Which DIY PCB Power Supply Kit?

[Huey]

[Cross post from the L20D IRS2092 Class D Thread]

I've ordered a L15D Amplifier Kit and now I need to put together a power supply for it. The amp is 150W RMS x 2. I may use it with a 4 ohm load in the future. It will be used for either 2 way speakers or subwoofers.

I have a 300VA 38-0-38 toroid transformer and I'm looking at buying a power supply PCB and component kit. The two I'm looking at are:

68pcs 220uF board: 68pcs 75V 220uf Original RUBYCON YXF Capacitor Power supply board kit /PSU kit #0506 13-in Other Electronic Components from Electronic Components & Supplies on Aliexpress.com

or this 8x 2200uF board: 63V 2200uF 8 Rectifier Filter Power Supply Board Kit | eBay

Any thoughts on which would be best? Or other suggestions?

 

[ticktock]

Bear in mind many of those chinese boards come with fake lytics. I doubt these are genuine Rubycons.

 

[gootee]

I think that you will need and want a lot more capacitance than either of those boards has. I think that you will want at least 22000 uF per rail per channel. And you would need much more than that to be able to get near 150 Watts without clipping, into 8 Ohms.

Also, since you are contemplating using it to power subwoofers, you can't skimp on the capacitance and still have any hope of reproducing low bass tones at high power. The peaks of bass tones are very wide; much wider than a charging period. So they can drain your capacitors as if you're consuming constant DC at the peak level. Without enough capacitance, the ripple voltage will then be quite large and it will lower the rated max output power. And the PSRR (power supply rejection ratio) of those amps is not that high, so too much ripple will also distort the output more than usual.

I would also look for a power supply that has a 35-Amp bridge rectifier (or two of them would be better) that bolts to the chassis.

Why not just build your own power supply? Ten 4700 uF caps for each rail would probably be good-enough. You could even leave space for ten per rail and only buy ten at first and install five per rail and see if it's OK for your application, and then only buy and add the other ten if you need them.

 

[Huey]

Thanks for the response gootee. I looked at my numbers again and used the Charlie Laub spreadsheet. I was using the entire capacitance value to calculate ripple voltage but it looks like I should have just been using half (aka just the capacitance on a particular rail).

Since this is my first DIY amplifier / PS project I'll probably just go with a little over 20,000uF per rail to start (around 5 - 4700uF capacitors). I don't plan on driving the amp too hard initially. I'll leave room for additional capacitors in case I do switch it up to sub duties.

Any other recommended components for a basic power supply besides the capacitors and a bridge rectifier (or two)? I plan on having a drain down resistor.

I'm looking at these 63V 4700uF capacitors right now from Cornell Dubilier: SLPX472M063E3P3 Cornell Dubilier | Mouser or these off of ebay: Rubycon USR Series 4 700uF 4700uF 63V Snap in Electrolytic Capacitor Box of 20 | eBay. How prevalent are fakes? That seller is US based and has perfect feedback.

Should I look at something else for a budget minded power supply cap?

 

[ticktock]

I'd prefer a reliable vendor like Mouser, RS etc. as they aren't neccessarily
more expensive, and consider these (if going for 4.700 µF) with at least
3 A ripple current:


ECO-S1JA472DA Panasonic | Mouser

ECO-S1JP472CA Panasonic | Mouser


LKG1J472MESCCK Nichicon | Mouser

LLS1J472MELZ Nichicon | Mouser

LKG1J472MESCAK Nichicon | Mouser


B41505A8478M002 EPCOS | Mouser

B41505A8478M000 EPCOS | Mouser

B41505A8478M007 EPCOS | Mouser

B41303B8478M2 EPCOS | Mouser

B41505A8478M007 EPCOS | Mouser

B41303B8478M7 EPCOS | Mouser


However, I'd go for one 10 or 15 mF lytic on each channel + one 1 - 2.2 µF MKP bypass.

 

[gootee]

Go with Mouser. (Always go with Mouser.) Get the lowest ESR (Equivalent Series Resistance) caps you can. But anywhere in the same ballpark will be fine. All of the brands that ticktock mentioned are great. Too many links for me to look at, right now. If you narrow them down and have any question just ask.

I think I read somewhere that the taller thinner ones are better than the shorter fatter ones (in case there happens to be a significant choice).

I know that many people like a small film cap to bypass the electrolytics. But I and some smart people disagree. And there's a whole in-depth thread about it, here. The impedance at the power supply output is not that important. The ONLY place it matters is RIGHT AT the points of load. So put your small decoupling caps THERE. If placed by the reservoir caps, the inductance of the rail conductors between there and the load means they can't do you any good, at the load, anyway. And film caps paralleling electrolytics is also a bad idea because it can form an LC resonance, with the un-damped C of the film cap and the inductances of the electrolytic and the power/gnd rail conductors. So put another 4700 uF (or at least 470 uF, but it depends on the current demands of the load device) and a small X7R ceramic in parallel right at each power pin of each active device (to ground).

For your power supply, you can do one other good thing: Make accommodations to have a resistor and small cap in series, between/across the terminals or connections of each transformer secondary winding, in case you need snubbers. If you are tempted to put a cap across each rectifier diode, just don't. Without a resistor in series, the best they can do is just lower any resonance's frequency. You could put a series RC across each diode. But one set across the secondary will solve any LC resonance problem and has the added benefit of better-limiting the high-frequency garbage that you put back into the AC mains.

I would also use BIG honkin' rectifier bridges, like the 35-Amp ones that you have to bolt to the chassis.

Also remember, very well, to keep the two wires or conductors of ALL pairs right against each other, everywhere. So, tightly-twist all transformer windings' pairs, and route the rectifier in-out pairs always with the absolute-minimum gap between them. If one transformer primary wire is supposed to go to a switch and/or fuse, you HAVE to take the other one with it (tightly twisted), and back, even if it looks useless to do so. They must not EVER have any open gap/space between them. (Otherwise, you have made some good antennas that will emit time-varying magnetic fields, which every other loop with any enclosed area will receive, inducing hum currents in those loops. See Faraday's Law.) If you have ANY wire or PCB trace that goes anywhere BY ITSELF, you're doing it wrong.

 

[garrbage]

There is no doubt that Rubycon YXF 75V caps are fake. Just look into the data sheet for the series. voltage goes ... 50, 63, 100 no such thing as 75.

So put another 4700 uF (or at least 470 uF, but it depends on the current demands of the load device) and a small X7R ceramic in parallel right at each power pin of each active device (to ground).

I agree with gootee but one thing. Never, ever use X7R in the audio applications, with all the issues of such dielectrics major one that biasing let's say 4.7uF X7R (1206) cap with 20-25 volts of DC you will get only about 1/10 of capacitance. It is not suited for audio.
If you need to bypass something then use PP/PE caps. If in a tight space you need to use ceramic do not use anything worse than NP0/C0G caps, yes they are expensive and difficult to get anything larger than 0.1 in 1206 package, but there is no other way

 

[MMAdams]

Rubycon 75v 220uF times 68 capacitor assembly

Does anyone have any hard experience with these Chinese capacitor assemblies? I have four that I have assembled but not tested and without resources to test these boards with.

Thanks for your help.