Soekris DAM1021 R2R DAC Measurements

[rebbiputzmaker]

I think the rabbi is working in the repetitive/last-man-standing argument tradition. It doesn't cut it here, fortunately.

If double blind is the listening standard, at least something should be attempted to go along with the measurements.

 

[rebbiputzmaker]

And you have done this to back up all of your subjective claims here?

Actually yes, many times, for many years.

 

[amirm]

Actually yes, many times, for many years.

Ah, please start a fresh thread and give us details of some of them.

 

[Wombat]

Actually yes, many times, for many years.

Rhetoric.

The references/details/results please.

Also, please expound on the audio and electronics credentials that you infer but never provide background to.

Edit: Amirm, you beat me to it.

 

[March Audio]

First, Amir's definition of level linearity is that over 0.1 dB level deviation is failure, which is just 1% level deviation, which can be affected by all kind of things, probably especially noise, but also measurement equipment.

What I really noticed was atomicbob's post #19 where the dac1541 -140 dB signal is still visible at -140 dB, how close to -140 dB is hard to see, but I would say it's pretty good, especially when compared to another R-2R DAC....

Some measurement, not just the dac1541 but also others, are disturbed by mains noise, and mains noise can come from many things, especially when we're looking at that low signal levels. And then there is the issue of balanced connections, there seems to be several ways of doing it, and it seems like that audiophile equipment is not doing it the same way as professional equipment, partly to be compatible with single ended signals....

You have to carefully evaluate measurements, t.ex. many people believe that R-2R DACs distortion of 0.01% is way to high, yet they listen to speakers with 1% distortion.... I would say that pure harmonic distortion is not that important, the problem is with non harmonic distortion, and some of those we haven't figure out how to measure, that's why most delta sigma DACs are described as "harsh", even when they measure perfect....

I do know this site is really about measurements, but you should listen to the devices under test too.... But it will then take way more time....

There are a few points to comment on here.

Whilst Amirs personal limit of 0.1dB may seem at first to be somewhat arbitrary and excessively tight, as we see many dacs that are linear to that figure to very low levels, then it becomes a reasonable target to aim for**. When high FFT resolutions and averaging are utilised random noise becomes a very small issue. In my second set of measurements I was also careful to test where there were no spuria, mains or otherwise. Yes of course the linearity of the measurement equipment is an issue, however we do see other DACs performing very well and showing to be linear at low levels where the 1021 is not. Also as mentioned previously, Amir and I are in the process of validating measurements against each other.

Im not sure what point you are making regarding the balanced connections WRT mains noise. I would expect a balanced connection to show very little/insignificant amounts of mains noise. With the 1021 we see higher levels than I would expect. Of course I was initially suspicious of the measurement set up, but when there is no problem found, and I measure other DACs with balanced or single ended output that show pretty much zero mains noise, and when SBAF measurements also show mains noise, its difficult to draw any conclusion other than it is a DAC issue. BTW I tested balanced, single ended and direct from the ladder with same problem noted.

I very much agree with your point regarding relative problems in a reproduction chain. Yes speakers have huge distortion by comparison, but we are testing the technical performance of the dac and many other dacs are vastly superior in this respect. If you are not concerned by harmonics at levels up to -86dBFS, then you are also by definition saying that 24 bit content is pointless. I might actually agree with that, but you will certainly end up in an argument with many subjective audiophiles .

Can you be more specific about these things we apparently cant measure? I simply dont agree that "most" Delta Sigma designs sound harsh. I dont see any basis for that statement. All of us here listen to devices. Where we differ is that we are very aware of the problems associated with subjective evaluation. From my perspective FWIW, this dac sounds fine, I would be quite happy to live with it. It sounds a little different to other dacs I have. However I dont hear "more" with it and I dont think it sounds more "natural" whatever that may be. Maybe the measured issues are below audible thresholds, but equally maybe they are what cause it to sound a little different.

 

[March Audio]

As mentioned previously I think the limit of the QA is about 22bits for this measurement. I will probably limit graphs to that in future. It will interesting to see how it correlates to Amirs measurements.

