Audio measurement gear

[Schackmannen]

Copied from the RME ADI-2 Pro Fs Manual:
"User’s Guide ADI-2 Pro FS – v 2.3", page 65

33.1 Analog Inputs
XLR
• THD @ -1 dBFS: -113 dB, 0.00022 %
• THD+N @ -1 dBFS: -110.6 dB, 0.00029 %

33.2 Analog Outputs
1/2 XLR
• THD @ -1 dBFS: -112 dB, 0.00025 %
• THD+N @ -1 dBFS: -110 dB, 0.00032 %
• THD @ -3 dBFS: -116 dB, 0.00016 %

Yeah I saw that just now, looks like I first found the manual for the non FS-version. Not sure why the FS-version has a bit worse THD performance since I thought the only big change was improved clock performance with the FS-version.

Edit: https://www.forum.rme-audio.de/viewtopic.php?id=26942 Found a thread where @MC_RME claims the THD+N and SNR specifications is the same for the FS and non FS-versions. Could you clarify this?

 

[Rja4000]

From what

As I wrote, it's not clear how they measured it (that is, with what gear).

RME explains that they have an AP analyzer, but that they do QA with the RME ADI-2 Pro itself and HPW-Works

From the same user manual
"34.18 ADI-2 Pro as Hardware I/O for Measurements

At RME not only the well-known references Audio Precision and Rohde&Schwarz are used, but also simpler, sometimes even more flexible or unusual solutions. A long-time favourite is HpW Works, a software analyzer and generator program that has been in use by RME developers for more than 20 years. Most measurement diagrams shown in this manual were done with it. In many cases we only use the expensive systems for verification. Not because HpW Works would measure wrong, but because of the limited hardware that serves as I/O.

That leads us straight to the ADI-2 Pro. One of its development goals was to make it so good that it can serve as hardware frontend for audio measurement software. Accepting a few limitations the hardware should be capable to easily measure most of the audio interfaces, DACs, ADCs and analog equipment that is in daily use by many. "

 

[KSTR]

but because of the limited hardware that serves as I/O.

That's the point were we are left in the dark...
I trust RME, though, and it seems they are confident enough of their specs not needing to show "industry standard" AP plots. For the sceptics, and for the sake of consistency, they should include those (or ones made with R&S) that match the values given on the specs page.

 

[wynpalmer]

That's the point were we are left in the dark...
I trust RME, though, and it seems they are confident enough of their specs not needing to show "industry standard" AP plots. For the sceptics, and for the sake of consistency, they should include those (or ones made with R&S) that match the values given on the specs page.

I just repeated a loop back test using REW- single ended with no doubling up on the DAC for quickness- I get @1kHz, -1dBFS THD+N -111.3.
Perhaps all of this is just a reflection of their attempt to worse case everything, which in some senses is laudable especially in a professional environment, but when you're a bunch of users trying to duplicate their results and nit picking everything it can get pretty confusing.
By the way, the 3rd harmonic dominates in my measurements. but it's c. 3dB less than the new value in their user guide plot for the ADC and pretty close to the original ADC number despite it being a loop back.

 

[Schackmannen]

@amirm Do you still have your ADI-2 Pro? Maybe you could try to get an FS-version to test and see what the differences are.

 

[MC_RME]

Some comments:

> RME explains that they have an AP analyzer, but that they do QA with the RME ADI-2 Pro itself and HPW-Works

Nope. QA means manufacturing / EOL (end of line) testing, this is done with APx525 and R&S UPD.

THD values are never one single value, but have a range. As you can see in ADC/DAC data sheets the THD or THD+N values are never guaranteed, but 'typical' and 'Min'. Often even the 'Max' value is missing. The values in our manuals should be seen as typical, and are usually a bit better in real-world.

That said the early generations of the ADI-2 Pro had a special trick to reduce THD at high input level (near 0 dBFS), which drove it to partly crazy values up to -120 dB at -1 dBFS. Later on we improved the input stage to reduce DC in DSD recording, which sadly required to remove that trick. The THD at higher levels is now typically -114 dB. Typically means there are small deviations possible, and as usual a few dBs lower in level THD is getting a lot better (back to -120 dB).

Regarding AP measurements: I plan to do a complete measurement series with my APx555B until the end of the year, released as extra PDF to download. Should be fun...

 

[Rja4000]

Some comments:

> RME explains that they have an AP analyzer, but that they do QA with the RME ADI-2 Pro itself and HPW-Works

Nope. QA means manufacturing / EOL (end of line) testing, this is done with APx525 and R&S UPD.

THD values are never one single value, but have a range. As you can see in ADC/DAC data sheets the THD or THD+N values are never guaranteed, but 'typical' and 'Min'. Often even the 'Max' value is missing. The values in our manuals should be seen as typical, and are usually a bit better in real-world.

