Another interesting r2r dac

[abraxalito]

I read a post on DIYA recently which said that the Soekris does do NOS when the right filter is selected, seems its NOS in terms of the switching of the R2R elements but the latches still update with data at 8fs, its just the same data repeated 8 times.

[nige2000]

oh....
i thought that was what oversampling was?

i thought the "nos" filter for soekris was just a digital filter bypass but still oversampled

i probably need a clarification on what oversampling and no oversampling (nos) is

i thought oversampling was a way to avoid the nos droop?

[abraxalito]

Oversampling is an increase of the data-rate, but just repeating the same sample eight times totally defeats the purpose of oversampling. That purpose being to move the image frequencies further out of the way of the audio band, making them easier to eliminate with an analog filter.

When you repeat the same sample eight times your image frequencies stay firmly put in the original place (i.e. starting at 24.1kHz). In order to open the frequency gap up you need to calculate new samples - at the very least for the intervening 7 samples so they're not just a repetition of the first one. The new samples are calculated by an FIR filter normally which is doing a frequency domain best-fit operation.

The NOS droop being fixed up is a welcome side-effect of oversampling but not normally the main purpose. Needless to say repeating the same sample doesn't fix it.

[nige2000]

ok thats clearer then

so the soekris then need droop correction with the 'nos' filter ?

i assume the audio gd r2rs will ship after the festival season

[abraxalito]

Yes - assuming this description is indeed what's happening in the FPGA, the Soekris will indeed need drop correction in 'NOS' mode. And ideally it'll need a steep analog filter afterwards or dynamics will suffer.

[nige2000]

dynamics arent great with nos filter as is

might be worth looking into doing something with it

[rickmcinnis]

Main amps are First Watt J2s (made into mono amps) and SIT 1s.
[/quote]

I'll have to investigate them to see if the PSUs are likely impacting their dynamics.

Power supplies are garden variety - I have converted the J2s into mono amp using choke input filters instead of the standard CRC. They are now LCRC - the R being .117 - more capacitance but not ridiculous.

The SIT1s will be converted to choke input during the year - they are already mono.

So your scheme is an "all in one" approach? DAC and amplifier?

Not yet, I'm not yet 100% happy with using the FDA chips as digital amps due to not being sure I've fully optimized the PSU. I'm still experimenting on that. But as DACs (digital amp with modified output filter and step-down transformer) they're my reference for D/A.

What you have shown leaves me confused as to how this is used as a DAC alone. (Which is not surprising)

Not surprising as its something a bit different that no-one else has done (rather like nige's output trafo volume controls!). But there might be some interest for those who're looking for the next step up in dynamics.

I meant "not surprising" in relation to my not understanding what you are dong, that is what is most unsurprising!

For transformers, there is Dave Slagle over here. He would use nickel laminations. One would think those better than ferrite but I have no real idea. What do you think?

Nickel is great but fairly costly. Not sure how DIY-able it is in terms of building your own cores. I use ferrite partly because its got lots of options in cores available but mostly because its cheap. It does have an advantage over nickel in that its non-conductive which makes for better isolation between input and output. This is fairly important in a DAC that's fed over a USB from a noisy PC.

That is an interesting bit of information. Dave is reasonable but a cheaper alternative is always welcome.

I certainly like the idea of a DAC with plenty of driving power.

That is indeed one of the great beauties of this approach - depending on how you do it the driving power can be immense. You can get low output impedance without use of NFB if you think that's important.

No NFB in the SOEKRIS that I can imagine. I am using complementary buffers for my low pass filter below 500 hz. The J2s have had their feedback reduced by 8 db and the SIT1s have none at all. So the only NFB I hae is that that is left within the J2.

I am far from annoyed by the SOEKRIS's politeness - it is not BAD but the idea of better is always intriguing.

Hope others are interested in what you are doing!

I heard back from the audio g-d fellow asking me for shipping information last week but nothing since.

[rickmcinnis]

Got something from audio-gd saying they can ship boards in February.

Anyone else going to try these out?

[nige2000]

Got something from audio-gd saying they can ship boards in February.

Anyone else going to try these out?

well to be fair there aint much point sending them before the festival season

ive a few ordered too, but heard nothing since i paid for em

might do something with the 1121 if i get a bit further on with some other stuff in the mean time

[adolfo.a.aguiar]

I bought 2 to build a NOS DAC.