[Agisthos]

I have been reading all these diy forums for a while.

It occured to me that while a lot of effort is going into listening to incremental improvements that black caps make e.t.c but at the same time most diy'ers are using transformers that they bought from Jaycar.

From what I gather hardly audiophile grade stuff.

Just in recent days Van Alstine Audio has updated his entire line of dac/pre/amps to "wide band, low impedance, multi-stage high voltage power supplies with separate regulators for each section of each circuit"

Looking at the comments he claims that the sound improvements are absolutely spectacular in every department. No doubt hyping his stuff but the idea is there.

Then I came across this Naim mod site http://www.users.zetnet.co.uk/avondale/
Which seems to specialize in power regulator upgrades and they stock AST Audio Grade transformers, which they claim are the best evar.

Just my thoughts. Prehaps Aksa could even do a seperate power supply kit?

[AKSA]

Agisthos,

This is a difficult issue.  I suspect, though can't be dead sure, that there are gains with higher quality transformers.  However, what are the criteria, that is the question?  A few issues:

1.  EIs have appalling external fields at idle, not too bad at full power.  They do, however, have poor frequency response from 10KHz on, which reduces hash intrusion at higher audio frequencies.

2.  Toroids have a FR up to around 200KHz, which is bad, but they have negligible external field at idle, ramping up at high current.  These are the most economical, with low profile, and I use them a lot.

3.  I have also tried Double C Cores, custom wound, with very good results, but hum can be an issue because of external fields.  But they are very good otherwise, though they reportedly take a LONG time to bed down.  When they do, the sound appears to improve.  This is anecdotal;  I'd need to look at a bigger sample, but I suspect the double C core is the best.

4.  How should we load the transformer;  1.8 Tesla, the limit, or down to say 1.2T.  And if EI/Double C core, should we use grain oriented steel, which greatly raises the saturation limit, permitting a smaller, lower profile core (like the transformers I use in the GK-1) with much less external field, or should we stick to toroids?

One thing is definitive;  I've verified this time and again.  It is very important to use two transformers, one for each channel, for full dual mono construction.  This is fundamental to good separation and imaging.

Lastly, are there differences between cheap Chinese trafos and locally made Harbuchs, which follow the Antrim design from the UK.  I have not heard them per se, but I can say that the best amps I heard used these trafos, and possibly the ultimate used a 250VA double C core.  Since an amplifier is a modulated power supply, it certainly makes sense that the quality of the trafo is important, but just how one trafo sonically differs from another is moot.  My impression is the following:

1.  Magnetic loading should not exceed about 1.5T.  More than this creates saturation on heavy passages, which flattens the waveform, injecting high order harmonics into the power supply which are difficult to remove and which add to the HF noisefloor, albeit subliminally.

2.  Core materials should be tightly clamped.  This is particularly true of the toroids, where a loose core will rattle and bang and generally create grief, especially with poor quality mains with residual DC, which overbalances the toroid's core, leading to buzzes during quiet passages.  The construction of the core costs money;  this is a classic area of saving.

3.  There is a proportion of copper to iron which is best for a given power rating.  If the wire is poor quality copper, or a little thin, both to save money, then the DCR of the windings increases and regulation suffers.  Generally, however, the larger the trafo, the better the regulation, and on a 300VA we look for 6.5% or lower.

4.  EI/Double C Core transformers can be scramble or regular wound.  Some even use interlayer partitioning for superior insulation.  They all differ in the interwinding parameters, and this will affect the coupling of HF hash into the DC circuits the trafos are supplying.

Cheers,

Hugh

[Agisthos]

mmmm I can't say I understand it all but interesting thanks Hugh.

Harbuchs then. And they are locally made?

I had seen recently that Norh are using a digital switching power supply that has no transformer at all. Reduce heat e.t.c The only company also doing that I read was Linn who released that integrated amp that was an inch high.

[T-Bone]

Agisthos.
Hugh takes a different approach in the design process. He engineers his products based on what they sound like rather than how they measure up ie; THD, IMD, Slew rate, PSRR ect. You can have the best "numbers" possible and still end up with a poor sounding piece of equipment. Hugh's earlier website addressed various questions pertaining to the the power supply and it's influence on the final sonic character of his AKSA 55 and 100. Keep in mind Hugh wanted this design affordable to the average working man. Many hours of testing using the human ear were performed to obtain the best possible performance using off the shelf power supply components. With an unlimited budget Hugh undoubtably would have used a custom one off everthing in his power supply design, but then the final price to you and I would have become astronomical. About 2 years ago I examined the possibilty of what you are suggesting. Custom built toroids and/or EI transformers as you might have guessed are very expensive. A pair of custom toroids easily cost more than Hugh's AKSA 55 kit. so I shelved the idea. More importantly I'm not real sure the balance that Hugh and his team created wouldn't be upset using a different transformer. Please keep in mind varying amounts of secondary voltages and VA were put to the test. Even the amount of reservoir capacitance along with many different rectifiers were tested (which is probably the most important ingredient) for the best possible sound. I still have one of Hugh's 100w power supplies and when mated with any BJT design I have tested is quite impressive over a single bridge rectifier with snubbers and an x-cap. This is not to say improvements cannot made going to a custom transformer just be advised you quite possibly might be opening Pandora's box.

