Signalyst DSC1

Do I need 3.3v on other pins: 7, 10 and 11.

Just on the pin 10, pin 9 is the other 3V3 pin, but it is not connected. Note the non-standard pin numbering on the Amanero connector! So pin 10 is the top-left pin (you can see trace on the solder side goes to C1).

And for SCJ1 I should short 1 and 2?

Yes... This makes the DSDEN signal control latch-enable on the 595. So when there is no DSD data, output latches are in high-impedance state.
 
mcluxun, sorry for my memory about R1, I never set it to 1K when I check it just now.
The gain was by value=432r of R806, Sounds really great but loudness still not enough.
When I try to chang R1 to 1K from 500r, distortion can be heard.15K/32=468.75r, 8K/32=250r you know. So set the gain on output stage is the right way.
 
I didn't get any notable increase in distortion by setting R1 to 1K. 2nd harmonic in the output is around -80 dB and higher order harmonics at lower levels. Increased level at the intermediate stages increases distortion a little since it is relative to level, but also increases SNR.

Reason why I changed my mind for this is that at 0-point half of the latches are on, thus 15K/16=937.5 and 8K/16=500. So on average, the source is half of the total resistance.

You can of course also change gain by placing suitable resistor at R806/R816 which are there, but marked "NI" (Not Installed) for that reason.
 
Sorry for my word expression. The source impedance is 468.75r on full range and the source current is 10.67mA p-p, mapped to R1 should be 10.67V p-p if R1 is 1K, gain=-2.13. Sounds seems been clipping distortion. Perhaps this is 'a case' on my side.

It shouldn't ever have more than about 75% of the elements in same state, otherwise the input signal is illegal...
 
To make it easier I ve actually shorted the relay.

Right now I'm still getting very loud noise. Any hint where should I start looking?

It would be good to verify the wiring if you are not using Amanero input board...

Of course supply voltages for all chips, check that the enable line at shift registers is stable so that the latches stay enabled. Scope or logic analyzer on bit clock and data lines at the shift register end. If you have spectrum analyzer, then checking out the output at resistor network.

This design should be pretty simple to debug even with just plain scope, since there's just one clock and two data lines...
 

mcluxun

Member
2010-11-06 7:07 pm
It would be good to verify the wiring if you are not using Amanero input board...

Of course supply voltages for all chips, check that the enable line at shift registers is stable so that the latches stay enabled. Scope or logic analyzer on bit clock and data lines at the shift register end. If you have spectrum analyzer, then checking out the output at resistor network.

This design should be pretty simple to debug even with just plain scope, since there's just one clock and two data lines...

So embarrassing.... it's a grounding issue.
here's the 1k sine wave
Jtsy1gn.jpg
 
Good tool! I am very interested to see what will happen after this:

In this picture there is still missing PSU feed for IC1/IC2 that is normally coming from the Amanero +3.3V output...

I tried to make a clarifying picture:
[IMGHTTPDEAD]http://www2.signalyst.com/tmp/dsc1-brd.png[/IMGHTTPDEAD]

(the DSC1 PSU board has suitable 3.3V output for the purpose)
 
In this picture there is still missing PSU feed for IC1/IC2 that is normally coming from the Amanero +3.3V output...

I tried to make a clarifying picture

(the DSC1 PSU board has suitable 3.3V output for the purpose)


Yes, I understand you point out. Perhaps the power circuit and the signal circuit should be consider in next layout. Expect Dsc-2.0.
 
Yes, I understand you point out. Perhaps the power circuit and the signal circuit should be consider in next layout. Expect Dsc-2.0.

I think I can add separate power connector for the 3.3V, but for the signals I prefer to have a direct plug-on daughter card for the interface instead of wires... I was thinking to add place for JLSounds OEM or I2SoverUSB board too. Or maybe just two PCB variants.
 
And ad844 for IV looks overwhelmed, lm6181 is the good one in use.

It should be quite easy to experiment with different ones. LME49713 has also been proposed and I will probably try it out (TO-99 version can be fitted).

But I kind of regret not sticking to AD797 for the filter stage, I've had very good results with it in previous designs. So mayble I'll go to single ones in the next version for more flexibility. I also realized later that duals used this way will limit the filter performance somewhat. AD797 needs a bit more complex and different surrounding design.

I'm not sure if I should also switch to SOIC-8 versions at the same time, but that makes socketed approach harder for those who would like to easily swap & compare. Opinions on this?
 

Terry Demol

Member
2002-04-07 3:12 pm
*
It should be quite easy to experiment with different ones. LME49713 has also been proposed and I will probably try it out (TO-99 version can be fitted).

But I kind of regret not sticking to AD797 for the filter stage, I've had very good results with it in previous designs. So mayble I'll go to single ones in the next version for more flexibility. I also realized later that duals used this way will limit the filter performance somewhat. AD797 needs a bit more complex and different surrounding design.

I'm not sure if I should also switch to SOIC-8 versions at the same time, but that makes socketed approach harder for those who would like to easily swap & compare. Opinions on this?

SOIC-8 to DIL8 mini pcb adaptors are an option and easily available.

797 has stability problems if not treated carefully. The real advantage of 797
is very low CM distortion so if your circuit has either inverting config and or
low source impedances you may get as good performance with other opa's.

Refer Samuel Groners opamp measurement paper.

Have you done any distortion measurements on this DAC. I am interested
where the linearity bottle necks are with this design. I am guessing with
DSD128 and higher it should measure very well.

T
 
797 has stability problems if not treated carefully. The real advantage of 797 is very low CM distortion

I know, I've been using it for long time in various DACs. It has also good GBWP suitable for active filters even in MFB type setup.

Have you done any distortion measurements on this DAC. I am interested
where the linearity bottle necks are with this design. I am guessing with
DSD128 and higher it should measure very well.

Yes, I have done some. These results are with the modifications I posted earlier:
Preliminary measurement results for the DSC1 DAC - Blogs - Computer Audiophile
More DSC1 measurement results - Blogs - Computer Audiophile

DSD256 is optimal rate for this as it is now.

I know how to improve this further. But I'll make a new board for those improvements. I'm quite happy with the result for first board iteration. Especially how it sounds... :) Full design in one month on my "free" time.