I have 32 polymer caps on my dual mono setup lol. Suppose your DAC/ADC recording setup is reliable, it's still not entirely clear whether a "smeared" sound is necessarily bad or less accurate. It seems to me that a "digital" sound is probably less "smeared" than "analog", but people don't always see it as a plus. One thing I know for sure is that my first 1021 build sounded very sharp but caused actual headaches due to listening fatigue. I would agree that live music is usually sharper and clearer, but perceived sharpness is not always better.
Wouldn't the soundstage sound "nervous", if the idea of out-of-sync clocks makes you "nervous"?
I didn't even think about that when I built it. I assumed that it should sound better, but after a few listening comparisons it was clear to us that the 2 clocks must be the cause. Actually logical, on the one hand you are hunting for even better jitter values and here two clocks run independently of each other.
I had a Lvds splitter and used the Spdif input circuit described in the instructions. Source was a LUMIN U1, speakers and amps from Tidal.
In my build the caps were solving the wrong problem the wrong way. It's essential that the grounding and general PSU quality are taken care of that surround the DAM1021.
I feel no need now to add any caps to the DAM board now though, high quality recordings sound great and have all the bass of the original recording in my build. This is confirmed by the DAAD loops, which make all the changes reproducible to me.
I use a +/-15VDC switch mode PSU which then travels through an LC filter, rectifier, big battery of parallel low ESR electrolytics and linear regulators lowering the voltage to +/- 7,3VDC.
The analog output board gets the +/-15VDC filtered by an LC filter followed by an RC filter.
Chassis connection for the audio ground of both the DAM board and the analog outboard is right at pin1.
More caps in parallel to the +/-4VDC rails may actually generate resonances, which you can only find with proper measuring equipment. Otherwise it is really all just guesswork.
I've conducted more experiments with different AES/EBU sources and changes to the digital connection (balanced/pin 3 grounded, shielded/floating, added or removed filtering caps etc.).
The essential thing: The DAM1021 is very sensitive to the minutiae of the incoming digital signal and all of these changes matter.
I've never been able to see jitter sidebands at a quarter of the samplingrate though. Either the DAM syncs and it works, or it doesn't. But the differences are very audible (focus, clarity, bass punch and extension, high end agressiveness , ...).
Two things are strange:
1. Drift
The DAM1021 drifts badly, though at a sub-sample (at 44.1khz) level. It makes the mixing of inverted files to gauge their difference impossible, resulting in flanging. Even DeltaWave (null comparator software) is unable to properly correct for it.
I don't have that problem at all with any other DAC.
2. Fed with a sinewave of 11025hz at 44.1khz sampling frequency (to check for jitter sidebands) the amplitude of the output changes to a maximum of 3dB below the rest of the (unfiltered) frequency range. It also drifts around. This is recording with the AD converter that clocks the sound card which in turn clocks the DAM1021.
This doesn't happen with other DACs either.
What's going on here and how can we fix it?
The essential thing: The DAM1021 is very sensitive to the minutiae of the incoming digital signal and all of these changes matter.
I've never been able to see jitter sidebands at a quarter of the samplingrate though. Either the DAM syncs and it works, or it doesn't. But the differences are very audible (focus, clarity, bass punch and extension, high end agressiveness , ...).
Two things are strange:
1. Drift
The DAM1021 drifts badly, though at a sub-sample (at 44.1khz) level. It makes the mixing of inverted files to gauge their difference impossible, resulting in flanging. Even DeltaWave (null comparator software) is unable to properly correct for it.
I don't have that problem at all with any other DAC.
2. Fed with a sinewave of 11025hz at 44.1khz sampling frequency (to check for jitter sidebands) the amplitude of the output changes to a maximum of 3dB below the rest of the (unfiltered) frequency range. It also drifts around. This is recording with the AD converter that clocks the sound card which in turn clocks the DAM1021.
This doesn't happen with other DACs either.
What's going on here and how can we fix it?
Well, I hope that the revised timing handling that Sören spoke of recently will see to that the buffer/PLL starts to really work as a barrier towards the outer world. Your findings are interesting and adds to the indications that there are room for improvements here.
Again - I would like to listen to these DACs with the clocks left untouched for many seconds if not a minutes - the clocks that that drives the I2S and s/pdif are so good nowadays that once the DAM PLL has picked up the center frequency of the incoming clock, a quite small buffer should suffice. (...and please no re-sync after every track!! - only after lock)
And by all means - make a movie mode that tracks the incoming clock better to reduce delay but make room for the full potential of the design by all means!!
Looking forward to the next (soon!?) FW update
//
Again - I would like to listen to these DACs with the clocks left untouched for many seconds if not a minutes - the clocks that that drives the I2S and s/pdif are so good nowadays that once the DAM PLL has picked up the center frequency of the incoming clock, a quite small buffer should suffice. (...and please no re-sync after every track!! - only after lock)
And by all means - make a movie mode that tracks the incoming clock better to reduce delay but make room for the full potential of the design by all means!!
