Computer Audio 2016

[Ken Moreland]

BTW I wasn't peeping, I thought I heard a significant improvement!

Great day's listening and comparing topped off with a super Indian dinner, well done to the hosts Anne and Tony.
Nige's Dac was the highlight for me and also the revelation that 2 Jitterbugs actually are better than one. The battery powered Regen too was impressive and all sounding good through the Quad 63's which I prefer to the 57's.
All good fun.
KM

[frd1996]

I also would like send big thanks to Tony & Anne for great reception. I came late and I missed the Hugo Dac and blind testing stage, but I was extremely pleased with what I heard from Soerkris dac - the true performance! Thanks everyone for making this fantastic evening.

[rickmcinnis]

Nice.....looks like you guys had a lot of fun! :)))

Just curious what was the battery supply for the Regen? Was is it some special John's mod? Any pictures?

Simple, really - bypass all regulators on the Regen & supply 3.3V battery supply directly to power clock & USB hub chip - no 5V supply considered or used as I don't agree with USB cable carrying 5V supply - it degrades the sound & any USB audio device that needs 5V Vbus is sub-optimal, IMO.

As Nige, the delusional one, says - a significant jump in sound.

Interesting, since the REGEN fellows think so highly of the TI regulator they use in the thing. Well, maybe one of these days someone will make a really good regulator? We can hope ...

[james]

Interesting stuff!

What's R2R when it's not reel-to-reel, or have you given up on digital completely...? :)

Let me know if the Big 7 is going in a skip, Pearse. I could give you the scrap value for it. There I was this morning thinking I should just nut up and start thinking about a Big 7, only for it to be immediately superseded!

Anyway, great report of what sounds like an entertaining evening.

http://tnt-audio.com/clinica/convertus1_e.html

"...
1.5-Different D/A converter models

Just to give an idea of the different possible implementations and just as a first approximation, we can say that there are different classes of D/A converters. Each class is subject to infinite detail variations, so that it is nothing more then a theoretical exercise.

Multibit converters: in these there is tipically a R-2R ladder circuit which is used to sum currents or voltages each with a weight proportional to a power of 2. The circtuit is extremely delicate as the precision of the ladder is absolutely essential, hence it is rather difficult to implement in practice with an high number of bits. The first CD players used this kind of circuits with oversampling: the interpolator predistorted (noise-shaped) the signal so that when it was converted by a 14 bit DAC and filtered the original 16bit precision could be achieved.

One-bit converters (also called "Bitstream" by Philips and "delta-sigma" by others). The structure is the same as above but oversampling 128X or 256X is required as the digital to analog converter has 1 bit only. The modulator converts the incoming N-bit sample stream into a 1-bit stream which represents the difference (indicated by the greek letter delta) between the current input sample and the reconstructed output, which is the integral (indicated by greek letter sigma) of the 1-bit output stream up to the last sample.

Feedforward or multistage converters (also called "MASH") are made up of a series one-bit converters, each working on the residual quantization error output by the previous stage, whose output is combined to give the main output.
..."

[jrling]

It's all Greek to me!!

[randytsuch]

Interesting stuff!

Multibit converters: in these there is tipically a R-2R ladder circuit which is used to sum currents or voltages each with a weight proportional to a power of 2. The circtuit is extremely delicate as the precision of the ladder is absolutely essential, hence it is rather difficult to implement in practice with an high number of bits. The first CD players used this kind of circuits with oversampling: the interpolator predistorted (noise-shaped) the signal so that when it was converted by a 14 bit DAC and filtered the original 16bit precision could be achieved.

..."

So it looks like this article was from 2000, and back then this was probably true.

The Soekris R2R has a 25 bit ladder (I just counted :) ).
Its possible these days because they can make very precise surface mount resistors for a reasonable cost now. And a FPGA does all the logic, and they are also relatively cheap.

Randy

[jkeny]

One of the interesting approaches of the Soekris is that it uses sign-magnitude signal format instead of I2S which confers more symmetry to the voltages around 0V . DACS that use sign-magnitude signals are a type of R2R DAC called co-linear DACs.

The advantage of these types of DACs is that they have better THD at low volume levels - hence the better rendition of fine small low level detail & a more realistic illusion.