Here's what happens on the sample level when the amplitude goes down with the DAM1021 playing back a 11025 Hz sinewave.
The recording very much looks like a sinewave, I wonder why the oscillator plugin reads this as a triangle wave (top clipped triangle at the amplitude low point).
The recording very much looks like a sinewave, I wonder why the oscillator plugin reads this as a triangle wave (top clipped triangle at the amplitude low point).
Attachments
Regarding sample points phase for a particular sinewave, it depends on how the sine wave was generated. If the generator puts the sample points in a particular phase, and if they are not later phase shifted in a filter or something, then then they should stay the same as when they were generated.
Here's an alternative to sniffing the input:
From my recollection the Si interface is well documented. You could even mock it -- pretend that you're an Si clock when really you're not -- then use the commands sent to select the right clock.
At least this would work on the dam1121 which has external clock I/O. Unlike the dam1021 I think?
Listening impressions
After a few days of listening I would share my impressions about the DAM1021 sound quality.
Of course they are my subjective impressions (like any other opinion posted on this thread) based on the comparison between the DAM1021 and the other DACs in my availability I have listed in a previous post.
My absolute reference to compare all the DAC is acoustic live music, mostly classical and jazz.
Firstly the conclusions so I can immediately collect a new series of insults.
The DAM1021 doesn't sound so bad, it sounds better than I expected.
It's probably worth the money you have to pay, however it is far from a state-of-the-art DAC. Even though I've never listened to TotalDac and MSB Tech DAC, I don't think the DAM1021 is in the same sound quality region, unless my Naim Audio CD3 and my DAC AD1865 were better than them.
I do not think so.
And now the detailed impressions of my listening session.
Source dependence
After a careful comparison between the spdif source (AUNE X5S) and the i2s source (RPI) feeding the DAM I can confirm the DAM1021 sound quality is source dependent.
With x 44.1 kHz material the performance is very similar for both sources, while with x48 kHz sample rate family the I2S source performs better.
IMHO the explanation is quite simple and is related to timing: the Raspberry PI 4 I'm using for I2S output uses a 54 MHz oscillator to generate audio samples.
This means it can operate a correct division with x48 sample rate family (for example 54MHz / 48 kHz = 1125), while it has to approximate when the sample rate family is x44.1 kHz (54MHz / 44.1 kHz = 1224.4897959....).
Finally the I2S output from the Raspberry is bit perfect but in case of x44.1 kHz sample rate family the timing is alterated by its PLL to get a correct division (the time is alternately shortened or lengthened).
The DAM1021 tracks the input with its PLL, this closes the circle.
Many users have reported this issue so they will have a better idea of the causes than mine.
Maybe the phase noise measurement will tell.
Raw output vs buffered output
Although there is not a big difference I prefer the raw output over the buffered output.
To my taste the raw output sounds more homogeneous while the buffered output seems to be affected by some artifacts which make it a little more angular (I don't know if the word "angular" can give an idea of what I mean).
It might sound better with a different buffer, I have not experimented on this.
Bass region
Compared to the other DACs the bass of the DAM1021 are a little booming and less deep and controlled.
On the whole the sound remains homogeneous however there is a certain intrusiveness in the mid-lower part of the spectrum, I believe in the 80-160 Hz region, which makes the sound less natural. Difficult to say whether it depends on an excess in the medium-low part or a deficiency in the deepest part.
Listening to the grand piano there seems to be a shortage in the deepest part of the audio spectrum, however it could also be due to a too slight midrange compared to the low range.
The CD3 has deeper bass and better aligned to the rest of the audio spectrum.
The AD1865 DAC stands in the middle, a little more redundant in the bass region but not as the DAM1021.
High region
The higher part of the spectrum is enough detailed with the DAM1021 although with complex music there seems to be some confusion between the details of the various instruments.
Listening to the orchestra this confusion in details increases and the whole sound becomes a little harsh.
With Hi-Res material this effect is a little less pronounced and the sound is less harsh, maybe because the sound becomes a little darker and closed in the higher part of the audio spectrum, with a few loss in detail.
The high region of the Naim CD3 is always well balanced and full of details and never becomes harsh, maybe a little closed but overall better than the DAM1021.
Again the AD1865 DAC stands in the middle, less detailed but never harsh.
Mid region
Here the difference between the DACs is more clear and audible.
The DAM1021 always seems to suffer from some sort of graininess in the midrange.
This effect is clearly audible listening to piano and voices.
The grand piano becomes more woody. Although the characteristic Steinway sound is still recognized with the DAM1021 it's partly clouded by this sort of graininess.
The spectacular voice of Mary Black is as veiled, losing the typical warmth and fullness that distinguish it.
