[Greg Erskine]

Happy xmas all,

Put simply, a Gainclone (chipamp) doesn't sound good with large filter caps unless you use a RC network in parallel with the filter cap. I think I remember Hugh saying years ago that large filter caps may destroy the sonic quality of the AKSA. So I was wondering if the same solution could apply to the AKSA.

Here is a link the a thread describing what people are doing with their chipamps and also note the link in the first post to TNT website recommending this mod for all amp power supplies.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=43423

BTW: For some reason, I have an aversion to fiddling with my AKSAs, so I haven't tried.

Thanks

[andyr]

Happy xmas all,

Put simply, a Gainclone (chipamp) doesn't sound good with large filter caps unless you use a RC network in parallel with the filter cap. I think I remember Hugh saying years ago that large filter caps may destroy the sonic quality of the AKSA. So I was wondering ...

Hi, Greg,

It's a bit simpler than the perhaps philosophical question "will huge a PS filter capacitance destroy the magic of an AKSA?"!!

When you switch on an AKSA, you get a large inrush of current firstly from the mains - which increases as you increase the VA rating of your power transformer.  That's why you need a mains fuse which has an amp rating larger than what you might think is required for an amp producing 100 watts a side.

Secondly, you get a high current flow from the rectifying diodes when the filter caps are charging up.  The amount of capacitance which Hugh has selected is balanced to the power rating of the rectifying diodes he supplies ... if you increase this too much, you will blow the diodes!  And yes, you could substitute something else instead of the diodes Hugh supplies ... but that would certainly be risking taking away some of the AKSA magic!

However, if you wanted to "improve" an AKSA power supply, there are a couple of things you could do:
*  use double the number of 2,200uF caps in place of the 4,700uF caps which Hugh supplies.  Multiple smaller caps lowers the ESR of the filter capacitance, which I understand is a "good thing".  But ... it takes up more space and costs more!
*  use 3 x 2,200uF instead of the standard 4,700uF on an AKSA 55 power rail ... or 5 x 2,200uF on an AKSA 100 power rail.  This small increase, I am sure, will not stress the rectifying diodes too much.
*  use something like a 47uF Solen as a bypass cap on each power rail.

Regards,

Andy

[Occam]

Some background info -

Rather than make you experiments derivative of something that was derivative in the first place, look at what the DiyAudio thread was based upon-

DVV's article which originally appeared in TNT a few years back-
'Designing your own power supply'
http://zero-distortion.com/start.htm

and for the numerate -
www.calex.com/pdf/3power_impedance.pdf
http://www.designers-guide.com/Design/bypassing.pdf

The DiyAudio thread is quite interesting, but as per usual, egos have become quite manifest. IMO, there is no universal solution, topology, or specific values. Specifically, because the critical damping of inductive instabilitiy (resonance, such that exists) must address all contributors, not only the capacitor's contribution, but that of the wiring, which is very implementation dependant. Nor is this a situation to throw 'audiophile approved' components at it. Snubbers call for miniscule inductancs in both resistors such as metal film, and specific snubber caps, Rifa PHE450, Vishay MKP1841, Illinois PPB/R, etc..... Inappropriate layout and components add more inductance than they mittigate.
FWIW

[Occam]

Happy xmas all,
Put simply, a Gainclone (chipamp) doesn't sound good with large filter caps unless you use a RC network in parallel with the filter cap.. ...

Not really. This is certainly true if one attempts to out Zen the folks a 47Labs.
These folks, in their minimalist implementation, realized that if they were to sucessfully implement that 9 component amp, (not counting the 100s that were in the 3875 chip) that relying on a single cap per rail  and the intrisically high PSRR of the chip, were constrained to around 1,000uf/rail in order to avoid instabilities. To adopt rail bypassing as specified in the National data sheets (already in the AKSA, and is universal standard practice) would have made the 47Labs less minimalist, less Zen, less marketable.  47Labs was brilliant in making the most of its constrained implentation.[I've heard the 47Labs amps quite extensively. They better any clones I've heard, and they're quite lovely. Just not my cuppa]

That being said, it would certainly be interesting to attempt to improve the PS with a RC damping network, though I believe it would ideally be optimized for both the PS and its specific wiring  to amp board.

