to negative even though it's negatively charged electrons moving in the opposite direction and you sort of accept that relatively it's the same thing.

1N4007 diodes are rated 1000V so your 16V isn't going to harm them. They are rated max 1A so that is the concern here, since the diode needs to pass double or more the current at AC in order to provide 1A output in continuous DC. If you want to draw lots of current you need bigger diodes, perhaps 3A ones. ... You don't need to "respect" the capacitors, since it's well under 20V which can't harm you in any way. Well, it would give you a zing if you licked it with your tongue, but your normal skin is far too resistive to give you any noticeable "shock" up to about 50V. Bottom line, you cannot feel 12V or even 24V or even the 35V from touching the main rail capacitors with 2 fingers.

That is totally irrelevant here, since 50/60 Hz is nowhere near RF. You have been watch too much "Mr Carlson" 😉

you use one side for the half wave rectification (with the diodes, capacitors, voltage regulators) and the other strictly for ground.

the word is „wall wart“ - it‘s a nickname i think!

Wall wart is what Americans call "Wall socket mounted transformers". The term comes from their ugly appearance, like a wart on your wall. n Australia we call them "plug packs", since it is a power package mounted on a plug.

If you can get suitable caps with a value of at least 14100μF, sure, you _technically_ can. All of those 4700μF caps are in parallel to to provide enough capacitance and add up to that. Mind you, finding a pair of caps that big might be difficult: I took a look on Mouser and had issues finding anything with suitable values, because 4700μF is already pretty big as caps go. There are potentially other ways of designing this circuit that could allow the use of fewer caps, such as using inductors alongside the caps for energy storage and smoothing, but that comes with its own set of issues, and I'm not super comfortable dealing with them myself!

thanks! If I'm correct, by this calculation, at 60hz you could realistically get away with one 4700uF cap per rail.

​@@roccotuna- Um, nope. The capacitance isn't halved, it's basically 50/60th.

You can use shunt regulators on the modules to get accurate voltages. Like LM4040 you can use the 10v variant to regulate CV stuff.

This is where you start learning about the joys of Zener diodes!

well the 7812 can push out 1A at max, and the 7912 1.5A. now it depends on how much current your modules draw, but with my PSU I'm powering around 7 or 8 modules. (keep in mind that some/all of them are shoddily designed and use more current than they'd really need to.)

@@MoritzKlein0 amps is a measure of current flow. Watts is a unit of power. The TO-220 package fixed regulators are rated for a max output of 15W. In the TO-3 package they are good for 20W Amps is the same though. Go figure. What you need to do is look at how professionally made things are constructed. Take a PC for example. They have a PSU that supplies rail voltage and then regulation happens for each circuit block (VRMs). Eventually you'll see the light and come around. With electronics you build a lot of power supplies.

@@1pcfred Note that the wall wart suggested also only puts out 1A @ 12VAC; I suspect this will be the limiting factor long before the linear regulators are, due to losses to heat in the vregs and passives. If somebody wants to scale up to a massive rack, it'd be a good idea to either get a professional PSU (something from meanwell, maybe?) or redesign this PSU to handle the higher load. That said, the eurorack community assures me that cross-patching between racks doesn't cause issues as the ground voltages get tied together, so building one of these for every 10 or so modules might also be an option.

@@oasntet powering large projects is a bit of a trick. Expecting to do it all from one regulated source is a big ask. Trying to do it that way you're setting yourself up for trouble. Local regulation is a more realistic strategy. With regulators so cheap and easy to use today it's the sensible way to go. When you look at pro designs they may have a half a dozen regulators on one circuit board. They use overkill to get it done. You have to be comfortable with the complexity when you're dealing with complexity. Otherwise things can get too complicated.

yup, more amperage is never a problem!

Thanks!!

i think with electrolytic caps there's not that much difference in quality. (my local electronics shop is always dangling higher quality components in front of my nose if i pick out cheap ones - but with elcos they never do that.) don't quite get your question regarding the ground connection. are you talking about the actual ground connection on your power outlet?

