First of all, I must add a couple of important details that I overlooked when I first wrote the article, or are related to modifications I implemented later.
The first issue is the fact that in the final version I have stripped away two of the four 680pF capacitors in parallel to the output of each channel. I prefer it this way. With four capacitors it seemed to me to have not enough very high frequencies.
Just to be clear: no real problem even though you leave all the four capacitors there, but I prefer it with only two.
Obviously by stripping off two capacitors the output filter cutoff is moved to higher frequencies. No fear: I have not yet found a system that gets in trouble for those high frequencies, which are indeed attenuated. Just to give you an idea, I have attached the result of a few simulations: green is the original Convertus output filter, red is the two capacitors only version and blue is the original Kusunoki filter. (4 capacitors but no following RLC filter).
Kusunoki cutoff is at lower frequencies, but the slope is really mild (6dB/octave vs. 18dB/octave in any Convertus version), so that at 500kHz and higher Convertus attenuation is higher than in the Kusunoki's design. Note: the RF choke used in theory could have some core saturation or similar problem, but as a matter of fact I have not been able to hear anything strange.
Second issue: ALL power supply transistors must have a heat sink. I have used improvised heatsinks, so any normal heatsink should be enough. Anyway you should better take care of the analogic power supply transistor, I have used a little metal slab 3cm high and 8 cm wide.
This for sure explains a certain death rate in the PSUs of the Convertus a few friends build, making the big mistake of trusting me...
And now I really think it is the case to answer to our Editor!!!
mmmh... after the Convertus review, given his unbelievable faultfinder attitude, I feel a little pardoned and in perspective ungrateful, but some things have to be said...
No, Lucio, no no and no. The PSU is important, but what makes a circuit sound well are not only side details, box, PSU, wiring, footers. The most important item of all is the circuit itself, the schematics, the underlying idea. Idea which, for the Convertus, is by Kusunoki, not mine.
It is absolutely true that apart the DAC module, everything else has been designed from "white paper" (probably before getting aware of the Kusunoki design): the Convertus PSUs have been designed using very well known concepts assembled in original way, all the digital circuit and its layout is based on the Crystal/Cirrus data sheets, application notes and evaluation cards, the LC output filter is an original idea (as far as an LRC filter can be an original idea...) initally designed for CS4329 and only adapted to TDA1543 and for sure the analog PSU voltage and the output filter modifications are part of that fine tuning which is always necessary to get the best out of any design.
On the other hand I am perfectly sure that even not taking as much care as I did for PSU, given the Convertus schematics, it is really difficult to get something far worse than what I was able to do.
Yes, it's true I spent months testing the prototypes, but when - at the very end - I selected this schematic the result came out immediately, in a hardly regular and optimized way: for example I did not know if I was going to like 1543 sound or not, so I mounted the converters on a little card with a male socket which I inserted in the CS4329 socket of the motherboard I had been testing; I had to make a few changes to the motherboard, but I would have been able to reverse them.
This is no ideal assembly, but I am still using it, which can tell you how much the design is solid and straightforward. With a better assembly technique I might have achieved even better results, but given the situation I am afraid that really few people would have been able to listen to the differences. I do not expect everyone to agree: Kusunoki himself normally gives very detailed informations about system assembly, as he evidently thinks that this a fundamental aspect to be optimized.
In the end, even though the Convertus sound has not been obtained by chance and all parts contribute to it, for me one thing is for sure: the credit for the beautiful Convertus sound goes - first of all - to Kusunoki's idea.
Below you find the articles on the Convertus (theory, schemes, components list etc.):
[Part 1] | [Part 2] | [Part 3] | [Part 4] | [Listening test]
Copyright 2000 Giorgio Pozzoli - http://www.tnt-audio.com
How to print this article