Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 KHz

[Greg Stewart]

No I2S?

Again, I do not know what source you are connecting to your dac and it might be that in your set up you need galvanic isolation. In my case, when connecting SPDIF signal generating chip (inside my cd transport) to dac input circuitry, I do not think I need galvanic isolation. What are you isolating from what? Think about meriads of chips inside your cd player or other source that communicate with each other without galvanic isolation and they are perfectly ok communicating.

So, what for exactly you need your transformers?

I assume this is like a Shigaclone, where I2S is not available? If it was, that would be the best signal source from your drive, IMHO!

Greg in Mississippi

 

[hannesie]

@ folkeb

Maybe a picture of the setup, so we can see how you did make the connections?

 

[formatcd3]

If you are building it directly into a CD player and taking SPDIF signal directly from some chip, most likely that is a TTL output and you could just hook it up to the DAM1021 TTL input. Just make sure it is at 3.3V level. There is no need to mess around with differential signaling.

For the DAM1021 differential SPDIF input, 1.2V DC bias is needed to bring the voltage levels into the FPGA allowed input voltage range. Even ignoring the potential ground loop issue that was mentioned, the SPDIF input will have a 1.2V DC offset. The output circuitry of the SPDIF source is not likely to be built for that. If one side of the SPDIF signal would happen to be grounded at the source end, you will basically have a short circuit. This is not a question of tweaking for subtle improvements, it's about potentially destroying your equipment.

As for SQ improvements with or without SPDIF transformer - what kind of DACs are you referring to? A transformer can affect non-reclocking DACs by introducing additional jitter. For DAM1021 that should not be an issue. On the other hand, a transformer also attenuates some noise, which will affect all DACs, reclocking or not. So in case of DAM1021 having a transformer could actually improve the sound quality.

Thank you for your advise, this is very helpful.

I first connected my cd transport electrical SPDIF (unmodified yet) output to the SPDIF input of dam1021 contacts 9 (SPDIF2IN) for positive and 11 for ground (J3 connector). Dam1021 would not lock. Checked all (six) power supplies of the dam1021 - all voltages were fine.
After that I connected SPDIF output of the cd transport to the SPDIF input of dam1021 contacts 7 (SPDIF1IN+) for positive and 8 for ground (SPDIF1IN-) (J3 connector). Dam1021 immediately locked. The combo - cd transport and dam1021 - work very well so far together. The sound is very good.

Next step will be modification of the cd transport SPDIF output circuit as I described in the previous posts (get rid of matching transformer, two caps and some resistors and replace all these with L-pad resistor network).

After that I will try to connect the cd transport via i2s directly to dam1021. Do you think dam1021 will recognize the i2s signal from the cd transport which most likely is not a 32 bit signal?

This will be followed by more cd transport mods.

 

[formatcd3]

I assume this is like a Shigaclone, where I2S is not available? If it was, that would be the best signal source from your drive, IMHO!

Greg in Mississippi

You are right, I will definetly try to hook up my cd transport to dam1021 directly via the i2s interface. As I mentioned in one of the previous posts, my only worry is compatability of cd transport i2s signal format with the dam1021 one. I do not think my cd transport signal processor generates 32 bit signals.

 

[TioFrancotirador]

thanks finally. is there a cheaper way though? this solution is kinda expensive. ive seen people connecting the output of a triodal trans directly to the board. with no regulaor circuit. i would like a solution that provides decent quality DC with low noise but cheaper than building 3 reflektor Ds specially since i will need another one for the USB I2S mudule. so we aretalking 4 pcbs

Try opening picture in "Open link in new window" option then it should work.
You will need 3 regulators. Two for dam1021 and one for USB I2S module. Like in the picture.
Good power is most important thing in your dac build, more important that dac/chip itself. Going with cheaper regulators here does it make sense. You would do better with some cheap dac like ak4490 and good psu like reflektors than expensive dac like dam1021 with poor regulators.

 

[folkeb]

@ folkeb

Maybe a picture of the setup, so we can see how you did make the connections?

