I listen to a fairly broad range of music from 90's pop to orchestral. Modern Kpop to 70's Rock. Is there a way to edit my post above? There were several grammatical mistakes I'd like to amend and I was fairly inebriated when I wrote it. IMO The Celibdache is a solid DAC. I think if the compression could be fixed to add more nuance you'd have something that could really compete at a higher price. Is this a power issue?
I think after the initial 30mins expires the only way to have your post edited is to ask a friendly moderator to do that.
The compression you report isn't something I've noticed in listening but George posted up some linearity measurements of the TDA1387 a while back (over on the lingDAC thread) which leads me to think that the DAC chip itself might be adding a little compression. For competition at a higher price I've decided to switch away from that DAC and persue designs with other chips, beginning with PCM56.
The compression you report isn't something I've noticed in listening but George posted up some linearity measurements of the TDA1387 a while back (over on the lingDAC thread) which leads me to think that the DAC chip itself might be adding a little compression. For competition at a higher price I've decided to switch away from that DAC and persue designs with other chips, beginning with PCM56.
I've probably mentioned it before but we've an I/V stage plus filter coming along soon, that could be used in conjunction with one of Miro's DACs that has +/-1mA output if you take out the opamps. Then you'll discover what the DAC chip's really capable of - it clearly shows how PCM56 beats an array of 100 or more TDA1387 and in turn, PCM56 is beaten by PCM58.
Regarding a PCM56 integrated DAC we're almost up to rev2 which is a paralleled pair of PCM56s, PCB is going out today or tomorrow.
Regarding a PCM56 integrated DAC we're almost up to rev2 which is a paralleled pair of PCM56s, PCB is going out today or tomorrow.
If you end up designing a new dac, I want to make one small suggestion. Please make the screw holes a bit larger so that it is the same size as the screw holes on the PSU components. This way plain old motherboard screws and mounting hardware can be used on it.
Thanks for the info! It’s also good to know my ears are still working properly haha! I probably wouldn’t be able to tell there was compression on lesser gear like my thx 789, always sounds compressed.
Last edited:
You are welcome Ren
Abrax, the nonlinearities I have measured were at the very low levels. I don't think they cause the sense of compression.
Attached is the measurement of a friend's Celibidache DAC. Red line is input/output relationship. Cyan is linearity. Horizontan axis is DAC's digital input level. Vertical axis is DAC's analog out
George
Edit. For the nit pickers, see 2nd attachment. The non linearity at high level is only 0.12db btn 0dBFS and -0.5dBFS input level. IMO, a high achievent!
Abrax, the nonlinearities I have measured were at the very low levels. I don't think they cause the sense of compression.
Attached is the measurement of a friend's Celibidache DAC. Red line is input/output relationship. Cyan is linearity. Horizontan axis is DAC's digital input level. Vertical axis is DAC's analog out
George
Edit. For the nit pickers, see 2nd attachment. The non linearity at high level is only 0.12db btn 0dBFS and -0.5dBFS input level. IMO, a high achievent!
Attachments
Last edited:
Good questions Claude. Firstly - because I only came across PCM58 after having developed a PCM56 DAC. Second, PCM56 is available in SOP16 package whereas PCM58 is only DIL28. Thus the board area required is considerably larger. Third, the PCM58s I have in my possession look to be recycled ones (solder on the leads) - this means I probably can't use a normal DIL socket for them, they'd have to be soldered in directly or go in a ZIF socket. Lastly, yes there is a price issue, the PCM56 is around 3X cheaper.
I am not giving up developing with TDA1387 now I know that PCM56 is better, I think there are DACs for applications and price-points.
My current working hypothesis is that PCM58 sounds better because its lower noise so perhaps paralleling PCM56s will reach PCM58 performance because paralleling improves the SNR. There is another engineering reason (beyond getting lower noise) for running these DACs in parallel and that's because their current output is subject to quite a wide manufacturing tolerance (30%). Paralleling already characterized DACs can help compensate for this output current spread - pair a higher output current one with a lower current one to get closer to the average. When run with an opamp I/V (aka 'voltage output mode') this pairing isn't necessary because the DAC has a compensated on-chip I/V resistor on-chip but with a discrete I/V it becomes an issue.
@gpapag : I agree with you, the levels are very low and I've not heard the compression myself. Thanks for the plots.
I am not giving up developing with TDA1387 now I know that PCM56 is better, I think there are DACs for applications and price-points.
My current working hypothesis is that PCM58 sounds better because its lower noise so perhaps paralleling PCM56s will reach PCM58 performance because paralleling improves the SNR. There is another engineering reason (beyond getting lower noise) for running these DACs in parallel and that's because their current output is subject to quite a wide manufacturing tolerance (30%). Paralleling already characterized DACs can help compensate for this output current spread - pair a higher output current one with a lower current one to get closer to the average. When run with an opamp I/V (aka 'voltage output mode') this pairing isn't necessary because the DAC has a compensated on-chip I/V resistor on-chip but with a discrete I/V it becomes an issue.
@gpapag : I agree with you, the levels are very low and I've not heard the compression myself. Thanks for the plots.
Ren, I wonder how the Ares would sound had it had a much less sized PSU.
I mean the capacity (internal resistance) of the power source plays a role in how a low power circuit evidently sounds, be it a RIAA pre, a line pre, a tuner, a DAC. I have experienced it many times with my diy projects
Abrax DACs all have an onboard PSU circuit which one would say “shields” the DAC from the outboard DC power source’s qualities.
Yes and no.
On my main system it is the opamp-less version of Abbado DAC which plays music. It takes 12Vdc and draws some 140mA. This is hardly 2W. With a 1.2Ah (90mOhm) Pb battery it plays good. With a 7Ah (26mOhm) battery, sound is much better. Smoother and more "bodied". And it becomes a bit more better with two such 7Ah batteries in parallel.
At least now I can claim I have a heavyweight (4kg) DAC
George
Attachments
Nice, you’re making me want to try some tweaks haha. Linear PSU time?Ren, I wonder how the Ares would sound had it had a much less sized PSU.
I mean the capacity (internal resistance) of the power source plays a role in how a low power circuit evidently sounds, be it a RIAA pre, a line pre, a tuner, a DAC. I have experienced it many times with my diy projects
Abrax DACs all have an onboard PSU circuit which one would say “shields” the DAC from the outboard DC power source’s qualities.
Yes and no.
On my main system it is the opamp-less version of Abbado DAC which plays music. It takes 12Vdc and draws some 140mA. This is hardly 2W. With a 1.2Ah (90mOhm) Pb battery it plays good. With a 7Ah (26mOhm) battery, sound is much better. Smoother and more "bodied". And it becomes a bit more better with two such 7Ah batteries in parallel.
At least now I can claim I have a heavyweight (4kg) DAC
George
Hi Jim - currently I'm running at some non-standard voltages, +/-5.3V on my PCM56s. That choice is the lowest legal voltage I could get within the constraint of the references for my voltage regulators being IR LEDs for lowest noise. Each LED is around 1.2V, there's a string of 5.
No I haven't played with the DAC's supply voltage, my focus has been more on the I/V stage than the DAC itself, other than giving it the lowest possible noise supplies.
No I haven't played with the DAC's supply voltage, my focus has been more on the I/V stage than the DAC itself, other than giving it the lowest possible noise supplies.
