Buffalo DAC (ESS Sabre 9008)

[Jean-Charles]

The counterpoint does have a very very faint hiss (pretty common with bipolar ccts). I have to put my ear close to the driver to hear it, and this is really only because I have the counterpoint connected to directly to the power amp. There is no analog attenuation to knock it down. I am controlling the volume digitally at the DAC.

Russ

Hi Russ,
I also used LEDs for voltage reference in some of my audio designs. They worked very well dc wise, but I found out they were noisy and introduced an audible hiss in the audio band.
Tubes are particularly sensitive to it.
I finaly got rid of them and raplaced these LEDs with other polarisation methods, the noise went down immediately.
I noticed the Conterpoint uses a lot of LEDs for polarisation, may be it could be a good start for investigation.

 

[mikelm]

a little cap across each LED or across a chain of LEDs will help.

 

[Russ White]

a little cap across each LED or across a chain of LEDs will help.

Yes, I tried this. It did know it down a bit. One thing I did note is that reducing the rail voltage made it nearly disappear. I might try other means of creating the bias voltages too. Or possibly just reduce the current through the LEDs?

 

[Jean-Charles]

a little cap across each LED or across a chain of LEDs will help.

It did not help much in my case. I suspect the noise spectrum was too wide. I did not checked it.

 

[Russ White]

Further update...

Here is something very very interesting. When I run one counterpoint at a time(not sharing a common supply) each is perfectly silent. No hiss at all. So I suspect some loop issue.

I will see if I can track it done.

 

[Russ White]

Eureka!

It seems power supply isolation is key.

I added 4.7R in sereis with the rails and then 47uf to each rail and gnd after the resistiors.

Hiss virtually gone (ear to tweeter can still detect tiny hiss).

 

[Jean-Charles]

Eureka!

It seems power supply isolation is key.

I added 4.7R in sereis with the rails and then 47uf to each rail and gnd after the resistiors.

Hiss virtually gone (ear to tweeter can still detect tiny hiss).

Great news Russ. Do you plan to push your investigations further, or if you stop there?

 

[Russ White]

Great news Russ. Do you plan to push your investigations further, or if you stop there?

Always further my friend.

 

[mikelm]

Great news Russ. Do you plan to push your investigations further, or if you stop there?

I always feel that technically separate supplies is the best route once the left & right channels have gone their separate ways to keep the earthing orderly - it's not often that this actually gives rise to a problem - but in ur case it seems that it has become significant - possible because of the high frequency hash coming out of the DAC.

I would be interested to hear if separate supplies subjectively sounds better than one supply because it sounds like a single supply is messing up ur low noise figures and for me low noise is one of this dac's biggest assets

cheers

mike

 

[BrianDonegan]

The Placid should also help a lot, and it will be easy to stack one with each CP module.

 

[Russ White]

The Placid should also help a lot, and it will be easy to stack one with each CP module.

Absolutely. I am sure it will.

I also changed a few resistor values to make the current through the main string a bit less.

Since that change even with my ear to the tweeter I have no hiss.

This is with the DAC directly connected to a Sympatico!!!

 

[Russ White]

Ok guys, this thing seriously rocks now. I think I have it pretty well tuned.

PCBs (and likely kits too) are available for anyone who has balanced gear and wants to give it ago.

I would not use it directly out single ended because you will get a turn-on and turn-off thump.

Through half a IVY(setup for bal/se conversion) should be just fine.

 

[Russ White]

Counterpoint sim results

This how the simulation says THD should be at 20kHz and 1Vpp(2vpp summed) balanced output.

And this with no global feedback!!!

Fourier components of V(+out,-out)
DC component:1.80661e-011

Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 2.000e+04 9.321e-01 1.000e+00 -12.42° 0.00°
2 4.000e+04 4.064e-11 4.360e-11 -15.38° -2.96°
3 6.000e+04 2.033e-07 2.182e-07 -28.08° -15.66°
4 8.000e+04 3.108e-11 3.335e-11 -50.12° -37.70°
5 1.000e+05 1.074e-09 1.152e-09 -64.70° -52.28°
6 1.200e+05 5.534e-11 5.937e-11 167.47° 179.89°
7 1.400e+05 1.686e-09 1.809e-09 91.83° 104.26°
8 1.600e+05 2.156e-11 2.313e-11 121.47° 133.89°
9 1.800e+05 1.537e-09 1.649e-09 -118.13° -105.71°
Total Harmonic Distortion: 0.000022%

 

[Russ White]

FFT

here is the simulation FFT.

I just measured the real thing with easy lab. The distortion and noise were below the floor of the ADC.

Cheers!
Russ

 

[Russ White]

So funny thing.

The last time I checked over the boards, I noticed that the supply terminal block on right side DAC were just sitting in place. I had forgotten to solder them in. I think this was the heart of the noise problem the whole time.

Doing some regression testing now.

Cheers!
Russ

 

[Russ White]

Excellent news. I have rechecked the original configuration with only some changed resistor values to adjust current through the diodes.

It is working perfectly now with one supply or two. No hiss, no noise.

Color me relieved. It seems it was the loose terminal blocks the whole time. While it was a bit embarrassing to make such a rookie mistake, its far better than the alternative.

Cheers!
Russ

 

[Yunick]

Hi Russ,

From post #719.

I have the input bias voltage set at 1.65V This does wonders for the buffalo!!!

Would you recommend this if I use the Pass D1 I/V ?

Thanks

 

[Russ White]

Hi Russ,

From post #719.



Would you recommend this if I use the Pass D1 I/V ?

Thanks

Yes.

 

[Yunick]

Thank you!

 

[okapi]

Re: FFT

here is the simulation FFT.

I just measured the real thing with easy lab. The distortion and noise were below the floor of the ADC.

Cheers!
Russ

i am quite amazed that you can achieve -165 to -170dB with your test set-up. What sound card are you using and what are the settings (how many bits and sampling frequency).

would you mind posting the response of your testing hardware on its own with the same 20khz stimulus (jump the output to the input)?

thanks