[feifan]

I did a frequency response test (using miniDSP EARS & REW) on my SMSL SU-8 v2 DAC and found that the phase performance is strange. I used 2 headphones: Sennheiser HD650 & Hifiman HE-4XX, both single-ended. I also compared the SU8 to 2 other DACs: Grace SDACB & XiangSheng DA-05B. All DACs were paired with the Monolith THX AAA 887 balanced headphone amp.

In all the tests (SPL, distortion, GD, Clarity, etc.) except for phase, results were similar. However, for both heaphones, the SU8 phase was -90 & -128 degrees. For the other 2 DACs, the phase was 81 & 53 and 80 & 52 degrees.

My question is, is the -90 & -128 degrees phase for the SU8 something to worry about?

 

[bigshot]

What makes you think it is the fault of the DAC and not the headphones or the cabling? Have you isolated those variables? Why not just measure the output of the DAC without headphones?

 

[feifan]

What makes you think it is the fault of the DAC and not the headphones or the cabling? Have you isolated those variables? Why not just measure the output of the DAC without headphones?

Thanks for the response, and apologies for the lack of clarity in my post. Go to https://feifan2.wordpress.com/2019/12/12/graphs-for-smsl-su-8-dac/ to view a graph of the different phase results.

Yes, I isolated the poblem to the SU8. The phase results for the headphones with the other two DACs were similar, in the positive degree range. It's only with the SU8 that the phase degrees are in the deep negative range for both heaphones.

The cabling remained constant across all tests.

Re "measure the output of the DAC without headphones" -- I'm not sure how to do that with (or without) the EARS rig.

Question: I'm not sure how the -90 and -128 phase degrees translate to SQ for headphones. I don't hear a difference in timing in my usual test tracks. Could the DAC still be fine with this apparent anomaly?

 

[bigshot]

I'm afraid the way you are stating the question, it's like an alphabet soup of models. But I think you have signal processing with the miniDSP in there. Why wouldn't you suspect that is the cause? Pare it back and test without so many variables in the chain. This isn't in my wheelhouse, but it looks to me like you are overcomplicating your rig so it's hard to know where the change is coming from.

 

[SoundAndMotion]

Yes, I isolated the poblem to the SU8. The phase results for the headphones with the other two DACs were similar, in the positive degree range. It's only with the SU8 that the phase degrees are in the deep negative range for both heaphones.
[snip]
Question: I'm not sure how the -90 and -128 phase degrees translate to SQ for headphones. I don't hear a difference in timing in my usual test tracks. Could the DAC still be fine with this apparent anomaly?

The SU8 is inverting the signal compared to the other two. Inverting is the same as adding or subtracting 180deg for all frequencies. In your case it's subtracting. You shouldn't hear a difference.

Re "measure the output of the DAC without headphones" -- I'm not sure how to do that with (or without) the EARS rig.

If you have a USB ADC, you would connect the DAC outputs to the ADC inputs and plug the ADC's USB in instead of the EARS. If you don't have an ADC, you can't.

 

[feifan]

I'm afraid the way you are stating the question, it's like an alphabet soup of models. But I think you have signal processing with the miniDSP in there. Why wouldn't you suspect that is the cause? Pare it back and test without so many variables in the chain. This isn't in my wheelhouse, but it looks to me like you are overcomplicating your rig so it's hard to know where the change is coming from.

Thanks for your reply, bigshot. I'm not sure if I follow you. The graph is a compilation of 6 different test runs -- not 6 runs all at once. All variables are kept constant, except for the switches in 3 DACs. I tested each DAC with 2 headphones. The purpose was to isolate the problem. Assuming that DAC1 is the SMSL SU-8 and DAC2 and DAC3 are the other two DACs, then DAC1 is the only one that registered in the deep minus range for phase (all 3 DACs were similar in the other dimensions -- SPL, distortion etc.). This was true for both headphones.

In actual informal listening tests, I don't hear phase differences with the SU-8 in the line-up despite the minus-degree results. Thus, I'm wondering what this phase inversion means. Do I have a defective DAC? If yes, how does this defect impact SQ? Is this defect benign as long as we don't actually hear it?

 

[bigshot]

Is the inversion some sort of noise cancellation?

The general rule of thumb of home audio is, "If you can't hear it, don't sweat it."

 

[feifan]

The SU8 is inverting the signal compared to the other two. Inverting is the same as adding or subtracting 180deg for all frequencies. In your case it's subtracting. You shouldn't hear a difference.



If you have a USB ADC, you would connect the DAC outputs to the ADC inputs and plug the ADC's USB in instead of the EARS. If you don't have an ADC, you can't.

SoundAndMotion, thanks for the response and the clarification. If the inversion is basically inaudible in actual listening tests, then would it be safe to say that this "problem" with the SU-8 is benign abnd simply a design quirk or defect?

I don't have a USB ADC. Can you recommend an inexpensive, simple, yet reliable ADC?

 

[feifan]

Is the inversion some sort of noise cancellation?

Good question. The test DACs and amp are all balanced (XLRs at the interconnects), but I used single-ended headphone outputs. I did earlier tests with the Sennheiser HD800 in balanced mode and found a similar inversion, so I didn't think balanced or SE mattered. But this test wasn't systematic, so I may need to retest with the HD650 and HE4XX with balanced cables.