Same situation with the strings, the sound of violins and cellos becomes harder and more anonymous, losing the harmonics that make it silky.
Conversely the Naim CD3 has spectacular midrange (thanks to the TDA1541A), there is no grain, the voices are warm and clear without any haze.
The Steinway grand piano comes out in all its magnificence with the full and round sound that distinguishes it.
The midrange of the AD1865 DAC is a little less spectacular in respect of the CD3 (the TDA makes the difference), but it is still more enjoyable than the DAM1021 and never becomes tiring.
The DAM1021 performs a little better with Hi-Res music but there is still a fair difference from the other two DACs that play redbooks.
With the dam1021 the fender lap steel guitar performed by Steve Howe seems to drill your brain, you get tired of listening immediately.
So the DAM1021 is not only source-dependent , IMHO is also music-dependent. Depending on the music you listen to the performances are very different, more balanced and pleasant with simple music, more lacking with complex music such as choirs and orchestra.
Ambience
This is another weakness of the DAM1021 compared to the other DACs under test.
The soundstage coming out from the DAM1021 is narrower and less deep than the other DACs.
To my taste the soundstage of the DAM1021 is a little compressed compared to reality. Although in many cases the voice or the solo instrument is just catapulted out of the stage there is still a certain compression of the space in all directions.
This feeling that is perceived by listening to the DAM1021 is very pronounced in the case of the orchestra. The orchestra appears to be confined within the space between the speakers and lacks depth. With small instruments group the sensation is less perceivable but still exists.
The Naim CD3 and the AD1865 DAC perform much better, the soundstage is enough correct with the right width and depth. The orchestra opens well beyond the speakers.
Is this related to THD? No, I know where this issue come from.
Keep in mind from the physical point of view space, time and speed of sound are strictly correlated.
Dynamic
The DAM1021 has good enough dynamic performance although it suffers a little playing complex music like full orchestra.
With the full orchestral sound the dynamic seem a little compressed but I suspect that it is not due to a lack of real dynamics but rather to a midrange too monochord and a little confused.
Overall, however, the dynamic contrast of the DAM1021 is almost comparable to that of the other two DACs under test.
The DAC inside the Aune X5S source is the worst performer of the group, it performs much worse than the DAM1021. So it does not worth to spend words about its sound.
In the end, to give an immediate idea of my listening impressions during the comparison of the DACs under test I would assign to each a numerical value that represents them over a scale with a range from 1 to 10:
Naim Audio CD3: 7.5
AD1865 Diy DAC: 7
DAM1021: 6 to 6.5
AUNE DSD1793: 5
The value attributed to the DAM1021 varies between 6 and 6.5 because as I said previously it's even music-dependent, so with some songs it sounds better while with others it sounds a little worse.
Remark: when playing HiRes music via spdif sometimes the DAM1021 loses the lock.
Next steps:
- phase noise measurement (as from the factory)
- ladder accuracy measurement (as from the factory)
- upgrade with a true FIFO buffer and DAC calibration
- phase noise measurement (after upgrade)
- ladder accuracy measurement (after upgrade)
- blind listening session with comparison against the TDA1541A reference system (both version: from the factory and upgraded)
After a few days of listening I would share my impressions about the DAM1021 sound quality.
Of course they are my subjective impressions (like any other opinion posted on this thread) based on the comparison between the DAM1021 and the other DACs in my availability I have listed in a previous post.
My absolute reference to compare all the DAC is acoustic live music, mostly classical and jazz.
Firstly the conclusions so I can immediately collect a new series of insults.
The DAM1021 doesn't sound so bad, it sounds better than I expected.
It's probably worth the money you have to pay, however it is far from a state-of-the-art DAC. Even though I've never listened to TotalDac and MSB Tech DAC, I don't think the DAM1021 is in the same sound quality region, unless my Naim Audio CD3 and my DAC AD1865 were better than them.
I do not think so.
And now the detailed impressions of my listening session.
Source dependence
After a careful comparison between the spdif source (AUNE X5S) and the i2s source (RPI) feeding the DAM I can confirm the DAM1021 sound quality is source dependent.
With x 44.1 kHz material the performance is very similar for both sources, while with x48 kHz sample rate family the I2S source performs better.
IMHO the explanation is quite simple and is related to timing: the Raspberry PI 4 I'm using for I2S output uses a 54 MHz oscillator to generate audio samples.
This means it can operate a correct division with x48 sample rate family (for example 54MHz / 48 kHz = 1125), while it has to approximate when the sample rate family is x44.1 kHz (54MHz / 44.1 kHz = 1224.4897959....).
Finally the I2S output from the Raspberry is bit perfect but in case of x44.1 kHz sample rate family the timing is alterated by its PLL to get a correct division (the time is alternately shortened or lengthened).
