PSU Design Check Please

[ANDYLASER]

I am thinking of making a small batch of PCB's for a universal amplifier PSU. I would be greatfull if you could check my design before I progress further. Thanks.

Component values are likely to change, I have just allowed for the bigest I may need.

 

[peranders]

Are you aware of that we already have a such thread? You could specify by yourself a PSU and see if the members here like it.

(Why do you write in CAPITALS in your profile? This means normally that your are SHOUTING or you are an american lawyer )

 

[ANDYLASER]

Writing in capitals was an old habit of mine from when I use dto teshnical drawings (on paper). Sometimes it does not seem wrong to me. I will change it though.

I couldnt find the thread you mentioned.

 

[AndrewT]

Hi,
that will work.
22mF 80V are not cheap.
What voltage are you planning?

 

[ANDYLASER]

I was planning on using this for a couple of different projects I have on the go. One requires +/- 40V, the other +/- 65V. I doubt I will use 22mf caps, more than likely 10mf will suffice. I just thought that it wouldnt hurt to plan the boards to take the larger cans if required.

Here is a link to the board layout.

Board Layout

 

[AndrewT]

40Vac giving about 58Vdc works for 63V rated smoothing caps, saving a lot of money and space.

 

[ANDYLASER]

I completely agree. But the larger amp will reguire the bigger caps. I thought it would be easier to product 1 multi purpose board rather than 2 different ones. Especially as I will probably need to order sufficient to use a sheet of material, maybe 12 -15 boards.

I am thinking of selling a few of the excess to recoup some of the costs. A more "adaptable" board may possibly find a few extra takers.

 

[jtkraj]

I think you should choose a Caps which will optimize the design. choosing a very high value caps are not necessary for this kind of application.

What is the current rating you are thinking of offering? Calculate caps which will give lesser than specified ripple noise.

you can even use an inductor in series, still the cost will be lesser that using 22mF caps.

 

[ANDYLASER]

Finally got round to doing a track layout. Any critique welcome.

 

[nigelwright7557]

Why have you used 8 diodes when you can do the job with 4 or 1 bridge ?

 

[Sch3mat1c]

1. Use a single FWB. We've been through this before, you are literally burning away twice the voltage for absolutely no gain.
2. Way too much capacitance. You'll have poor power factor (0.5 or less!) and the possibility of induced ground loop anywhere near the power lines. If your circuit actually needs this little ripple, it's honestly a terrible circuit, reconsider your priorities.
3. Those poly caps won't do anything, even if the electrolytics have terrible ESR.
4. The layout shows hair thin AC tracks, was that intended to replace the fuses at the other end? 🙂

Tim

 

[abraxalito]

Any critique welcome.

Don't use groundfill, use point to point (star earthing). Don't assume that because a schematic is drawn in a certain way, the layout must follow it.

 

[Redshift187]

I designed something similar to this a while ago, but as there was no interest and I can't afford to just have 1 or 2 made, it was abandoned. Let me know if you would like the Eagle files (it looks like that's what you're using).

Info post:
http://www.diyaudio.com/forums/chip-amps/147713-interest-check-15a-unreg-psu.html

Current revision:
http://www.diyaudio.com/forums/chip-amps/147713-interest-check-15a-unreg-psu.html#post1886187

 

[abraxalito]

Don't use groundfill, use point to point (star earthing).

Take no notice of what I said, groundfill is fine for power supplies, I've just spiced it. I do recommend though that in place of the big caps you use a few smaller ones in parallel, and put series resistors of lowish values between groups of them. This gives a worthwhile improvement to the HF ripple (where it matters more because the chipamp's PSRR is falling off with increasing frequency).

 

[ANDYLASER]

Thank you guys, I shall do some more playing with the design.