• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Do interconnect cables make a difference? A null test result

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#1
With all the discussions about cables, I thought DeltaWave could be used to run a null-test comparison. Here's a quick test of three interconnects.

Equipment
Apogee Element24 in loopback mode @ 24/96k. Interconnects (DUTs): GLS Pro audio 6ft patch cable, Mogami Gold 6ft, and Monoprice AES/EBU 5ft cable:).

Results
The first number is the RMS difference between the two captures, the second number is the correlated null, both in dB. Original2.wav comparison is the original digital file compared to the captured file with the specific interconnect:

1559657891895.png


With the largest RMS difference between Mogami Gold and GLS interconnects (-99dB), the two files match, bit for bit, for over 57% of all the samples in the two captures.

Discussion
Different interconnects appear to have a measurable effect on the captured audio.
  1. When each interconnect is compared to itself (yellow highlight) the values are pretty much at the limit of the DAC/ADC ability of Element24 to capture
  2. When compared to another interconnect, the nulls are 8 to 10 dB worse the RMS difference, but a smaller difference in the correlated null.
    This means that the error is primarily in amplitude and not in timing/frequency
  3. The difference between DAC/ADC captures and the original digital are significant (-42dB RMS) but very consistent across interconnects. The main effect here is obviously in the processing in the DAC/ADC and not due to interconnects.
Method
Three captures were recorded using each cable, interleaving cables during the test to ensure external factors didn't unfairly affect any one cable. The loopback connection was using an XLR-terminated cable (DUT) inserted between Mic input and the line output of Element24. I used this audio file from the Gearslutz AD/DA evaluation thread. Audacity was used to capture the output from the ADC stage. DeltaWave was used to compare the captures.

THD+N for Element24 appears to be around -112dB, as in the below, off-the-screen photo. These are the same settings I used to run the null tests:
element24.jpg
 
Joined
Apr 19, 2016
Messages
43
Likes
54
#2
The difference between DAC/ADC captures and the original digital are significant (-42dB RMS) but very consistent across interconnects.
42dB seems a suspiciously high value to me. Are you sure that the levels are identical, both signals optimally aligned and that there's no drift (sample rate difference)?
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#3
42dB seems a suspiciously high value to me. Are you sure that the levels are identical, both signals optimally aligned and that there's no drift (sample rate difference)?
As sure as I can be :) The software does the hard work. Remember that there are additional filters in DA/AD process that will affect null values when comparing original, unfiltered file and one recorded through the loopback interface. These do not affect the comparison with two loop-back captures. By the way, dbA (A-weighted) null values are about 10dB better for original/loop-back comparisons. See below (Mogami vs original2.wav):

1559664669062.png


Also remember that the null comparison is computed for the full range of the capture, i.e., 0-48kHz
 
Last edited:

dreite

Active Member
Joined
Aug 10, 2018
Messages
145
Likes
138
Location
Bremerton, WA
#4
'You ever try using a real-time program like ARTA in the differential ("FR2") mode?
It allows to A/B two DUT's (cables in this case) directly and yields precision amplitude/phase plots showing the difference.

Dave.
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#5
'You ever try using a real-time program like ARTA in the differential ("FR2") mode?
It allows to A/B two DUT's (cables in this case) directly and yields precision amplitude/phase plots showing the difference.

Dave.
I have not, but sounds like it should work.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,674
Likes
2,603
Location
The Neverlands
#6
Can you show the loopback results of the system with the 3 different cables in the loopback ?
Or different cables on L and R ?
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#7
Can you show the loopback results of the system with the 3 different cables in the loopback ?
Or different cables on L and R ?
Are you looking for something other than what I posted?
1559675679301.png


Different cables on L and R will add cross-talk as another variable, but I can certainly do that.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,674
Likes
2,603
Location
The Neverlands
#8
Was looking for 3 distortion plots (multitone ?) of the ADC-DAC loopback of each cable to see how much the loopbacks varied, in what frequency range and how much mains hum etc. would be leaking in.
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#9
Was looking for 3 distortion plots (multitone ?) of the ADC-DAC loopback of each cable to see how much the loopbacks varied, in what frequency range and how much mains hum etc. would be leaking in.
It wasn't mains frequency. I tried notch-filtering out 60Hz and that didn't change much in the results. The differences are mostly under 200Hz, as the nulls become much more consistent after high-pass filtering. I'll post some details later.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,952
Likes
5,172
#10
I've not posted them, but I've done the same testing as Paul here. I get very similar results.

About his loopbacks vs the original his numbers are normal. I'm not sure I understand all the reasons why. I've got a couple pieces of gear that will get loopbacks in the mid 50's range which is about the most you see with anything. Just as a for instance in the multi-generation tests, you'll see something in the mid-40 to 50 range between any two generations. It remains relatively constant. In tests of 3rd vs 3rd generation repeat or 5th vs 5th generation repeat you get those results near 100 db.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,952
Likes
5,172
#11
I think some of the variability is changes in phase at both high and low end. Most interfaces roll off somewhere between 10 and 20 hz. That effects phase vs originals enough to reduce the null. So rolling off below 200 hz makes sense as giving better results. Also random noise is pinkish to brownish so the most variability will be noise down below 200 hz.
 