The DAM1021 tracks the input with its PLL, this closes the circle.
Many users have reported this issue so they will have a better idea of the causes than mine.
Maybe the phase noise measurement will tell.
Raw output vs buffered output
Although there is not a big difference I prefer the raw output over the buffered output.
To my taste the raw output sounds more homogeneous while the buffered output seems to be affected by some artifacts which make it a little more angular (I don't know if the word "angular" can give an idea of what I mean).
It might sound better with a different buffer, I have not experimented on this.
Bass region
Compared to the other DACs the bass of the DAM1021 are a little booming and less deep and controlled.
On the whole the sound remains homogeneous however there is a certain intrusiveness in the mid-lower part of the spectrum, I believe in the 80-160 Hz region, which makes the sound less natural. Difficult to say whether it depends on an excess in the medium-low part or a deficiency in the deepest part.
Listening to the grand piano there seems to be a shortage in the deepest part of the audio spectrum, however it could also be due to a too slight midrange compared to the low range.
The CD3 has deeper bass and better aligned to the rest of the audio spectrum.
The AD1865 DAC stands in the middle, a little more redundant in the bass region but not as the DAM1021.
High region
The higher part of the spectrum is enough detailed with the DAM1021 although with complex music there seems to be some confusion between the details of the various instruments.
Listening to the orchestra this confusion in details increases and the whole sound becomes a little harsh.
With Hi-Res material this effect is a little less pronounced and the sound is less harsh, maybe because the sound becomes a little darker and closed in the higher part of the audio spectrum, with a few loss in detail.
The high region of the Naim CD3 is always well balanced and full of details and never becomes harsh, maybe a little closed but overall better than the DAM1021.
Again the AD1865 DAC stands in the middle, less detailed but never harsh.
Mid region
Here the difference between the DACs is more clear and audible.
The DAM1021 always seems to suffer from some sort of graininess in the midrange.
This effect is clearly audible listening to piano and voices.
The grand piano becomes more woody. Although the characteristic Steinway sound is still recognized with the DAM1021 it's partly clouded by this sort of graininess.
The spectacular voice of Mary Black is as veiled, losing the typical warmth and fullness that distinguish it.
Same situation with the strings, the sound of violins and cellos becomes harder and more anonymous, losing the harmonics that make it silky.
Conversely the Naim CD3 has spectacular midrange (thanks to the TDA1541A), there is no grain, the voices are warm and clear without any haze.
The Steinway grand piano comes out in all its magnificence with the full and round sound that distinguishes it.
The midrange of the AD1865 DAC is a little less spectacular in respect of the CD3 (the TDA makes the difference), but it is still more enjoyable than the DAM1021 and never becomes tiring.
The DAM1021 performs a little better with Hi-Res music but there is still a fair difference from the other two DACs that play redbooks.
With the dam1021 the fender lap steel guitar performed by Steve Howe seems to drill your brain, you get tired of listening immediately.
So the DAM1021 is not only source-dependent , IMHO is also music-dependent. Depending on the music you listen to the performances are very different, more balanced and pleasant with simple music, more lacking with complex music such as choirs and orchestra.
Ambience
This is another weakness of the DAM1021 compared to the other DACs under test.
The soundstage coming out from the DAM1021 is narrower and less deep than the other DACs.
To my taste the soundstage of the DAM1021 is a little compressed compared to reality. Although in many cases the voice or the solo instrument is just catapulted out of the stage there is still a certain compression of the space in all directions.
This feeling that is perceived by listening to the DAM1021 is very pronounced in the case of the orchestra. The orchestra appears to be confined within the space between the speakers and lacks depth. With small instruments group the sensation is less perceivable but still exists.
The Naim CD3 and the AD1865 DAC perform much better, the soundstage is enough correct with the right width and depth. The orchestra opens well beyond the speakers.
Is this related to THD? No, I know where this issue come from.
Keep in mind from the physical point of view space, time and speed of sound are strictly correlated.
Dynamic
The DAM1021 has good enough dynamic performance although it suffers a little playing complex music like full orchestra.
With the full orchestral sound the dynamic seem a little compressed but I suspect that it is not due to a lack of real dynamics but rather to a midrange too monochord and a little confused.
Overall, however, the dynamic contrast of the DAM1021 is almost comparable to that of the other two DACs under test.
The DAC inside the Aune X5S source is the worst performer of the group, it performs much worse than the DAM1021. So it does not worth to spend words about its sound.