[AKSA]

Agisthos, Brad,

Some very interesting questions and comments in your posts, and Brad, of the many posts I have seen over the years perhaps yours comes closest to describing how I designed the AKSA and related products.

All Aspen products were designed by ear.  Not just mine, but others too, some of them veritably glowing golden in broad daylight!  If this seems ad hoc, consider the folly of designing solely for good specifications, and then steadfastly insisting the product is good even when it sounds terrible and consumers vote with their feet!  This has been the case for many years,  with scores of amplifiers designed for good specification in the hapless pursuit of 'good sound'.  

In truth, high end engineers have at times taken on an almost surreal priestliness to their craft, and this would be admirable if it had delivered the goods.  A justifiable view might now be that we are measuring the wrong things, and still setting our clocks by the measurements, and flailing about wondering why the sonics are below par.  I have taken a very pragmatic view, engineered something very simple in most cases, and then by careful dimensioning, component choice, and layout, 'voiced' something which sounds very good.  I must stress that others have believed in my approach, encouraged me, many of the AKSAphiles here in fact, and in particular Darl Singh of NZ, whose input with the GK-1 was worth his weight in gold.  (Tragically he has lost considerable weight since those days, Darl we need to talk!!)

The odd thing I've found has been that after the voicing, the careful layouts and choices, the exhaustive listening tests, Robert L. Jones of Newport Beach CA did some measurements on the 100W AKSA in December 2002 and came up with this (thank you sincerely, Robert!):

***********************************************************
Total Harmonic Distortion at 1 kHz into 8 ohm resistive load using HP339 Distortion Analyser
0.2% at 0.1W; 0.08% at 1W; 0.007% at 10W and same to clipping.
At 10kHz and 1W into 8R resistive: 0.042%; at 20kHz and 1W: 0.045%.

Not exceeding 67dB below (0.045%) at any frequency to 20KHz for 10W to 95W into 8R. Distortion profile is second and third harmonic, both 'musical', with all other distortion components, including intermodulation, immeasureable and below the noise floor.
***********************************************************

These results were obtained about a year ago on a stock AKSA 100WN, and show that the voicing procedure has, if anything, enhanced the engineering specification, which is, frankly, to die for.  I believe these figures demonstrate that an audio amplifier can be designed using sonic criteria and still meet rigorous engineering parameters, and that this design approach is at least as valid as any other.  Of particular pride to me was the distortion profile, something I have been rabbiting on about for years, as it is this profile which the ear seems to home in on, rather than individual levels of harmonics or absolute magnitude of the distortion, which as we know with SET amplifiers is very high, often over 2%.

The power supply is very important, and adds the icing on the cake to a good design by ensuring very low noise injection into the power rails.  It's not well understood, but it appears that high speed switching noise from power diodes creates a wash of RF spikes which enter the amplifier and prevent it from delivering high resolution in the audio band.  It makes good sense that an amp cannot deliver sonic detail if it is noisy;  what is not realized so universally is that even if this noise is ultrasonic detail will be lost in the audio band.

Transformers can and do make a difference.  This was never clear to me until I had some very expensive double C cores made for a 100W amplifier.  At first the sound was disappointing, for two months in fact.  Then something clicked, and now the sound is just cosmic.  For the life of me, the only thing different about this amp (which I built to order) was these transformers.  However, and I'm reserved about this, it does appear there is an inexplicably long delay 'breaking in', whatever that means.  I have no explanation for this, but I will probably sit in parks with my fist under my chin thinking on it for years to come........

In closing, Brad comments on cost.  I could probably build an amplifier for the devotee which costs thousands in parts alone.  No problem.  Give me carte blanche and I will built it coated in platinum!! (I won't enjoy it, though, because it goes completely against my frugal nature!)  But I cannot guarantee it will please everyone, since tastes do differ, and it will be very poor bang for the buck.  I've said this before, but most in DIYers want an economical, spare design which beats the big names - fair enough, we all want the best we can get - and I wouldn't sell too many!!

Cheers,

Hugh

Hugh

[Seano]

Harbuch's are indeed local....

http://ourworld.compuserve.com/homepages/harbuch/harbuch.htm

[AKSA]

jThanks muchly, Sean!  Agisthos, my apologies;  I'd forgotten.

Harbuch, run by Peter Buchtmann, is a Sydney company, website as given by Sean.

Cheers,

Hugh