The general rule of thumb of home audio is, "If you can't hear it, don't sweat it."

Haha. Good point. Still, I'm curious and want to know what's up. Once again, thanks for the helpful responses.

 

[castleofargh]

If you use the same interconnect cables and headphones, then the DAC does this and there is no need to measure more. But if you want to anyway, or if it's possible that a cable is the cause, just plug the DACs straight into the mic input of your computer and run REW with that loop without amp or headphone. You'll see if the impulse response peak is up or down(and the opposite of the other DAC if one is out of phase). It's probably a better way to look at this than your phase graph, as some internal delay could be enough to shift things depending on your settings. While the peak of the impulse response doesn't care if the DAC lags a little before sending the signal.


Having the signal inverted goes unnoticed for almost everybody, and is still not so obvious to detect when we're aware of it IMO. There are also headphones and IEMs that end up that way, because the cable wasn't wired with the attention it deserved. When they mess up inverting only one channel or worst, it's immediately noticeable by ear, so they more rarely end up selling those. But inverted phase will usually show up only if you go looking for it. So I'm not surprised when I encounter some.
Now I couldn't tell you if your device is a lemon, or if they have designed their DAC that way(oops). They probably know the answer to that, or you could try to ask if other owners and pray that the 3 guys in the world who own one and can handle REW, end up reading your post. ^_^

 

[feifan]

If you use the same interconnect cables and headphones, then the DAC does this and there is no need to measure more. But if you want to anyway, or if it's possible that a cable is the cause, just plug the DACs straight into the mic input of your computer and run REW with that loop without amp or headphone. You'll see if the impulse response peak is up or down(and the opposite of the other DAC if one is out of phase). It's probably a better way to look at this than your phase graph, as some internal delay could be enough to shift things depending on your settings. While the peak of the impulse response doesn't care if the DAC lags a little before sending the signal.


Having the signal inverted goes unnoticed for almost everybody, and is still not so obvious to detect when we're aware of it IMO. There are also headphones and IEMs that end up that way, because the cable wasn't wired with the attention it deserved. When they mess up inverting only one channel or worst, it's immediately noticeable by ear, so they more rarely end up selling those. But inverted phase will usually show up only if you go looking for it. So I'm not surprised when I encounter some.
Now I couldn't tell you if your device is a lemon, or if they have designed their DAC that way(oops). They probably know the answer to that, or you could try to ask if other owners and pray that the 3 guys in the world who own one and can handle REW, end up reading your post. ^_^

castleofargh, thanks for the enlightening response and touch of whimsy . Lightens the load. (Pun intended.) I'll need to decide in the next day or two if I'm going to keep the SU-8 or return it for a refund. After a few more test runs with balanced cans, I'm going to stop testing altogether and simply trust my ears. Out of curiosity and when I can find the time, I'll be futzing with the DAC-computer/REW loop to to see how that works. Fascinating stuff. Thanks!

 

[feifan]

Is the inversion some sort of noise cancellation?

bigshot, your question about noise cancellation made me realize that I should've tested with the balanced versions of the cans instead of the single-ended. The results (see the added graph) were more or less similar, but with the results flipped.

 

[SoundAndMotion]

SoundAndMotion, thanks for the response and the clarification. If the inversion is basically inaudible in actual listening tests, then would it be safe to say that this "problem" with the SU-8 is benign abnd simply a design quirk or defect?

I don't have a USB ADC. Can you recommend an inexpensive, simple, yet reliable ADC?

The SU-8 is inverted compared to the other 2. I don't know if the design quirk is with the SU-8 or the other 2 (less likely, I suppose). Don't worry about it either way.

There is a wide range of stereo ADCs that would work for you, ranging from just a few bucks to $2500 (Lynx Hilo) and up. A lot of people use and are satisfied with the inexpensive Behringer USB DAC/ADC interfaces (UCA222 or UCA202) that run about $20.
If you are the DIY type, you can make an excellent ADC using the PCM4222EVM with a low noise (LT3045 based) power supply (± 15V and +5V). You'll need a S/PDIF to USB converter, unless you have a S/PDIF (coaxial or TOSLINK) interface on your computer. If this last bit is gibberish, ignore it.

 

[feifan]

The SU-8 is inverted compared to the other 2. I don't know if the design quirk is with the SU-8 or the other 2 (less likely, I suppose). Don't worry about it either way.

There is a wide range of stereo ADCs that would work for you, ranging from just a few bucks to $2500 (Lynx Hilo) and up. A lot of people use and are satisfied with the inexpensive Behringer USB DAC/ADC interfaces (UCA222 or UCA202) that run about $20.
If you are the DIY type, you can make an excellent ADC using the PCM4222EVM with a low noise (LT3045 based) power supply (± 15V and +5V). You'll need a S/PDIF to USB converter, unless you have a S/PDIF (coaxial or TOSLINK) interface on your computer. If this last bit is gibberish, ignore it.

The Behringer USB DAC/ADC interfaces (UCA222 or UCA202) sound like a winner. It'll be one of my projects for the winter break. Thanks!