All "other" aspects which did not fit somehow in the other topics...
Parallel DAC Chips, why I believe it sounds better
Everyone who knows my DDDAC1543MK2 project, knows that I have gone very far here in paralleling DAC chips. Last test we did with
240 chips in parallel and still we heard improvement! This seems in contradiction to the linearity calculations I make on my old web site,
where the maximum based on the SQRT(n) principle was round 60 DAC chips....
Why did it sound better than? I have been thinking a long time about this and came to the following conclusion. It has to do with Jitter reduction...
Interesting, as there has been opponents of my project (they never build my project for sure ;-) who claimed that paralleling chips COULD NEVER sound better,
as so many chips, firing at the same time would cause lots of jitter...
So this is MY theory....
Just think about one single DAC chip doing the digital to analog conversion. After loading a new sample, the DAC will fire and create the next analog level.
We all know that this point of firing time is being troubled by jitter, so the actual time the DAC fires is a fuzzy moment and the
time difference is what we call the jitter.
The moment of conversion is influenced by jitter
Now think about many many DACs in parallel. When the next sample is about to fire,
one of the DACs will actually be the first one to start a chain reaction. A small analog current is starting to flow, than the next
DAC fires and the current increases and so on and so on...
Just for your understanding. Every sample, this will be a totally different sequence of DAC chips (as of the jitter ....) Exactly here lies
the benefit, as the analog signal will build up every time roughly the same with the middle of the signal jump at the same spot (well almost, as we do not
use an infinite number of chips) You could simply say, the jitter is being reduced by the number of chips. On top we have the improvement
of better ( SQRT (#chips) ) linearity, so that is an extra.
Thanks to the sequential firing of the DAC and parallel current output, the jitter reduces
I recently have been reading, that someone was done to the opinion, that this sequential building up of sample output, would cause a
rattling sound. This of course is totally wrong, as rattle would mean there is pause between the samples and that cannot be the case. Assume the
jitter would cause a bandwidth where the dac chips will output their sample of 1nS (that is pretty poor...but just assume worst case), now look at the minimum
length of sample. This is worst case 1/192kHz which is roughly 5uS, 5000 times the jitter bandwidth. No rattling to be expected ;-)
The effect is less when the number of chips starts increasing. Do not forget, these are NOS DACs which are anyway less influenced by jitter as you might know...
But anyway, this insight pushed the design of the 1794 DAC also in paralleling DAC chips. Here it is a bit more difficult to go extremely far in
number of chips (for technical AND budget reasons, but I also found out that the 1794 chip, as a single chip, starts at a clearly higher level than the TDA1543,
so we have an advantage to start with. I started with 4 boards with double DACs and the improvement over one board (Deck) is clearly audible.
Now up to 8 boards, see if that works fine :-)