Advice on PSU design

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,
I designed many DACs and ADCs, but now its the first time i make an diyamp.
At the moment i am designing the PSU, that is fed by an sedlbauer 300va torroid. After that i use a 400v 35a bridge rectifier and about 20000uF of caps to fill the ripples (dont know how to call it, my english is not that well).
Now my question :
Besides that they should be low esr, is the difference in caps so big ? I know that in signalpath the difference between a panasonic fc and a cheap cap is groundbreaking. And if you take an audiograde elna its a whole new world.
Is the difference also so big in psus ?

Thanks
Tom
 
I prefer Nichicon KG. They sound much more transparent than Panasonic FC, very audible. And FC are superior to general caps.

You may want to consider a CLCCCC or such. But be very careful to choose an inductor that won't saturate during peaks. You may want to consider something that has beyond 17a capability, and you only need a small amount to be useful such as 15uh. That was recommended to me, the 15uh.
 
You may want to consider a CLCCCC or such.

Definitely - I'm a big fan of series inductors. Without them adding more caps to your supply becomes kinda futile. Use multiple smaller caps (i.e. not the kind with screw terminals aka 'coke cans') to get better noise rejection.

But be very careful to choose an inductor that won't saturate during peaks. You may want to consider something that has beyond 17a capability, and you only need a small amount to be useful such as 15uh. That was recommended to me, the 15uh.

Interestingly I only yesterday built a filter with 13uH inductors. They were wound on no.26 iron powder toroidal cores (cheap). Iron powder is a great choice for supply line chokes because of its soft saturation characteristic - essentially it has distributed air gaps. OTOH ferrite is a rather poor choice (unless over-sized) because it has a very sharp saturation characteristic.

When using supply line chokes, pay attention to their DCR - too low and your supply could amplify, rather than attenuate, ripple. There's a temptation to wind with the thickest wire a core will accept but often, thinner wire will give better damping.
 
I knew that many low capacitance caps are better than one large, so i use 6 3900uF at the moment. How is series inductor meant ? In series to each cap or where should i place it ?
You know psu design very well, so i ask one more thing : what about snubbers at the bridge rectifier ? I read about them, but dont really know if i should use them. In most industrial designs i have seen they are not used.

Thanks for the help guys.
 
Good question - it depends fairly critically on the particular caps. If they have low ESR then more damping is needed in the inductor.

The filter I just designed, I measured the ESR of my 2200uF caps, it was about 40mohm. The 13uH inductors were 20mohm, this had about optimum damping. Going to lower ESR caps would mean putting thinner wire on the inductors.

I should add - the particular filter I built was CLCLCLCLC. Putting caps in parallel means a bigger value inductor could be used.

Added a crop from a pic of the internals of my amp to show what the filter looks like. The 2200uF caps are underneath.
 

Attachments

  • justfilt.jpg
    justfilt.jpg
    27.2 KB · Views: 85
Last edited:
You could use resistors yeah but they don't give such good rejection at higher freqs. This is more important with SMPSUs than with linear though.

I don't have any 10mohm-ish valued resistors in my parts stash, its much easier for me to just choose a few gauges thinner wire :)
 
Last edited:
Given you have 6 off 3900uF, I'd suggest have CLCCCCC. So this means one cap before the inductor and 5 afterwards.

I see. Now i know what you meant by CLCCCCC. So they should be all in parallel. After the high capacitance caps i wanted to use two smaller ones to filter hf interference. Does this make sense to you ?
The last thing i am looking for and find no real help is what input filter to use before torroid. I have seen Ls and C and R in parallel. What would be suitable for mains filtering ?
 
Adding smaller caps in parallel with multiple 'lytics doesn't make any sense at all to me. The purpose of the series L is just that - to filter HF interference. Inductors do quite a good job of that.

Before the toroid I'd advise a common-mode choke. Look for one with the appropriate current rating - too high is worse, not better. Also pick one with segmented windings as these tend to reject higher frequencies better.
 
The PSU is directly in the signal path in most amps.

Please reference or show a schematic diagram for a mainstream power amp that illustrates this seemingly odd engineering feature.

In fact modern power amp circuits usually provide substantial rejection of power supply noise. The influence of power supply characteristics is generally greatly reduced by the nearly omnipresent negative loop feedback.
 
What if I use regulators? Dont I hear the regs then more than the caps?
Im testing at the moment if i should better make the psu regulated. The only reg with low Noise and 5a i found is the lm338. Is there any better that i just did not find with 5a?
Should i stabilize only with the caps or are regs a good way?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.