Tír Na HiFi

nige2000 wrote: ↑Tue Dec 04, 2018 4:19 pm I've got 2 270s in atm
If I remember right anything between 100 and 150r seemed to work just fine

Here's a LP filter which gets rid of a whole load of ultrasonic stuff. It doesn't do NOS correction - do you need that? It can be tweaked to add some HF boost to counteract the first-order hold function within the DAC. Here I've reduced the I/V resistor slightly as the filter adds extra compliance over a naked resistor. The 13ohm resistor should be reduced by the DCR of the 1.2mH inductor. To drive a transformer in voltage mode I'd suggest a MOSFET buffer - a current source loaded source-follower along the lines of my headphone buffer. The best MOSFET for this will be SSM3J56 which comes in a very very cute package about 1mm long. The CCS is best built with bipolars as they have lower capacitance.

nige2000 wrote: ↑Thu Dec 06, 2018 1:37 am https://www.diyaudio.com/forums/diyaudi ... oject.html
just thinking out aloud
abrax what you think of this?

I had a look at it a while back when I was designing I/V for my own DAC. While I like MOSFETs in I/V stages this one seems to waste a whole lot of power by running from rather high rails. As far as I can see there's really no need for such a waste - the FETs can be biassed with CCSs rather than using fat, sweaty-hot resistors. I also would suggest some more modern MOSFETs (i.e. Toshiba) which have higher transconductance (or alternatively run at lower bias for the same gm), the ones being used there might have been SOTA a decade or so ago but fortunately FETs are improving over time.

abraxalito wrote: ↑Fri Dec 07, 2018 2:48 pm

nige2000 wrote: ↑Thu Dec 06, 2018 1:37 am https://www.diyaudio.com/forums/diyaudi ... oject.html
just thinking out aloud
abrax what you think of this?

I had a look at it a while back when I was designing I/V for my own DAC. While I like MOSFETs in I/V stages this one seems to waste a whole lot of power by running from rather high rails. As far as I can see there's really no need for such a waste - the FETs can be biassed with CCSs rather than using fat, sweaty-hot resistors. I also would suggest some more modern MOSFETs (i.e. Toshiba) which have higher transconductance (or alternatively run at lower bias for the same gm), the ones being used there might have been SOTA a decade or so ago but fortunately FETs are improving over time.

was also looking at the t pears mercury even though im not sure what it is, nearly no info on it afai can see

this the one?

what sort of output impedance should i look in a buffer 500ohm or less?

i think ive j327,35's,k339s to hand

ill get cracking on lt spice then so

Yes, that's the one. You can use J327Rs they're only slightly poorer (higher capacitance) than the J56 but you can run them hotter. I'd run the bias nice and toasty, say about 40mA to get an output impedance down below 1ohm, then use a series R (say 4.7R) to keep it stable with the parasitic capacitance of the trafo.

I also found improving the PSRR of the CCS makes a slight improvement. This is done by adding an extra R - Elvee on DIYA pioneered this approach.

If the DC bias conditions are favourable you might be able to dispense with the input cap, I don't know enough about the PCM1794 though. Only one output cap is needed when driving the trafo with two buffers, do check its polarized correctly or use a non-polar.

Dc offsets are circa +/- 2.75v on the differential outputs at 133ohm iv
Think it's either 800 to 1000mv output ac

Am I right in thinking if I introduce a negative supply I can use the same fet buffer on the negative output to drive a traffo without caps?

Pics from a good days work yesterday, have almost 7 DAC boards ready now and going through touch ups and finishing by Nige and Fran.
Got the cases and just starting to look at layouts and build next.

5 of the 7 boards

And again

Playing with layout for PS on right side of case

PS components but actual Transformer will be smaller

Supercaps fed off Regulator Board

She's coming together quite nicely!!

Great to have that bit of space in the case - if we do end up with an output stage of some sort it be easy to fit it no matter what we use.

Looking good, guys - great progress
Any listening impressions from the get-together?

Nige, what regulator are you using?
Does it stands up well to the near short circuit of the caps when depleted?

jkeny wrote: ↑Sun Dec 09, 2018 8:23 pm Looking good, guys - great progress
Any listening impressions from the get-together?

Nige, what regulator are you using?
Does it stands up well to the near short circuit of the caps when depleted?

Coz all the regulators are in series we opted for plastic backed regs to save worrying about shorts on the heatsink

https://uk.farnell.com/stmicroelectroni ... dp/1087151

These super caps are an absolute pig to charge up. 360 farads
Maybe 20 min( have not timed it yet ) til full charge
Regs go to about 65c without heatsink
So we will be putting heatsink on these

Only I've heard the version with zfoils. As I didn't manage to debugg it before home time sat.
Id say the soekris is sweating though

nige2000 wrote: ↑Sun Dec 09, 2018 9:31 pm Coz all the regulators are in series we opted for plastic backed regs to save worrying about shorts on the heatsink

https://uk.farnell.com/stmicroelectroni ... dp/1087151

These super caps are an absolute pig to charge up. 360 farads
Maybe 20 min( have not timed it yet ) til full charge
Regs go to about 65c without heatsink
So we will be putting heatsink on these

That's a long charge time - are you sure the regs aren't being internally thermally limited?

Only I've heard the version with zfoils. As I didn't manage to debugg it before home time sat.
Id say the soekris is sweating though

Against the tweaked Soekris, that's high praise