Amir, how are you with the Explorer2 measurements for the comparison?

**So on the same scale

DAM1021

Motu 8A

Chord Mojo

 

[soekris]

There are a few points to comment on here.

Whilst Amirs personal limit of 0.1dB may seem at first to be somewhat arbitrary and excessively tight, as we see many dacs that are linear to that figure to very low levels, then it becomes a reasonable target to aim for**. When high FFT resolutions and averaging are utilised random noise becomes a very small issue. In my second set of measurements I was also careful to test where there were no spuria, mains or otherwise. Yes of course the linearity of the measurement equipment is an issue, however we do see other DACs performing very well and showing to be linear at low levels where the 1021 is not. Also as mentioned previously, Amir and I are in the process of validating measurements against each other.

Im not sure what point you are making regarding the balanced connections WRT mains noise. I would expect a balanced connection to show very little/insignificant amounts of mains noise. With the 1021 we see higher levels than I would expect. Of course I was initially suspicious of the measurement set up, but when there is no problem found, and I measure other DACs with balanced or single ended output that show pretty much zero mains noise, and when SBAF measurements also show mains noise, its difficult to draw any conclusion other than it is a DAC issue. BTW I tested balanced, single ended and direct from the ladder with same problem noted.

I very much agree with your point regarding relative problems in a reproduction chain. Yes speakers have huge distortion by comparison, but we are testing the technical performance of the dac and many other dacs are vastly superior in this respect. If you are not concerned by harmonics at levels up to -86dBFS, then you are also by definition saying that 24 bit content is pointless. I might actually agree with that, but you will certainly end up in an argument with many subjective audiophiles .

Can you be more specific about these things we apparently cant measure? I simply dont agree that "most" Delta Sigma designs sound harsh. I dont see any basis for that statement. All of us here listen to devices. Where we differ is that we are very aware of the problems associated with subjective evaluation. From my perspective FWIW, this dac sounds fine, I would be quite happy to live with it. It sounds a little different to other dacs I have. However I dont hear "more" with it and I dont think it sounds more "natural" whatever that may be. Maybe the measured issues are below audible thresholds, but equally maybe they are what cause it to sound a little different.

The balanced three pin XLR connector have signal+, signal- and earth. The questions is how to wire them, in the professional world earth is protective earth, in the audiophile world the earth pin is mostly connected to signal gnd.... The receiving end are supposed to be a transformer using signal+ and signal-, which result in excellent common mode rejection. But most modern inputs are using active balanced to single ended conversion, resulting in much worse common mode rejection, due to used resistor tolerances, and worse, capacitor tolerances which really ruin the common mode rejection at lower frequencies where you need it the most.... See http://www.rane.com/note151.html for more.

The DIY dam1021 build seems to do the XLR wiring the professional way, which can result in higher mains hum if the input is not transformer coupled.... And who knows how much the three main transformers affect that build, and how the wiring is.... But it's not anything you can hear, the human ear is pretty insensitive at low frequencies at low levels.... See https://en.wikipedia.org/wiki/A-weighting for more.

And just fyi, distortion do not have anything to do with dynamic range, especially with a Sign Magnitude R-2R DAC where distortion is relative to actual signal level, not full level, so 24 bits do matter....

This is not the place to start long discussion about what you can hear or not, just want to point out there is a big difference between measuring using sine tones and playing actual music which are very complex..... Most with just a basic knowledge of electronics can take a $3 dac chip, add a couple of external parts and create a DAC that measure extremely good using simple tests, but don't sound that good when playing music.... Also, there are large difference to what people can hear, I t.ex. can't hear shit above 16 Khz. While at some of the Audio Shows I have meet people with fantastic ears.... T.ex., one guy immediate after listening to my DAC asked if it had something else than a brickwall fir filter, and was happy when I switched to the butterworth filter....

And yes, I might try to discredit Delta Sigma DAC's more than they deserve, I kinda have to . But I have also meet many people that agree that a good implemented R-2R DAC beats any Delta Sigma DAC.