Hi, I'll post pictures tomorrow. My setup is basically just + to pin 5 and - to pin 6 from the 4:1 mux. When I connect the same to my 1543dddac it works. For initial testing I use a cd player spdif to the mux.

 

[formatcd3]

Hi, I'll post pictures tomorrow. My setup is basically just + to pin 5 and - to pin 6 from the 4:1 mux. When I connect the same to my 1543dddac it works. For initial testing I use a cd player spdif to the mux.

Dam1021 locks on when I connect my Kenwood cd transport to the SPDIF input of dam1021 contacts 7 (SPDIF1IN+) for positive and 8 for ground (SPDIF1IN-) (J3 connector). I tried to connect digital out of my Philips CD951 cd player same way and dam1021 would not lock. I will try other SPDIF inputs of dam1021 to see if it will work with the Philips.

 

[folkeb]

Dam1021 locks on when I connect my Kenwood cd transport to the SPDIF input of dam1021 contacts 7 (SPDIF1IN+) for positive and 8 for ground (SPDIF1IN-) (J3 connector). I tried to connect digital out of my Philips CD951 cd player same way and dam1021 would not lock. I will try other SPDIF inputs of dam1021 to see if it will work with the Philips.

I meant pin 9 and 11 of course, to the ttl spdif. I'll try 7&8.

 

[folkeb]

And pin 7&8 locked ☺ My dam is now working. My question is why is this working?

 

[hannesie]

........... ..................

 

[TioFrancotirador]

Hi,

I have Sokris rev 4 board. I already power the board with direct -/+5V (skipping -/+5V onboard regs) and everything works fine.
I wanted to power Vref directelly from Salas Ref D shunt regs running at -/+3.85V
However when I do, it works, it gives the sound, but with very very very low volume level.
I connected the external Vref power to four lines (2 plus, 2 negative) only through middle 100uf capacitors like in picture. (open in new tab to see the picture)

An externally hosted image should be here but it was not working when we last tested it.

Any ideas why is so? Soren?

 

[soekris]

Hi,

I have Sokris rev 4 board. I already power the board with direct -/+5V (skipping -/+5V onboard regs) and everything works fine.
I wanted to power Vref directelly from Salas Ref D shunt regs running at -/+3.85V
However when I do, it works, it gives the sound, but with very very very low volume level.
I connected the external Vref power to four lines (2 plus, 2 negative) only through middle 100uf capacitors like in picture. (open in new tab to see the picture)

An externally hosted image should be here but it was not working when we last tested it.

Any ideas why is so? Soren?

We had it up before, probably the hardware mute circuit, it need minimum -6V on PWR_A-, although -5V probably will do....

I vref mod I have no opinion about, as usually I can't recommend it as the rev4 boards already are better than any external regulators....

 

[formatcd3]

Hi,

I have Sokris rev 4 board. I already power the board with direct -/+5V (skipping -/+5V onboard regs) and everything works fine.
I wanted to power Vref directelly from Salas Ref D shunt regs running at -/+3.85V
However when I do, it works, it gives the sound, but with very very very low volume level.
I connected the external Vref power to four lines (2 plus, 2 negative) only through middle 100uf capacitors like in picture. (open in new tab to see the picture)

An externally hosted image should be here but it was not working when we last tested it.

Any ideas why is so? Soren?

I had the same issue with my rev 2 board. I was powering same way as you do through 3.3V LDOs (one per channel). When I measured the vref voltages they where below 1.9V on each rail (for clarity on each row of shift registers) what is obviously not enough. The sound level was very low.

Then I powered each row of shift registers (four in total) with its own floating LDO providing either +3.3V or -3.3V. So I have now two LDOs per channel. This is logical as the LDOs just replace the removed opamps that were providing +4.0V and -4.0V to the corresponding rows of shift registers. The LDOs output voltages are matched to <0.1% (per channel). I am planning to further improve the matching accuracy to see if this will improve the SQ (I could probably manage to get just above 0.03% with my multimeter and can potentially be matched down to 1uV with better instrumentation based on the LDO specs).