​@@MoritzKlein0 I was asking myself the same thing when I thought about it for a while. The AC wall wart has two output cables for AC, right? When you connected the oscilloscope in the video, you said to connect the ground connection to just one of the two AC cables from the wall wart. Is the same true for the rest of the ground connections in your circuit? I am just wondering, because I find similar circuits that use 12-0-12 V AC trafos and connect ground to the 0V. Those seem to mostly use bridge rectifiers.

@@christophschmid2456 yeah you choose one cable to be ground - and that cable is ground for everything else, including the modules you power with it. i think what you're referring to might be what's sometimes called "earth ground". it's basically a third connection on your power outlet that leads into the actual ground. as far as i know that's mainly used for safety reasons, but it might also work as "0V" in the circuits you described.

@@MoritzKlein0 No, he's talking about using a transformer having a centre-tapped secondary, with the centre tap used as the 0V point and a bridge rectifier across the outer connections. That gives you a split supply with full-wave rectification. But that can't be done using a wall wart with only two wires coming out of it.

Nevermind, now I got it: we are separating +V and -V from the same AC "pin", the other AC "pin" would be GND.

16V is definitely too low because the wall wart is pushing out >17V. it would work for a short while i think, but the capacitor‘s lifespan will be severely reduced. best practice is to use caps that are rated for 2X the maximum voltage going into them, because this ensures that they’ll last for a long time!

Thank’s ! One last thing, it worked fine for a few days and then the lm7812 stopped working causing the +12v rail to drop to 0, so I wondered what might have caused this ? I replaced it so I will see if it continues to do this

@@Cake11223344 did you attach a heatsink? what were you powering (meaning how much current did you draw)?

@@MoritzKlein0 yes with a thermal paste too, I was powering a breadboarded vcf (clone of the ms-20) so only one module

@@Cake11223344 weird. breadboard or soldered?

depends on how much power your modules need. for me, one of these will power 6-7 modules.

haven’t tried powering my circuits this way, so i can’t really give advice here, sorry! did you ground all the unused schmitt trigger inverer‘s inputs?

@@MoritzKlein0 yes!!! Maybe there's a diode protection on the power supplies affecting the circuit of the synth

@@poetadiffidato it sounds more like a grounding issue to me.. usually when the schmitt trigger spazzes out it‘s because of fluctuations on the power rails

I will try to eliminate this fluctuation! Thank you!!!

a few people suggested that, but afaik it does only work if your transformer has a center tap. otherwise you can't easily get a ground level.

in general, yes, though i would advise that you use a multimeter to make sure the voltages are right.

Uh, I made a bitter mistake: From checking the AC my Oscilloscope was on "AC". In order to see the right graph ("mountains with cut valleys") I had to switch oscilloscope to "DC"...

depends on how much current you want to draw from your power supply. the regulators are able to handle 1.5 A each. but they‘re fine with supplying less.

two reasons: first, the wallwart‘s peaks are around 15, not 12V. second, it’s good practice to use caps that can handle at least double the voltage since they will last longer that way.

some things will be slightly off with +/-9V (volume, gate voltages etc), but it won't break the circuit.

I think I answered my own question. The reason this isn’t the case is because you are referencing ground to not the zero V line, but the “wire you decided to call ground”

You have one lead from the "wall wart" that has diodes hooked to it. You have one lead from the "wall wart" with no diodes hooked to it. This "no diodes" lead is your "nominal ground".

@@kensmith5694 Thank you!

the one i use in other videos deliberately has smaller caps as i use it to only power a single circuit at a time!

@@MoritzKlein0 I have so much to learn still. I have built this power supply and will build it again. I’ve used it to power the oscillator 40106 that you made. It’s all working… but it’s noisy. I’ll do some trial and error. Thank you for these great videos.

you mean because the connectors will be exposed? shouldn't be a problem - the voltages coming out of this power supply are harmless.

@@MoritzKlein0 yes, i mean that. Thats all i need to hear :) Thanks!!

No, as long as it can supply *_enough_* current. Getting an adapter that's unnecessarily big will just cost more money.

yes, the trouble with batteries is simply that they get weaker over prolonged use (i.e. the voltage drops). if your circuits are very wasteful, that happens sooner rather than later. on the plus side, if you create a short circuit, not much will happen besides the batteries getting warm. (while getting depleted super quickly!)