In the end, to give an immediate idea of my listening impressions during the comparison of the DACs under test I would assign to each a numerical value that represents them over a scale with a range from 1 to 10:
Naim Audio CD3: 7.5
AD1865 Diy DAC: 7
DAM1021: 6 to 6.5
AUNE DSD1793: 5
The value attributed to the DAM1021 varies between 6 and 6.5 because as I said previously it's even music-dependent, so with some songs it sounds better while with others it sounds a little worse.
Remark: when playing HiRes music via spdif sometimes the DAM1021 loses the lock.
Next steps:
- phase noise measurement (as from the factory)
- ladder accuracy measurement (as from the factory)
- upgrade with a true FIFO buffer and DAC calibration
- phase noise measurement (after upgrade)
- ladder accuracy measurement (after upgrade)
- blind listening session with comparison against the TDA1541A reference system (both version: from the factory and upgraded)
Thanks for your extensive report.
I follow the logic behind your source-dependance. I assume that you have a RPi 4. For earlier versions the same might apply. They run 19,2 MHz clocks. For 48 kHz that's a factor 400 and for 44,1 kHz it's not divisible.
Which makes me wonder if it would bring any improvement to resample on the RPi before sending it over I2S. (I'm a non-oversampling kind of guy so that doesn't sit quite right with me... or is this just zero padding?)
Second to get rid of that ~100 Hz veil, be sure to bake or install some filters without the DC blocking IIR.
What was the filter setting you used? All of the stock ones have DC blocking on them I believe, so no difference there but rather to complete your report.
Third, I know there are quite a bit of power supply mods on this thread. Myself I can't comment as I'm using a dam1121, but this may help in the jitter and THD departments.
Finally and I cannot stress this enough, do install other filters as part of your next step. Again the differences are clearly audible -- in a good way -- and there's a lot to choose from, whether you're an OS or NOS kind of guy.
I follow the logic behind your source-dependance. I assume that you have a RPi 4. For earlier versions the same might apply. They run 19,2 MHz clocks. For 48 kHz that's a factor 400 and for 44,1 kHz it's not divisible.
Which makes me wonder if it would bring any improvement to resample on the RPi before sending it over I2S. (I'm a non-oversampling kind of guy so that doesn't sit quite right with me... or is this just zero padding?)
Second to get rid of that ~100 Hz veil, be sure to bake or install some filters without the DC blocking IIR.
What was the filter setting you used? All of the stock ones have DC blocking on them I believe, so no difference there but rather to complete your report.
Third, I know there are quite a bit of power supply mods on this thread. Myself I can't comment as I'm using a dam1121, but this may help in the jitter and THD departments.
Finally and I cannot stress this enough, do install other filters as part of your next step. Again the differences are clearly audible -- in a good way -- and there's a lot to choose from, whether you're an OS or NOS kind of guy.
The Total Dac is pointed out at the beginning of this thread as one of the reference DAC, I have never heard it so I cannot say anything about it.
BTW, what is wrong in the Total Dac measurements?
Phase noise measurement
I'm measuring the phase noise of the most crucial signal for the DAM1021, the LRCK (both input spdif and I2S, both sample rate families x44.1 and x48 kHz).
I cannot measure the MLCK for 2 reasons:
- it runs at 45/49 MHz so it's out of the TimePod measurement range
- the MCK output is LVDS so the voltage it's too low to be loaded with the 50 ohm input of the TimePod
I'm measuring the phase noise of the most crucial signal for the DAM1021, the LRCK (both input spdif and I2S, both sample rate families x44.1 and x48 kHz).
I cannot measure the MLCK for 2 reasons:
- it runs at 45/49 MHz so it's out of the TimePod measurement range
- the MCK output is LVDS so the voltage it's too low to be loaded with the 50 ohm input of the TimePod
Attachments
All NOS-Dacs show similar measurements - if they are working really without digital-filters (= no upsampling, no oversampling): Link
Please, read carefully:
"I don't think the DAM1021 is in the same sound quality region, unless my Naim Audio CD3 and my DAC AD1865 were better than them."
Ah Ok, the disaster is the THD.
Sorry, I didn't think about it because I usually listen to music with tube amps with 2% THD.
Not just the THD which could be benign. The FIR overshoots and is flawed. There's tons and tons of ultrasonic tones. spurious tones in IMD. multi-tone shows less than 16-bit noise resolution. ...
So unless you have some ultrasonic fetish this is not really acceptable performance. Perhaps the analog parts are all right but the digital design, filter and overall implementation seems very sucky indeed. The soekris 2541 linked earlier runs circles around it.
My DAM1021 has the filter coming from the factory.
As I said my PC doesn't provide RS232 so I cannot install different filters.
Il will try later.
Well, it wasn't me who mentioned the Total Dac as a reference.
Please, ask Soeren to remove it from the list of the reference DACs.