Tír Na HiFi

If the DAC's single-ended I'd go for a PNP emitter follower or darlington (depending on what's prior to the buffer) loaded by a CCS, into a transformer via a coupling cap.

If you need a schematic I can come up with one in LTSpice.

I know I would greatly appreciate abraxalito's ideas on an output stage.

On the DIYAudio thread for this DAC SOEKRIS mentions all of the things it is missing, most of which I am glad it is missing. He mentions there is not a clock on the board. Which is plain. So I guess that FPGA is doing the clocking duty?

Is there a way to add a clock or do you (and abraxalito) think this is needed at all? I certainly would not mind it not being necessary.

Looks like if I set it up for push/pull (which I know will make abraxalito sick) it does require lots of soldering and being the lazy fellow I am I would be glad to not do this.

As far as an output stage, especially for the single board version, I am considering a variation on ZEN MOD's idea of an autoformer, between two buffers. A 1:2 should be enough or do you think 1:3 would be better? How much gain do you think needs to added? I have tried this idea and it works very well in another application.

I guess the simpler version of the buffer with a pair of 2SK170s with a bipolar power supply so one does not have to use an output capacitor.

Did you use any low pass filtering with your initial trial?

Do you plan on removing the 1.8 volts regulator? I guess with the FPGA handling the clocking duty maybe it is more important for this to be battery powered? But what battery? Can't imagine a 26650 being very happy at 1.8 volts.

I got the tracking number for mine today.

Looking forward to hearing how you proceed.

i used this method to familiarise myself with the boards


i think the cpld chip gets shift register clock from input bck by skipping the relevant number of cycles
think its 1.8v level clk signal not 3.3v
maybe a better signal might do something here?

think output is 800mv so could be 1:2 gain easily

an ltspice schematic of a gain stage would be most welcome

So this DAC, unlike the SOEKRIS, requires the output cap? I guess the advantage of the SOEKRIS's bipolar supplies is no cap required?

10uf seems really big and expensive to get a good one.

I guess one could use a gapped output transformer and pay about the same amount of money.

abraxalito, what would you do?

I will start with a one board per channel setup and setup the other two boards for push/pull and hear what the difference is.

the shift register clk is going to be the same frequency as lrclk 44.1 48 88.2 96 192....384khz
on the potato 374 ff has a few spare lanes usually we only use 3 bck lrck and data
we could fully utilize the ff to reclock a bck lrck data + an extra lrck for shift register clk for each speaker channel (or x2)
that uses up the full 8 lanes of the 374

tested linking lrclk to shift register/ff earlier and works good

rickmcinnis wrote: ↑Thu Feb 16, 2017 9:52 pm So this DAC, unlike the SOEKRIS, requires the output cap? I guess the advantage of the SOEKRIS's bipolar supplies is no cap required?

10uf seems really big and expensive to get a good one.

I guess one could use a gapped output transformer and pay about the same amount of money.

abraxalito, what would you do?

I will start with a one board per channel setup and setup the other two boards for push/pull and hear what the difference is.

i used a 1uf film for now,
think we will get it as good or better than the soekris abeit maybe a little different sounding

if push pull works out a trafo will get it into single ended

On the website (as you well know) the designer of this DAC intended this to be used push/pull to mimic the 1704. He thinks very highly of the chip.

I doubt I have ever heard one unless that was the chip used in the THETA Gen V.

So with a transformer one would not connect the outputs together and send each board separately to the input of the transformer? Funny thing is the losses in the transformer would negate the extra gain and one would be back to using some kind of gain stage.

The transformer would take care of the low pass filter!

I want it to sound different. I think the only reason we do this is because we have tired of the sound we are hearing. None of it is right so different is better only because it is different.

I certainly like the simplicity of the thing.

Excuse the rambling.

i heard sima66 Wadia 931/922 i think that was 4 1704s per channel, maybe that was more detailed than this
this probably wins the lowest noise least fatiguing award
maybe were attuned to hearing integrated noise and miss it when it goes

I think many people have become so accustomed to the noise they think it is "right". Adds "excitement" to the music!

Seems they put all of their packages on DHL. They gave me the tracking number but it gives no indication of when it will get here.

Good stuff. Messing around with speaker drivers right now but after that I'll put in an order...