I like what you do here, but you should also try improving measure methods, if you can get closer to match measurements to what you hear it would be a great step forward....

 

[March Audio]

Erm the screen in xlr is a shield to be connected to the equipment chassis and pin1. It does not get connected to signal gnd. There is no differentiation between professional and audiophile.

Transformers have nothing to do with this, there is no requirement for them to be used and no, active does not lead to worse cmr. You are starting to concern me now with these sort of comments. As an example the Motu 8a is active and shows no issues with mains on balanced connection.

Motu 8a balanced output noise floor measured on QA401

Motu 8A balanced input noise floor

I dont understand your comments about distortion and dynamic range. My point was that if there are multiple distortion components, harmonic and IMD at high levels, how are you supposed to hear the more subtle lower level signals contained within a 24 bit recording?

Regarding more complex music waveforms, lets consider the imd performance of your dac with just 2 tones shall we? All I see is that your dac adds significant signals that arent in the original recorded waveform. The more complex the waveform (music signal) the more complex the distortion and spurious signals will be.

1021

Motu 8a

So what are you saying? Your dac sounds better because of all the spurious signals? If it sounds different to you than other dacs have you stopped to wonder why?

Before you start implying that measurements dont represent what we hear you really do have to produce some evidence that your personal subjective impressions have validity, or represent what you think they do.

 

[soekris]

Erm the screen in xlr is a shield to be connected to the equipment chassis. It does not get connected to signal gnd. There is no differentiation between professional and audiophile.

Transformers have nothing to do with this and no, active does not lead to worse cmr. You are starting to concern me now with these sort of comments.

I dont understand yor comments about distortion and dynamic range.

Regarding more complex music waveforms, lets consider the imd performance of your dac with just 2 tones shall we? All I see is that your dac adds significant signals tbat arent in the original recorded waveform.

So what are you saying? Your dac sounds better because of all the spurious signals?

Before you start implying that measurements dont represent what we hear you really do have to produce some evidence that your personal subjective impressions have validity.

To clarify, I was talking about the XLR connector pin 1.

A perfect active balanced input can have great cmr. But as you should know, they don't exists. They're normally done using opamps, resistors and input capacitors, and any unbalance there will affect cmr, most use 1% resistors and 5% capacitors, which will result in less than 50 dB cmr, lower at 20 hz. Measurement equipment often use better parts, but still nowhere close to the cmr of transformers.

I'm sorry you don't understand my comments about distortion and dynamic range, can't do anything about that, as I say, they just don't have anything to do with each other.

As you said yourself, IMD is mostly a product of Distortion, so of course it will look bad on a dam1021, especially when noise is mixed in.... But the question is if anything 90 dB down is bad when it's all a function of the basic tiny unlinearities of the DAC.... Think about a signal in the time domain instead of the frequency domain.

It's just not my "personal subjective impressions", a lot of people on diyaudio have built DAC's using the dam1021 modules, and are very happy with the results. The advantage of R-2R DAC's are that they play the music like it was recorded, one sample at a time with all the bits, without funny processing like your typical 5th order delta sigma modulator.... Remember, the reason why everybody switched to delta sigma DACs was not because they sounded better, but because they were cheaper....

 

[March Audio]

Your comments re cmr simply arent true. A real world example is above with a modest cost professional audio interface. It shows no mains related signals on its balanced input or output.

I think its you that does not understand the point I was making, it had nothing to do with dynamic range.

The frequency and time domain are intrinsically related. Your point makes no sense.

Im afraid the measurements unequivocally show that your R2R dac does not play the signal as it was recorded. It adds a significant quantity of signals that are simply not in the original.

Again, on what basis do you say that DS sounds inferior? Becauase a number of audiophiles say so? Sorry but thats not compelling evidence. However there is clear evidence to indicate that your dac may sound different or have a noticeable "personality".

 

[rebbiputzmaker]

To clarify, I was talking about the XLR connector pin 1.

A perfect active balanced input can have great cmr. But as you should know, they don't exists. They're normally done using opamps, resistors and input capacitors, and any unbalance there will affect cmr, most use 1% resistors and 5% capacitors, which will result in less than 50 dB cmr, lower at 20 hz. Measurement equipment often use better parts, but still nowhere close to the cmr of transformers.

