I've tried plenty of more modern ones and also discrete op amps and always went back to it. I talked to the manufacturer of my current large-format-console about using a "better on paper" op amp in the summing position of my console (most important chip in the entire thing), and he agreed that, while not the best on paper, the 5534 sounded better there than the newer and better speced ones. 5532/5534 is what he uses in all of his high end gear, that plenty of studio people buy and use.
The preoccupation with a single metric, harmonic distortion, is miguided IMO. This DAC, which has very high THD by modern standards, is a good indicator that there are other things to worry about.
And implimentation is absolutely crucial. With the 1602 based output stage on the DAM1021 with uncompensated feedback paths I wonder about borderline instabilities that might be responsible for the diminished sound...
Thanks for the schematic, Soeren! The OPA1656 is one I haven't tried, it looks very good indeed on paper.
I'm not sure what you mean about uncompensated feedback path, the 22 pF capacitors in your schematic don't have anything to do with compensation....
The opa1602 is supposed to be good... But there are many opinions....
The new opa1656 is also supposed to be really good, I like the fact that it's CMOS, it's amazing how far pure CMOS has come, you should really try it, although you might need an adapter PCB as it's dual and SMD only....
OPA1656: High-Performance CMOS Audio Op Amp
For a new diy project, the dam1941 appears as the only viable option for a balanced output dac, driven directly by the resistor network.
However, I am confused by the recent product line prerelease of the dac2541. As it is mentioned, this is NOT a DIY product. If this meant to indicate that dam1941 will not be outphased and retired thats great. However, the obvious overlap remain as the source of confusion. A consequence in my view is that OEMs are left behind.
A clarification or, better still a positioning of the diy & OEM targeted offering would help.
However, I am confused by the recent product line prerelease of the dac2541. As it is mentioned, this is NOT a DIY product. If this meant to indicate that dam1941 will not be outphased and retired thats great. However, the obvious overlap remain as the source of confusion. A consequence in my view is that OEMs are left behind.
A clarification or, better still a positioning of the diy & OEM targeted offering would help.
Last edited:
No, I just had a little too many dam1941 in stock, price is back to normal again.... At some point all products will be updated as stock runs out, but the dam1941 is not first, next will be the other in the dac series, then the dam1021...
Vital upgrade to the output stage: Resistors to pull the op amps into Class A.
In theory this results in a small reduction in measureable distortion, far below THD of the DAM1021.
But in praxis this gets rid of any trace of harshness and makes for even better dynamics.
My DAM1021 now officially eats for breakfast every other DAC I have ever heard.
Thanks Soeren for the DAC and TNT for the filter.
In theory this results in a small reduction in measureable distortion, far below THD of the DAM1021.
But in praxis this gets rid of any trace of harshness and makes for even better dynamics.
My DAM1021 now officially eats for breakfast every other DAC I have ever heard.
Thanks Soeren for the DAC and TNT for the filter.
Attachments
I tried to force the connection by Pin J3-3 to ground and ... nothing! The led continues to flash and has no signal.
Possible help to identify the problem?
I've conducted more tests with the DAM1021:
1.
On my original Rev 4 board I had added 12 1.800uf Panasonic FR low ESR caps to VREF, otherwise it is stock.
Replacing this board with another unmodded Rev 4 board running the same firmware and filters I noticed it lacking some energy/dynamics in the bass, while at the same time displaying a cleaner high end, closer to the original source (I made recordings for comparison with a discrete R2R AD converter). The perceived thinning of the bass cannot be corrected for with EQ.
Did anyone else notice this (additional VREF capacitance audibly worsening performance in the high end)? Maybe it is necessary to use better quality caps?
2.
I'm now firmly in the camp of zero aliasing linear filters. Since both my ADC and the DAC allow me to adjust filters, my experiments showed the technically accurate filters not only sounded more realistic, but also proved the differential signal to be much lower (= better correlation) and with less pronounced artifacts.
1.
On my original Rev 4 board I had added 12 1.800uf Panasonic FR low ESR caps to VREF, otherwise it is stock.
Replacing this board with another unmodded Rev 4 board running the same firmware and filters I noticed it lacking some energy/dynamics in the bass, while at the same time displaying a cleaner high end, closer to the original source (I made recordings for comparison with a discrete R2R AD converter). The perceived thinning of the bass cannot be corrected for with EQ.
Did anyone else notice this (additional VREF capacitance audibly worsening performance in the high end)? Maybe it is necessary to use better quality caps?
2.
I'm now firmly in the camp of zero aliasing linear filters. Since both my ADC and the DAC allow me to adjust filters, my experiments showed the technically accurate filters not only sounded more realistic, but also proved the differential signal to be much lower (= better correlation) and with less pronounced artifacts.
3.
The DAM1021 is changing its sound with different incoming clocks. In the case of my system, it is clocked from an RME HDSP9652. While the RME card does reclock by itself, I have noticed with other DACs that a superior external clock provides a more focused image even travelling through the reclocking RME card. The same is true with the DAM1021. Though reclocking happens twice in the chain, there is still an audible difference. Strange especially, since in both cases the clock signal is transported together with the digital audio signal via transformer balanced AES/EBU / S/PDIF lines.
