[Rom]

Folks G'day,
These idea has been bothering me for a long while now and reading the all the upgrade possibilities but not a single one mention of having two transformer for a single channel for the AKSA.
The main idea was to have separate power supply for the output transitor completely separate from the driver onwards.
These way the large current drain of the output transistor won't have much effect on the power supply for the front end.
Since I don't have my AKSA with me at the moment I can not make these experiment myself to see if there will be improvement or even sonic benefit ,so as for now it will remain as an idea.
Any inputs Hugh, et all?
P.S. please don't blame me if you burn your AKSA along with your speaker!when you try these, as on the subject above it's just an idea.

Rom

[Malcolm Fear]

So, you want a transformer for each transistor on the board. If this worked, then you would think 8 transformers would be the go, for the AKSA 100?
I have no idea. But then I am of the school - if it looks like a colourful earring, it is a resistor (and Riken Ohms are the prettiest), otherwise it is a capacitor.
I can't see any reason why a power supply for each transistor would be of benefit.

[JohnR]

I think he means separate for the output stage and the input stages.

[AKSA]

I think it's probably a good idea, Rom.

However, I'd go for just one transformer with two sets of windings to power both left and right diff pair/VAS.

It certainly has merit.  Wanna give it a go?

Cheers,

Hugh

[Brian T.]

Rom,

Having a separate power supply for the output devices does, indeed, bring sonic benefits.  I am just about to purchase an AKSA 100 to replace my venerable 1982 John Linsley Hood amplifier.  This has split power sources and you can hear the difference.  (I conducted a test back in ’92 when one of my transformers died.)  However, I do listen to Classical music at high levels.  The 1812 overture is very good for showing up transient response!  Your mileage with the AKSA/programme material may, of course, be different.  The PSU was one transformer with two sets of windings.  See (if it doesn’t come out below:)

http://www.brian.tuckey.btinternet.co.uk/Amp_cct.jpg

Brian  
 
[Edit: removed giant image, it was making the thread unreadable -- theborg]

[Greg Erskine]

I have seen in some designs that use a regulated PSU for the first two stages (class A?) and a normal unregulated PSU for the output stage (class AB).

In theory, this sounds reasonable and logical, but what are we trying to do... make a power supply more complicated than the actual amplifier.

A case in point, over in gainclone land, you can make a very simple amp with about 7 components then you can build the "ultimate" PSU like this.

http://greggbaker.net/ultimate_ps_gc_1.gif

Sort of defeats the purpose of keeping it simple I think.

Greg

[Rom]

Hi all,
Thanks for all the reponse.

JohnR is exactly right of what I intend to do.

Hugh,
Can you elaborate these further.

However, I'd go for just one transformer with two sets of windings to power both left and right diff pair/VAS.

Is these the same as what Brian T here is saying or is it what JohnR's?

Brian T.
Thanks for the response, I'm no designer, just a little hobbyist and your test really give more credible result that doing these on the PS should be a big benefit sonically.

Greg,

I have heard the original Gainclone and the whole nine yards of 47 lab equipment and I was not impress with it, the sound was so clean like being in a sterile room and the sound did not bit me as what Hugh has pointed out to me before. will it be musical?
The intention here is not really to make the AKSA more complicated but to extract all the sonic we can get out of it. I think these is what we all gunning for.

The idea here was to introduce a second power supply just as what we have on the original design and not to  use any regulator at all as what Hugh have on his FAQ, these way I was only using an already proven design.

Hey Mal,

these may be not your cup of tea but having these exotic parts unavailable to me I had to do something diffrent. Anyway these are just more ideas that we as an AKSAphile are here for sharing what we think would be beneficial to all of us. I wish I have all the parts like you to try out.

Regards
Rom

[AKSA]

Hi Rom,

Probably the best way to do this on a 55W AKSA is as follows:

1.  One 160VA transformer supplying the left channel drivers/outputs.
2.  One 160VA transformer supplying the right channel drivers/outputs.
3.  One transformer - a very small one of around 15VA would be just fine - supplying the left channel diff pair/VAS, and
4.  One transformer, as above, supplying the right channel diff pair/VAS.

This is four separate transformers, four separate rectifiers, and four separate filter cap banks.

Leave the existing power supply pcb with Nichicons as it is for each output stage, but for the low current transformers, I'd suggest around 400uF per mA of storage capacity per channel.  So, use 1A ultrafast soft recovery diodes - I suggest BYV26C (I have plenty, would you like me to send some?), and use 440uF capacity for each rail.  I'd suggest 2 x 220uF quality caps, say Rubycon ZL series or Starget, or Hitano EXR, at 50VW.  You can build a nice little bridge with these diodes just by weaving the leads together in a diamond, and bypass with a couple of 1uF film caps for good measure.  You should use small 15VA 25-0-25V ac secondary transformers, identical voltage to the big trafos.  They are not expensive;  but they might be called 50Vac centre tapped rather than 25-0-25Vac.

These trafos should all be wired together;  four in parallel at the primaries, for a single power cord to a single mains switch.  If you want to get cute, you could have one power cord for each channel - it's been done before and there is no reason it would not bring small enhancements to imaging.

I hope this answers your questions, but if you have more, email me privately.  I'm interested to hear how this would sound.  


Cheers,

Hugh

[JohnR]

I'm finding this quite interesting  

A couple of extra little transformers and associated bits would not add that much to the cost...

[Brian T.]

Sorry for the big image guys!      I couldn’t see the effect straight away, as the web hosting servers of our ‘wonderful’ privately owned monopoly telco was down over the weekend.  

Brian

[eico1]

Don't forget to control power-up/down sequence between the supplies like a reversed bias diode from low voltage to high..

Sometimes the second dc supply is stacked on the lower voltage high powered one, ie a 15v on top of a 70 volt to give 85v to supply the input stage. Also, might as well power the driver stage with a third. I think the real benefit is increased linearity due to the higher voltages rather than the extras regulation. Most voltage gain stages have very high power supply rejection factors with all the current sources and such. Of course many fet designs benefit from being driven from a high voltage stage too.


steve

[ginger]

The usual reason for doing this is to prevent the front end diff amp / VAS supply from dipping when the output stage delivers high currents on transients (particularly on loud bass notes).

The series diode in the supply to the front end in the AKSA 55 and 100 designs will prevent this occuring.

That is when the main supply dips due to large output excursions, the front end rail stays up as current cannot flow backward toward the output stage - during the main supply dip time the front end is being powered from stored charge in the rail capacitor - C3 I think it is. This might also account for part of the noted (on other threads here) sonic improvement when using a superior cap such as the Blackgates or Cerafines etc. in this position.

The series diode feeding the front end power is a technique widely used in valve guitar amps which have "loose" or "soft" power amp supplies. It helps maintain good transient response or attack through the preamp and phase splitter.

Will be interested to hear what results are obtained if anyone tries this mod and will be watching this thread for a report.

Cheers,
Ginger