I'm sorry you don't understand my comments about distortion and dynamic range, can't do anything about that, as I say, they just don't have anything to do with each other.

As you said yourself, IMD is mostly a product of Distortion, so of course it will look bad on a dam1021, especially when noise is mixed in.... But the question is if anything 90 dB down is bad when it's all a function of the basic tiny unlinearities of the DAC.... Think about a signal in the time domain instead of the frequency domain.

It's just not my "personal subjective impressions", a lot of people on diyaudio have built DAC's using the dam1021 modules, and are very happy with the results. The advantage of R-2R DAC's are that they play the music like it was recorded, one sample at a time with all the bits, without funny processing like your typical 5th order delta sigma modulator.... Remember, the reason why everybody switched to delta sigma DACs was not because they sounded better, but because they were cheaper....

Thank you for your posts and the info you are trying to convey. In the real world what you are saying makes sense.

 

[Purité Audio]

In the real world listeners often enjoy various types of added distortion .
Keith

 

[Blumlein 88]

First, Amir's definition of level linearity is that over 0.1 dB level deviation is failure, which is just 1% level deviation, which can be affected by all kind of things, probably especially noise, but also measurement equipment.

What I really noticed was atomicbob's post #19 where the dac1541 -140 dB signal is still visible at -140 dB, how close to -140 dB is hard to see, but I would say it's pretty good, especially when compared to another R-2R DAC....

Some measurement, not just the dac1541 but also others, are disturbed by mains noise, and mains noise can come from many things, especially when we're looking at that low signal levels. And then there is the issue of balanced connections, there seems to be several ways of doing it, and it seems like that audiophile equipment is not doing it the same way as professional equipment, partly to be compatible with single ended signals....

You have to carefully evaluate measurements, t.ex. many people believe that R-2R DACs distortion of 0.01% is way to high, yet they listen to speakers with 1% distortion.... I would say that pure harmonic distortion is not that important, the problem is with non harmonic distortion, and some of those we haven't figure out how to measure, that's why most delta sigma DACs are described as "harsh", even when they measure perfect....

I do know this site is really about measurements, but you should listen to the devices under test too.... But it will then take way more time....

I was taken aback somewhat when I saw Amir calling it at .1 db deviation. Though equally arbitrary I would have chosen 1 db. It is not like we are listening to -80 or -90 db or lower signals. Play a tone at that level in your home rig and you'll hear nothing. Where the linearity might impact sound is when it is poor enough it corrupts signals at -60 db which we could hear especially in low level detail or reverb tails as they fade out. In fact it would be interesting to hear an artificial hall ambiance fade to black on devices to see if it sounds different with different low level linearity. Say between the Soekris and the Yiggy and a good delta-sigma based DAC like the Topping or Exasound.

Now I do question the idea delta sigma DACs are harsh from non-harmonic distortion. What's the basis for that? What do you think is going on? There are certainly ways to detect non-harmonic spurious signals with various test signals.

 

[amirm]

Play a tone at that level in your home rig and you'll hear nothing.

Just a note here that a lot of our membership uses headphones and there, lot lower level sounds/distortions are audible than they are with speakers.

 

[rebbiputzmaker]

In the real world listeners often enjoy various types of added distortion .
Keith

Yes, you are entitled to your opinion. I feel measurements are important,but somehow should be correlated with sound.

Just a note. I do find it interesting how some here can vacillate at will between
objective /commentary anywhere, anytime with nary a word here, while others have great difficulty saying almost anything.

 

[rebbiputzmaker]

Just a note here that a lot of our membership uses headphones and there, lot lower level sounds/distortions are audible than they are with speakers.

Maybe it would be a thought to measure what is the threshold you can hear with you headphones to get a baseline.