Joined
Apr 19, 2016
Messages
43
Likes
54
#12
Remember that there are additional filters in DA/AD process that will affect null values when comparing original, unfiltered file and one recorded through the loopback interface. These do not affect the comparison with two loop-back captures.
IME the DAC/ADC chain should be quite transparent. After all it's the de-facto setup for out-of-the-box mastering with analogue outboard gear. Would it be possible to provide two test files (the original and one of your loop-back files) ? I'd love to listen and see why the null is only -42dB :)
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#13
Was looking for 3 distortion plots (multitone ?) of the ADC-DAC loopback of each cable to see how much the loopbacks varied, in what frequency range and how much mains hum etc. would be leaking in.
The same comparison as before but with a high-pass filter of 200Hz shows a much improved null result, pointing to lower frequencies as the main differentiator between the interconnects. Surprising!? Maybe something specific to Element24, or perhaps due to how these interfaces are designed, as Dennis said. What is curious is that the same error under 200Hz does not seem to affect multiple captures with the same interconnect....
  • Mogami vs Monoprice AES/EBU improves RMS null from -102dB to -111dB
  • Mogami vs GLS XLR improves from -99dB to -107dB
  • GLS XLR vs Monoprice improves from -105dB to -108dB
Mogami vs Monoprice:
1559680724606.png


Mogami vs GLS:
1559681004844.png


GLS vs Monoprice AES/EBU:
1559681382451.png
 
Last edited:

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,674
Likes
2,603
Location
The Neverlands
#14
I nulled cabled (about 30 years ago) with this device:

dist analyzer.JPG

The zero adjust knob is a 25 multiturn pot and even with that pot it was difficult to fully null.
As you can see the difference signal could be amplified 20dB or 40dB.
Even at the 40dB gain setting I could only hear a faint noise (from the 40dB amplification) and maybe very faintly some music.
This suggests that the null would have been at least -90dB and maybe could have been even better with more patience and a finer pot.
Ethan showed the same result so I am quite certain that cables should null well below -90dB.
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#15
I nulled cabled (about 30 years ago) with this device:

View attachment 27234
The zero adjust knob is a 25 multiturn pot and even with that pot it was difficult to fully null.
As you can see the difference signal could be amplified 20dB or 40dB.
Even at the 40dB gain setting I could only hear a faint noise (from the 40dB amplification) and maybe very faintly some music.
This suggests that the null would have been at least -90dB and maybe could have been even better with more patience and a finer pot.
Ethan showed the same result so I am quite certain that cables should null well below -90dB.
They null to better than -100dB. And my volume control goes all the way up to +100dB in DeltaWave :)

1559682566551.png
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#16
IME the DAC/ADC chain should be quite transparent. After all it's the de-facto setup for out-of-the-box mastering with analogue outboard gear. Would it be possible to provide two test files (the original and one of your loop-back files) ? I'd love to listen and see why the null is only -42dB :)
Here are the files, posted under fair use doctrine. I'll remove them in a few days. The files are 32-bit/96KHz, exported directly from DeltaWave.

orig2_ref.wav: original wave file from Gearslutz, upsampled to 96kHz in Audacity
mogami_b_comp.wav: Mogami Gold loopback capture of orig2 file using Element24
orig2_mogami_delta.wav: after matching phase offset, drift (none was found) and amplitude, this is the delta file (orig2 - mogami_b_comp).

Sorry, this was done in a bit of a rush, so hopefully I did it all right. Let me know what you find.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,952
Likes
5,172
#17
If you haven't used Deltawave it is quite a nice bit of software. You can listen to the difference files and choose gain as shown. When nulling cables against each other, and when nulling a cable now vs the same cable 5 minutes from now, you have to turn the gain up to 70 db or more. Then you hear noise, sometimes very faintly hints of the music fading in and out within the noise.

Also for those wondering about a loopback vs the original, there is a long running thread (like over 8 years) using the music Paul used here over on the Gearslutz forum. The results list are updated every couple of months or so. Here is the most recent list:
https://www.gearslutz.com/board/showpost.php?p=13947046&postcount=1743

The best difference in any of the loopbacks is about -59 db RMS.

Now there are some results near or at -80 db, and in those cases one device provided the master clock between DAC and ADC. I've a theory about that. I think if you use the same length clock cable as the analog cable the clock signal takes closer to the same time to reach between the devices and they are more in synch. So the recorded signal is better aligned in time to begin with. If you do a loopback, the ADC and DAC are completely in synch, but the analog signal took a few nanoseconds to get between them. So I think software can't quite match as perfectly as we wish. You also have some devices where an external clock was sent to both devices. Yet in those, like in a loopback, best results are -59 db or worse. I suppose we could test this if we had the same cable and compare 1 meter with 3 meters or some such.

In these cable tests where it is one cable loopback vs another cable loopback, you also have a situation where the timing relative to one another is better than in a loopback vs original. You also would have the same alterations due to phase differences this way at the low and high frequencies.

Deltawave does an excellent job of gain and time alignment, but I'm wondering if that isn't part of the limit right now.
 

pkane

Addicted to Fun and Learning
Patreon Donor
Joined
Aug 18, 2017
Messages
560
Likes
518
#18
I think there's phase and frequency differences at play here. Here's the gain-matched/phase-aligned version of the original file compared to one passed through the loopback with Mogami Gold.

You can see the difference below 20Hz (blue is the original, white is the loopback). You can also see the low-pass filter effect produced by Audacity diverging past 20kHz with ADC filter used in Element24. I'm sure these two differences account for at least some of the unimpressive null values.

1559697035966.png
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
6,863
Likes
3,032
Location
Riverview, FL
#19
How does the result look with a log sweep?
 
Top Bottom