At 3.3V vref I have high output signal levels that I feed directly from the dam1021 r2r resistor network into my power amp (through a passive resistor (10k series Vishay Z foil naked resistor) based volume attenuator though).

 

[formatcd3]

We had it up before, probably the hardware mute circuit, it need minimum -6V on PWR_A-, although -5V probably will do....

I vref mod I have no opinion about, as usually I can't recommend it as the rev4 boards already are better than any external regulators....

Soren, did you measure the accuracy of 4V vref voltages supplied to each row of shift registers of a channel?

 

[TioFrancotirador]

Formatcd3 and Soren, thanks for responses.
My scenario in more details:
I power dam1021 with -/+7V DC through standard inputs, so PWR A+ and A- is -/+7V.
I am using buffored outputs for now from onboard opamps.
I lifted Vout leg #1 from onboard -/+5V regulators and instead of that I used -/+5V external shunt regaultors in that place.
In such configuration everything works fine.

Then, as I wrote, I tried -/+3.85V for Vref directelly. Even 2 x (-/+3.85V) one pair per channel as sugessted by formatcd3. At the same time -/+7V is connected to the board to power digital and opamp buffors, so PWR A+/A- shows +/-7Vs (it that what mean soren here)?
I did not update the firmware yet. Do not even know with what version rev4 comes with. Maybe it is the firmware issue if it is not the newest already.

I just wanted to try different shunts for Vref and see how they sound. I tested bunch of them in my previous builds and I know they differ from each other sonically. Some I got has 1mohm Zout (versus 25mohm onboard), wide bandwithch and I can put it right on the board close to shift registers so for me it is worth trying. Whether it brings and improve something ... I do not know yet, however it is worth trying. You never know untill you try.

Any further thoughts?

 

[uncola]

when adjusting volume using the soekris built in attenuation feature.. does it still reduce digital audio resolution or does r2r magic do it like analog attenuation non destructively?

 

[formatcd3]

Formatcd3 and Soren, thanks for responses.
My scenario in more details:
I power dam1021 with -/+7V DC through standard inputs, so PWR A+ and A- is -/+7V.
I am using buffored outputs for now from onboard opamps.
I lifted Vout leg #1 from onboard -/+5V regulators and instead of that I used -/+5V external shunt regaultors in that place.
In such configuration everything works fine.

Then, as I wrote, I tried -/+3.85V for Vref directelly. Even 2 x (-/+3.85V) one pair per channel as sugessted by formatcd3. At the same time -/+7V is connected to the board to power digital and opamp buffors, so PWR A+/A- shows +/-7Vs (it that what mean soren here)?
I did not update the firmware yet. Do not even know with what version rev4 comes with. Maybe it is the firmware issue if it is not the newest already.

I just wanted to try different shunts for Vref and see how they sound. I tested bunch of them in my previous builds and I know they differ from each other sonically. Some I got has 1mohm Zout (versus 25mohm onboard), wide bandwithch and I can put it right on the board close to shift registers so for me it is worth trying. Whether it brings and improve something ... I do not know yet, however it is worth trying. You never know untill you try.

Any further thoughts?

On rev4 board I believe you can connect vref power supplies (taking into account the required polarity of each row of the shift registers) to, for example, capacitors C167, C170, C173 and C176.

 

[soekris]

when adjusting volume using the soekris built in attenuation feature.. does it still reduce digital audio resolution or does r2r magic do it like analog attenuation non destructively?

It's the Sign Magnitude magic that make the digital volume control that good, the DAC have 28 real bits so you do loose bits as the volume goes down, but there are plenty of bits even at -60 dB....

 

[formatcd3]

Changing dam1021 polarity

Did anybody implemented yet dam1021 polarity switch? I am not sure if dam1021 FW supports this function.

 

[soekris]

Soren, did you measure the accuracy of 4V vref voltages supplied to each row of shift registers of a channel?

Yes, they track very precisely. I also once did an experiment (posted looong time ago) where I made the + and - vref 0.1% of each other, that resulted in only increased 2nd harmonics at 14 db above the difference, so 0.1% difference resulted in -74 db 2nd harmonic.