 

[gepardcv]

To clarify the XLR pin 1 issue: @BE718, could you please confirm that both the dam1021 build and the Motu 8a you measured short pin 1 to the chassis (ideally right next to the connector)? These things are definitely not consistently wired in all equipment, and can introduce (audible!) hum when done incorrectly. It’s hard to tell what’s going on from the photo on page 1 of the thread. There may also be PSU and grounding scheme problems unrelated to the DAC itself (which is what I think @soekris was alluding to).

 

[amirm]

Maybe it would be a thought to measure what is the threshold you can hear with you headphones to get a baseline.

I did that test and from what I recall, I could hear -90 dB test tone.

 

[RayDunzl]

I did that test and from what I recall, I could hear -90 dB test tone.

With what SPL produced at odB?

---

I hear nothing (1kHz) on headphone (HD650 full available volume from DAC)

Very faint sound on speakers - wide open. Preamp level 151, disappears at about 121/151

Assuming 138V peak swing = 0dBfs - 700W @ 4 Ohms

-90dB = 0.000032 x 138V = 0.004416Vpk

@4 Ohms = 0.0000049Wpk

Phew.

Wide open at Listening position: estimating 120dB SPL at odBfs tone. Nice third harmonic pops up.

And the "all but inaudible" level: 25dB yields 115db for full scale tone.

Tone:

Conclusion:

Can be heard in isolation (by me too), with some effort.

 

[DonH56]

A few random thoughts...

I've not (or only rarely) used linearity plots like that; other tests show the same thing, just a different plot results. As a way to show relative differences among DACs I like them. Note 0.1 dB ~ 1%, so 1% deviation is not an unreasonable criteria, but only about 0.3 bits if you reference to SNR based upon quantization noise. You could argue for almost anything but a consistent reference is good enough for me and I can always read the graphs for more info or to use my own metric. It's a relative comparison so as long as the basis is the same it works for me.

It is probably worth noting that correctly wiring pin 1 and in fact just using XLRs (balanced) does not guarantee no ground loops or common-mode problems. And transformers don't always exhibit great CMRR plus have plenty of other issues. Neither active nor passive differential designs are perfect, you're always making trades among compromises. Great and awful examples of either type (active or transformer) exist, natch.

The Rane note does not really address transformer vs. active differential circuits, or barely, unless I missed it (possible, quick skim). It does address the issues with certain schemes to convert between balanced (differential) and single-ended (unbalanced?) using transformers or active circuits. In those cases the active circuit implementation is critical and it is true rarely provides decent CMRR. In fact that is one thing I have decried for years, the proliferation of various quasi-differential schemes marketed as having the same benefits as fully differential, which includes a number of consumer and pro audio components that do not implement a truly differential transmitter and receiver but throw in a TRS or XLR jack and call it "balanced". Even something as simple as a resistor on the "cold" leg so the impedance looking in is roughly the same as the "hot" leg is marketed as balanced based on the "balanced" impedance but in practice CMRR is not much better than a single-ended connection (i.e. almost nil, often as low as 6 - 20 dB).

Finally, a number of us have thought about why certain measurements and DACs measure better or worse. As a designer I am curious and always wondering "why" and how it could be improved. As a consumer, I am more interested in a consistent comparison among all the units. If there is a problem with the test scheme that consistently favors some units then that should be fixed. But, if the "problem" is in how we as consumers use the unit, and the test is an accurate reflection of real-world usage, then as far as I am concerned the poorly-performing unit has a real issue. I have designed plenty of things that test awesome on the bench but then I had to redesign or fix because in the real-world environment they failed. When my "perfect" ADC fails because the radar's T/R switch leaks some of the multi-kW output through the RF and IF chain all the way back and overloads my input, I have to come up with a fix, and chances are it won't be replacing the T/R switch or telling the RF front end guys they have problems. I'll figure out the leakage path and fix it.

If a consumer DAC only has good performance with my PC doing nothing else but driving it, or I have to use special power supplies and specific cables/adapters to avoid power or ground noise coupling in an average audio system to meet the published specs, then in the real world that DAC (or whatever) has failed me. As an engineer and a consumer.

IME/IMO/FWIWFM/YMMV/my 0.000001 cent (microcent)/etc. etc. etc. - Don