Looking forward to the next (soon!?) FW update
//
This was the conclusion some modders came to, including myself, which resulted in adding a potato flip-flop between my USB/I2S converter and the 1021 even though it's not supposed to make a difference. It is audible.
removing the isolators and adding the flip-flop elevated the sound quality substantially with my board.
idk. it was just circuits wired up on a wooden board. I used a Chinese knockoff of the sigma 22 PSU, and just listened to the buffered output. Didn't really bother to wait for things to heat up either. Accidentally fried the mosfets when i put it in an enclosure, so that PSU is no more. Haven't felt anything strange since
it's also not sharp as in off tone. Just fatiguing
well f- me. I don't have any proper equipment for this. I always thought that the theories supported the addition of polymer caps specifically. It's in fact the best known mod on this thread. Anyway, i'm not sure if I trust my ears anymore even if I do have DAAD to setup an ABX. Were you able to pass ABX on the recordings, or just AB? Not saying that there is definitely no audible difference if you can't pass ABX
f- me, really?...
I don't want to be presumptuous, but there have been many times that I was absolutely sure there was a difference in AB but can't pass ABX when it comes down to it. The mind is not a simple waveform analyzer
Maybe you're just not feeding your dam1021 right, it can get pretty upset when you do that. It's important that you spend the time to talk to it and understand what it actually likes. My dam1021 has been pretty happy with me.
If you do find out why, let me know
Last edited:
I don't think it is an issue of the digital connection. It is an issue on how the Fifo of the DAM works. I had similar experience with using always the same connection, see
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz - Page 734 - diyAudio
post #7334
My explanation is that the DAM memorizes its last internal clock frequency (the 22.57..MHz + adjustment) for a given sample rate family (say 44.1kHz).
If the incoming frequency matches, by coincidence, the stored frequency the DAM clock will stay stable.
If the incoming frequency has some offset (but is stable) the DAM will adjust its own frequency in the attempt to match the incoming frequency. However there adjustment will result have some overshot ... it adjusts in the other direction and enters in an undamped oscillation of small amplitude.
I was able to reach a stable state by changing the sample rate forth and back when the oscillation was at its crossing point, so that the exact frequency of the incoming signal was "stored". This is of cause only a "solution" for experiments.
@ living sounds
I think you are aiming at the wrong solution of your problems.
As you work with recorded data the "right" solution would be asynchronous data transfer and one clock running at fixed frequency, the clock for the DAC(s). So a setup as it is realized (for 2 channels) in the dam1941 with USB connection.
I see however that for 32 channels the USB-Audio-Class-2 approach has not enough bandwidth. So you would need, as far as I see it, to build your own digital interface including drivers (and then use multiple dam1121 with a common external clock). Not an easy task but as you use it professionally perhaps worth the effort.
P.S. and it would be best then to disable the Fifo and run the DAM1121 at fixed frequency. Perhaps you can talk Soeren in providing you with that, as he has it essentially already for the 1941.
I think you are aiming at the wrong solution of your problems.
As you work with recorded data the "right" solution would be asynchronous data transfer and one clock running at fixed frequency, the clock for the DAC(s). So a setup as it is realized (for 2 channels) in the dam1941 with USB connection.
I see however that for 32 channels the USB-Audio-Class-2 approach has not enough bandwidth. So you would need, as far as I see it, to build your own digital interface including drivers (and then use multiple dam1121 with a common external clock). Not an easy task but as you use it professionally perhaps worth the effort.
P.S. and it would be best then to disable the Fifo and run the DAM1121 at fixed frequency. Perhaps you can talk Soeren in providing you with that, as he has it essentially already for the 1941.
Last edited:
Thanks alot! This looks exactly like my problem. I was able to hit a stable status yesterday once and have a recording to prove it. I then switched off the DAM1021 and on again, and it oscillated wildly again.
You can easily see if the DAM1021's clock is stable by looking at the amplitude of a sinewave one quarter of the samplerate (11025hz at 44.1khz samplerate). If its amplitude matches the rest of the frequency range below it and isn't modulated (to a minimum of 2-3 db below the amplitude of other frequencies) the clock is stable.
The fact that the DMA1021 is actually capable of syncing perfectly (as it sometimes does) indicates that this behaviour should be considered a bug in the DAM1021's firmware and needs to be fixed.
Soeren, is there a way for you to change this? Thanks!
When I ordered the DAM1021 I had gotten the reply that all parallel units would behave the same, so if they received the same clock they would be in sync. Now I wonder if this would actually be the case.
USB is out of the question, I will not use anything other than RME PCIe cards in my studio.
I'll repost my findings form a i2c trace I received form a fellow forum member...
The fastest update seem indeed to be 0,1 Hz as Sören stated. 400 Hz came from somewhere else...??
Have to admit that now when I'm looking at these I cant really say what the right-most one shows ...
//
The fastest update seem indeed to be 0,1 Hz as Sören stated. 400 Hz came from somewhere else...??
Have to admit that now when I'm looking at these I cant really say what the right-most one shows
...//
Attachments
Last edited: