Gainclone Powersupply question.

[Garbz]

Firstly excuse me asking this here, but I hardly consider it worthwhile to subscribe to the relatively slow gainclone forums to ask a single question which could probably be answered by anywone (other then me) anyway.

I'm whipping up a 2x40watt amp using the LM4766 Overture chip (chose it because I found it in my parts bin). It doesn't need to be quality it just needs to work.

So anyway powersupply wise I will feed it with a 160VA 18+18 torroid. This will be followed by a single diode bridge, which will be followed by ... ahh here comes the question.

I won't be using a regulator with this amp. It needs to be cheap and nasty, so i'm trying to get an idea of what capacitors I need. My original scribble chose 4x2200uf (2 per rail +/-25v), with 1000uf caps localised next to the chip itself. Now here's the problem. My first experience with tube amps produced a noticable hum using 1 4700uf cap.

What type of filtering do I need for this? Should I go a CRC filter of say 4700uf, 10ohm, 2x4700uf? That should drop the corner freq to about 1.5Hz and eliminate most of the hum correct? Like I said i'm looking for the cheapest way to make a ~4.5Amp PSU.

 

[rellik]

another method if sound quality is not crucial

you could pick up 2 astron psu's on ebay for really cheap and wire em up into dual supply, and they got some with pretty high amp ratings

 

[nikongod]

i think the powersuply noise rejection of a tube is pretty crap (generalization) compared to a piece of solid state gear.

 

[amb]

Garbz, just the rail caps should be fine with no CRC or CLC filtering. In fact I advice against putting any resistance in series with the rails. Make sure your power switch and diode bridge are sufficiently rated as well. If you want a bit more reserve, you could go up to something like 10000uF per rail.

 

[MaxW]

Dont use that much capacitance. You will loose detail if you go over about 1500uF per rail per chip (well with LM3875/LM3886 anyway) which is what the original gainclones used. Its been discussed and tested many many times on DIY Audio.

If you want to use more then use a snubber:
http://www.chipamp.com/images/ps-rev3-sch.gif

 

[amb]

Quote:

Originally Posted by MaxW Dont use that much capacitance. You will loose detail if you go over about 1500uF per rail per chip (well with LM3875/LM3886 anyway) which is what the original gainclones used. Its been discussed and tested many many times on DIY Audio.

Curious. Was the "testing" objective? I fail to see how adding a healthy dose of rail capacitance could cause loss of detail in an unregulated supply. 1500uF is woefully low for speaker amps IMHO, and makes for rail voltage perturbations under load. Maybe the extra "detail" perceived with such low capacitance is actually dynamic distortion.

 

[MaxW]

Quote:

Originally Posted by amb Curious. Was the "testing" objective? I fail to see how adding a healthy dose of rail capacitance could cause loss of detail in an unregulated supply. 1500uF is woefully low for speaker amps IMHO, and makes for rail voltage perturbations under load. Maybe the extra "detail" perceived with such low capacitance is actually dynamic distortion.

Hehe, I know where you're coming from. By tested I mean LOTS of people tested it and came to the same conclusion. There are threads that are pages and pages in length about this subject most is over my head but Im sure you'll be able to understand.....

"Lately I've been thinking...
Why the hell don't these chips like high capacitance?
Why do they sound bad?
The reason is so simple that in the end I just wanna kick myself.
Bigger caps have higher inductance, and the amp will have lower damping factor at mid/high fequencies.
The thing is: these chips are really sensitive to this.
Go higher than 1500uf per rail/chip and you start noticing that the midband and treble magic goes away.
The more capacitance you add, the worse it sounds."
From here: http://www.diyaudio.com/forums/showt...threadid=43423

That thread is one of many and is 43 pages and counting, have a read

 

[Voodoochile]

Quote:

Originally Posted by Garbz My original scribble chose 4x2200uf (2 per rail +/-25v), with 1000uf caps localised next to the chip itself.

To comment on this part- I interpret this as you are considering ~25v caps for this application. I think you really ought to consider 35v+ instead, as you are skating awefully close to 25v after the bridge.

If you are just referring to the approximate V per rail, but considering larger voltage rated caps already, then nevermind!

 

[DaKi][er]

Quote:

Originally Posted by MaxW Hehe, I know where you're coming from. By tested I mean LOTS of people tested it and came to the same conclusion. There are threads that are pages and pages in length about this subject most is over my head but Im sure you'll be able to understand..... "Lately I've been thinking... Why the hell don't these chips like high capacitance? Why do they sound bad? The reason is so simple that in the end I just wanna kick myself. Bigger caps have higher inductance, and the amp will have lower damping factor at mid/high fequencies. The thing is: these chips are really sensitive to this. Go higher than 1500uf per rail/chip and you start noticing that the midband and treble magic goes away. The more capacitance you add, the worse it sounds." From here: http://www.diyaudio.com/forums/showt...threadid=43423 That thread is one of many and is 43 pages and counting, have a read

Just straight capacitors of high value will do this, but carlosfm came up with the snubber to add in to allow for high value caps on the gainclones which has been proven many times over and over to work very well by the countless people who have used it and even the theory and measurements have been brought up about it as well in another thread (I think started by peranders if you want to search)

Garbz - there are heaps of threads on DIYAudio, start with the one linked by MaxW

 

[amb]

Quote:

Originally Posted by MaxW Bigger caps have higher inductance, and the amp will have lower damping factor at mid/high fequencies.

If that really is the only issue then it's easy to cure... just bypass the big caps with smaller, low inductance ones.

Anyway, thanks for the pointer to the diyaudio forum thread, I'll read through it later. I used to read that forum regularly but have given it up for two reasons: 1. Too much ego amongst some members; 2. Not enough time to keep up with so many forums.

 

[mono]

I'll 2nd (3rd, 4th, etc) the notion of adding lower value fast (film or ceramic) caps, as close as reasonably possible to the amp IC.

 

[Garbz]

oh definatly the small decoupling caps were never in questions.

No i didn't mention the voltage of caps I was going to use but I'm perfectly aware to go larger. Infact my choice was 50v 4700uf caps. Looks like I can at this point simply toss in 1 per rail, maybe bypass with lowesr caps, and finally hit the chip itself with some 0.1uf films.

Another thought just occured to me today. The amp will be a replacement output stage for a fubared receiever with an intermittent channel (i've already sprayed 2 cans of freeze on it and am at this point writing it off). But the receiver will still be used as a pre-amp, and should in theory be perfectly good powersupply wise.

This is going to sounds stupid but here goes anyway: Is it "safe" to assume the power going through 2 giant caps in the receiver is the higher voltage for the output stage? If this is the case and they measure at ~+-25-40V then i should be able to just tap the power at the base of the caps and feed that into my gainclone right? That would save me ~$100. I'll post pictures of the receiver guts tomorrow.

 

[amb]

Quote:

Originally Posted by Garbz This is going to sounds stupid but here goes anyway: Is it "safe" to assume the power going through 2 giant caps in the receiver is the higher voltage for the output stage? If this is the case and they measure at ~+-25-40V then i should be able to just tap the power at the base of the caps and feed that into my gainclone right? That would save me ~$100. I'll post pictures of the receiver guts tomorrow.

Not only is that a good assumption, chances are those rails provide power not only to just the output stage but the entire power amp section (depending on how fancy a receiver you have).

 

[Garbz]

Sherwood. Definatly not fancy.

 

[Garbz]

Ok we're back. And I think we may be back to square one.

Here's a pic of the sherwood innards. The two caps towards the left are the ones i'm talking about. They are right next to the output stage.

THe output voltage is +/- 60v which is 120v total. The problem I have is according to the datasheet of the LM4766 supply voltage (|vcc|+|vee|) is 20-60v which A) may be cutting it too close, and B) would probably cause a bit too much heat output for my liking.

On top of that I don't know any easy way to turn 60v to +/- 30v. Does the resistor divider in a basic CMoy still work with high voltage high current applications ?

So anything useable here? Or am i back to my 150VA +/- 25v + rectifier + few caps?