[mb]

IMO there's a large dose of "perspective" in the argument too. The capacitance used in the chipamps is really puny -- 1500-2200uF which I've found does have a negative impact on performance, esp. bass, but does provide the nice mid/trebles that the chipamps are pretty good at. I was in sync with Carlos' argument that the chipamps offered with small suppy capacitance have significant problems driving many speakers. I heard an Audiozone (Peter Daniel) amp struggle with a pair of small VMPS ribbon monitors. A pair of Ampzillas did fine, but so did my AKSA55N .

I have used 12000uF on my chipamp main supply, followed by 1000uF Rubycon very low impedance ZL and film cap bypasses close to the chip. The results were very good, but nowhere close the my AKSA55N. Implementation of the RC snubbers at the chipamp end has narrowed the gap significantly.

As a matter of interest, I have added 2200uF in the form of low ESR Panasonic FCs parallel to the NCC 4700uF on the standard supply. To my ears there's been no downside and my 55N now sounds much more "burstproof" than before. My diodes were already higher rated than than the Philips that Hugh supplied, so I think I'm ok as far as diode ratings go. The piggybacking of the 2200uF/50V FCs is *not* pretty. Downright ugly, in fact, but sonically it's one of the best mods I've put into my 55N.

Someday I might get down to putting in Dejan / Carlos' RC snubber on my 55N. If anyone tries it out, do let us know!

[Greg Erskine]

Thanks andyr, Occam and mb

andyr

My AKSA is a 55 so there should be a little headroom in the diodes as long as I don't increase capacitance to far. Thanks for the warning and extra info.

Occam

I've read the tnt and zero-distortion power supply articles many times over the years but never made the connection carlos did. Thanks for the extra two links, these should prove very useful.

As far as the 47Labs, I think going minimalist means the selection of the components is more significant.

mb

I got my old test GC out of the cupboard on the weekend and added the RC snubber across 10,000uF filter caps and was very surprised and pleased with the improvement. How can a resistor and cap make such a difference? I still don't rate the improved GC in the same league as my standard AKSA 55. The results have kindled my interest so I might try a similar RC snubber on one of my other "non-GC" amps, then the AKSA.

I'm so happy with my AKSA I really don't feel the need to improve it, but curosity.......

Thanks again.

[mb]

mb

I got my old test GC out of the cupboard on the weekend and added the RC snubber across 10,000uF filter caps and was very surprised and pleased with the improvement. How can a resistor and cap make such a difference? I still don't rate the improved GC in the same league as my standard AKSA 55. The results have kindled my interest so I might try a similar RC snubber on one of my other "non-GC" amps, then the AKSA.

I'm so happy with my AKSA I really don't feel the need to improve it, but curosity.......

 ...

It's almost a case of which of us will blink first, succumb to the curiosity.

[SamL]

My diodes were already higher rated than than the Philips that Hugh supplied, so I think I'm ok as far as diode ratings go.

What did you replace the stock BYQ28E-200 diodes with?

Sam

[mb]

My diodes were already higher rated than than the Philips that Hugh supplied, so I think I'm ok as far as diode ratings go.

What did you replace the stock BYQ28E-200 diodes with?

Sam

Schottky diodes with 10A rating per leg (vs. 10A combined on the BYQ28E). I'm using a 55N, so it was possible to get by with diodes with 100V rating. I'm rather partial to schottkys.

[Greg Erskine]

and for the numerate -
www.calex.com/pdf/3power_impedance.pdf
http://www.designers-guide.com/Design/bypassing.pdf

Occam, I had a chance to read through these two articles and found them great reading. Clear and at a level where I could understand the concepts.  

Thanks
Greg