The DAM1021 is changing its sound with different incoming clocks. In the case of my system, it is clocked from an RME HDSP9652. While the RME card does reclock by itself, I have noticed with other DACs that a superior external clock provides a more focused image even travelling through the reclocking RME card. The same is true with the DAM1021. Though reclocking happens twice in the chain, there is still an audible difference. Strange especially, since in both cases the clock signal is transported together with the digital audio signal via transformer balanced AES/EBU / S/PDIF lines.
Hi!
I makes me happy that the scientific route is successful - been done right, it should be. I'm also longing for an A/D that I can provide filters with a safe guard to >=fs/2 input. There is one in the making here - soon hopefully.
Yes, the PLL is not effective, we have seen the behaviour by eavesdropping on the I2C lines to the clock. It changes a lot due to that Sören priorities movie watching instead of superior sound which is really surprising given the initially stated goal and vision for the product. I hope he releases a new FW where there is a switch between movie and ludicrous mode.
I also hear a difference when I pass the output from a Squeezbox Touch via an Ian buffer with a quite good NDK (selected!) clock in it.
//
I makes me happy that the scientific route is successful - been done right, it should be. I'm also longing for an A/D that I can provide filters with a safe guard to >=fs/2 input. There is one in the making here - soon hopefully.
Yes, the PLL is not effective, we have seen the behaviour by eavesdropping on the I2C lines to the clock. It changes a lot due to that Sören priorities movie watching instead of superior sound which is really surprising given the initially stated goal and vision for the product. I hope he releases a new FW where there is a switch between movie and ludicrous
mode.I also hear a difference when I pass the output from a Squeezbox Touch via an Ian buffer with a quite good NDK (selected!) clock in it.
//
I can second your findings. I added 12000uF of polymer caps to Vref of my 1121, and while in this case the bass improved, it became steely at the top end in a way that I preferred my cheap SD player to it, because it had a much more natural tone.
Lately I got it off the shelf gain, out went the Polymer and in just 4 470uf Elna Silmic, s o less capacitance than before. Mucho better, the harshness went away almost completely, and there was no loss in bass either. So obviously it is not so much the amount of capacitance, but the quality.
Still I was not convinced, because it still lacked musicality and timing.
Then a major step was the power supply of the USB board I use, the Waveio. Just added a simple 5V Lt3045 reg, and the sun started to shine. Now it is really good, super resolution, and faithful to the sound of instruments without being harsh or compressed as it was before.
So also the on board reclocking is not doing miracles, and it matters very much how you feed this dac. It actually is super sensitive to any change in the chain as it seems.
So happy that it's DIY and I could fix the problem, so I can finally enjoy this otherwise fine piece of engineering.
Well, fine apart from some things I would do different...
Lately I got it off the shelf gain, out went the Polymer and in just 4 470uf Elna Silmic, s o less capacitance than before. Mucho better, the harshness went away almost completely, and there was no loss in bass either. So obviously it is not so much the amount of capacitance, but the quality.
Still I was not convinced, because it still lacked musicality and timing.
Then a major step was the power supply of the USB board I use, the Waveio. Just added a simple 5V Lt3045 reg, and the sun started to shine. Now it is really good, super resolution, and faithful to the sound of instruments without being harsh or compressed as it was before.
So also the on board reclocking is not doing miracles, and it matters very much how you feed this dac. It actually is super sensitive to any change in the chain as it seems.
So happy that it's DIY and I could fix the problem, so I can finally enjoy this otherwise fine piece of engineering.
Well, fine apart from some things I would do different...
I run my 1121s unmodifyed. When I got these and compared to the 1021 it was so immediately obvious that the 1121 did something very right compared to my moded 1021. But as I recall it, I did like the modded 1021 better overall than the original (first rev). But maybe it was the top end...
//
//
I firmly believe there is an optimal amount of capacitance for every component in a system where the best balance of deep bass, dynamics, speed and agility and high frequency extension is reached. It is always about choosing the compromise which works best. In my 1021 i chose the "transistor mod" and only a small amount of additional capacitance in cleaning up Vref.
If you read about Peter Daniels improvements of his 1941 you will notice he only uses minimal but high quality capacitance.
Could you please describe the transistor mod and what and how much capacitance you used in detail?
I read about the transistor mod and saw pictures before, but it wasn't really clear what had been done.
I now added only 4 1.800uf caps on top of the capacitors on the side. It is good design practice to have the bigger electrolytic caps close to the PSU and the smaller ceramic caps close to the component that is supplied. There might also be less dirt dumped into the audio ground spilling into the signal this way...
The anemic sound of the stock board is completely gone and listening to music is fun again. Bass is full and engaging and overall sound is organic and relaxed. But there is also some high end sparkle back again that shouldn't be there.
The anemic sound of the stock board is completely gone and listening to music is fun again. Bass is full and engaging and overall sound is organic and relaxed. But there is also some high end sparkle back again that shouldn't be there.
Last edited:
Yes, OSVA. Well, it seem to have good performance. Good that you are on the train there
To have a rec/play chain with no alias/mirroring will I think be a revelation :-D
The whole party has been around BW up till now - crazy... if you have heard a good FM stereo direct live transmission.
//
To have a rec/play chain with no alias/mirroring will I think be a revelation :-D
The whole party has been around BW up till now - crazy... if you have heard a good FM stereo direct live transmission.
//
Last edited: