Best electrolytics for power supply

[valvesound]

Hi all,

I have a dual split supply for a class A amp which I would like to upgrade.
When I started with the project I used "cheap" electrolytics as I was not sure at the time what type of power supply I would end up with (CRC, CLC, regulated, cap multiplier). The plan was to upgrade components such as capacitors once I was sure about the make-up of the power supply.

Well I ended up with a simple CRC and I think it is adequate. There is no hum from the supply as the amp has a pretty good PSRR.

Q: I have decided on 3 types of electrolytics and would like to know what would be best sounding. Elna silmic 2, Nichicon KG gold tune, Mundorf Mlytic AG.

I currently have 54mF per rail (8 x 6800uF) which my PCB allows for.

I have read many reviews regarding Elna's in the power supply being superior but they are quite expensive and the biggest value is 3300uF at 50V. Thus (8 x 3300uF = 26.4mF).

The Mundorfs I can get 6800uF so would have 54mF / rail.

The Nichicons KG's are the cheapest and 6800uF are also available.

I have also considered mixing Nichicon and elna's 4 of each, would this be any good?

Any suggestions from people with actual experience (not the theory on low ESR etc.) would be much appreciated.

Thanks

[Atilla]

Use whatever you can source easily, you can afford and will fit your PCB. They're all quite nice. For a serious PSU you can consider Panasonic as well. You can get Panasonic and Mundorf in lower profile (shorter) sizes as well, which can make them a lot more convenient sometimes. IIRC the Elnas and the Nichicons tend to be quite tall at higher capacities.

I prefer the Mundorfs, because they're prettier :P

[valvesound]

Hi Atilla, thanks for your reply...."I prefer the Mundorfs, because they're prettier :P"

LOL..I actually prefer the cap that sounds prettier ;-)

[AndrewT]

I have this pet theory that the first C of a CLC or CRC is there just to smooth the mains. It's most arduous duty is handling ripple current. Select for that alone.

The last C of a CLC or CRC is connected directly to the amplifier. It has the most profound effect on amplifier performance after the on board decoupling. Select the last C for sound quality/amp performance.
The amp is to some extent isolated from the first C by that impedance (L or C) and thus has far less effect on performance.

But I tend to use normal, commercial, cheap, caps for the whole PSU. I only get pernickety with the decoupling caps which must be located and type chosen to perform correctly.

[Atilla]

Well,. let's put it this way. I'm using Panasonic caps in my chipamp and it sounds perfectly fine. Nothing fancy is in the supply there.

I'm using Mundorfs in my DCB1 and it's an absolutely brilliant buffer preamp. To be fair though, it's got a nice shunt regulator so I could have stuck anything in their place and it it wouldn't make a difference as long as my voltage didn't sag drastically.

But if you've got money to spare, packaging does matter, studies have proven. At least inside your brain

[valvesound]

Thanks Andrew, this is the sort of information I am after. I have decoupling 100nF panasonic polyprops on each of the "cheaper" commercial electrolytics at the moment. Do you think I would have an sonic benefit by replacing lets say the last 4 caps with more expensive audio rated electrolytics?

[valvesound]

Originally Posted by Atilla

Well,. let's put it this way. I'm using Panasonic caps in my chipamp and it sounds perfectly fine. Nothing fancy is in the supply there.

I'm using Mundorfs in my DCB1 and it's an absolutely brilliant buffer preamp. To be fair though, it's got a nice shunt regulator so I could have stuck anything in their place and it it wouldn't make a difference as long as my voltage didn't sag drastically.

But if you've got money to spare, packaging does matter, studies have proven. At least inside your brain

Hi Atilla, I did a google search on electrolytics in power supplies and which sounded better if at all. The article I stumbled upon tested a pre-amp with regulated supply. The guy had 5 types of caps of which the 3 I mentioned were part of. These caps were before the regulator and they were switched in with a 5 way switch so that he could do A-B-C-D-E testing.

Now I would also have thought that caps before the regulator could'nt have an effect on audio but according to his tests it did and he concluded the Elna's to be best sounding before black-gate and the rest.

I was amazed...or rather confused

[Atilla]

Ah, right. The latest edition of Linear audio has a great article on tests like that and how to draw conclusions on the results. Or more like how not to draw conclusions on the results.

I'm not touching the 'subjectively perceived benefits of power supply capacitors brand' with a 10-foot pole.

Following what AndrewT says here is what I've mostly been doing and then getting good quality caps - in terms of ESR, tolerance, reliability - when possible.

Mr. Pass says he likes Elnas for where he has to use electrolytics, so I've used them here and there. There's no chance in hell I'll make a PSU out of Slimic2 though, unless I strike some fantastic bargain. There are cheaper ways to make a power supply than that.

[valvesound]

I catch your drift Atilla.....a person should stop somewhere with actual benefit vs cost.

[DF96]

Originally Posted by valvesound

Now I would also have thought that caps before the regulator could'nt have an effect on audio but according to his tests it did

Possible explanations are: regulator has poor isolation at higher audio frequencies (remember it has internal NFB which must be stabilised), caps injecting noise into poor ground arrangements, placebo effect.