That said the early generations of the ADI-2 Pro had a special trick to reduce THD at high input level (near 0 dBFS), which drove it to partly crazy values up to -120 dB at -1 dBFS. Later on we improved the input stage to reduce DC in DSD recording, which sadly required to remove that trick. The THD at higher levels is now typically -114 dB. Typically means there are small deviations possible, and as usual a few dBs lower in level THD is getting a lot better (back to -120 dB).

Regarding AP measurements: I plan to do a complete measurement series with my APx555B until the end of the year, released as extra PDF to download. Should be fun...

Hi Marcus
Thanks for your reply, and sorry for my confusion.
So, if I understand, the ADI-2 pro and HPW-Works are used rather for quicly assessing some development options ?

My topic here, at the beginning, was about difference in measurement for IMD SMPTE at low level (around -60dBFS).
Would you have any measurements for that ? Or ballpark figure ?

 

[MC_RME]

My topic here, at the beginning, was about difference in measurement for IMD SMPTE at low level (around -60dBFS).

The measurement I found from you a few pages back looks perfect (as expected at such low levels, zero IMD). Amirm never measures IMD at -60 dBFS as FFT, so waddayumean?

 

[Rja4000]

The measurement I found from you a few pages back looks perfect (as expected at such low levels, zero IMD). Armirm never measures IMD at -60 dBFS as FFT, so waddayumean?

Hi
Yes, as I understand it, what we measure is basically noise (-66dB under-72db, that can only be noise!)

Amir is publishing an XY plot of IMD according to level and his measurement is higher by 15dB.
So, the difference is probably just a matter of averaged vs peak value.
His figures are probably more accurate (51+72=123dB makes more sense than 138) and I'm trying to replicate them.

Anyway, the main use of this analysis is to make sure there is no "incident" when you lower the level.

 

[KSTR]

EDIT: a possible way to measure it halfway correctly by itself could be (THD sans noise):
- ADC : Feed -20dBFS to DAC at +24dBu and set ADC to +4dBu, so ADC distortion will dominate (but DAC noise will dominate, too)
- DAC : Feed 0dBFS to DAC at +4dBu and set ADC to +24dBu, so DAC distortion will dominate (but ADC noise will dominate, too)

Here are the results. 96kHz sample rate, 512k FFT, bin-locked, using a 10Hz offset to better see the differences. Plot scale is dBc (normalized to carrier) for more direct readout.

DAC/ADC running at -1dBFS test level (0dBFS is a bit rude ;-).
ADC is more than 10dB better in this test -- which may not be entirely representative as still some addition could be going on (or cancelling, though less likely as levels and phases must match for this to happen) but I think we can see the tendency.

Below the same test but at -10dBFS test levels for ADC and DAC. Situation has changed, at that level the DAC (running -30dBFS, actually, +20dB makeup gain to have the ADC at -10dBFS) seems to fare a bit better but also develops some spuriae (more on this in a seperate post).

Comparing both plots we can see that the DAC seems to improve more than the ADC from having some clearance to 0dBFS.

For a better isolated test of the ADC performance in loopback style an IMD measurement, using 2 DAC channels mixed together passively, should work out well (on the to-do list).


For the sake of completeness, the previous plot with added spectrum for ADC and DAC both running at -10dBFS (+13dBu setting). ADC seems to dominate H2 whereas the H3 level seems to show some cancelling and H4 some constructive addition.

Overall, the Adi-2 Pro is impressive! Things can be tested down to -110dB distortion quite reliably when we run about -10dBFS levels for DAC and ADC, with no additional tricks.

 

[wynpalmer]

Hi
Yes, as I understand it, what we measure is basically noise (-66dB under-72db, that can only be noise!)

Amir is publishing an XY plot of IMD according to level and his measurement is higher by 15dB.
So, the difference is probably just a matter of averaged vs peak value.
His figures are probably more accurate (51+72=123dB makes more sense than 138) and I'm trying to replicate them.

Anyway, the main use of this analysis is to make sure there is no "incident" when you lower the level.

Here are the results. 96kHz sample rate, 512k FFT, bin-locked, using a 10Hz offset to better see the differences. Plot scale is dBc (normalized to carrier) for more direct readout.

DAC/ADC running at -1dBFS test level (0dBFS is a bit rude ;-).
ADC is more than 10dB better in this test -- which may not be entirely representative as still some addition could be going on (or cancelling, though less likely as levels and phases must match for this to happen) but I think we can see the tendency.View attachment 30493

Below the same test but at -10dBFS test levels for ADC and DAC. Situation has changed, at that level the DAC (running -30dBFS, actually, +20dB makeup gain to have the ADC at -10dBFS) seems to fare a bit better but also develops some spuriae (more on this in a seperate post).
View attachment 30494

Comparing both plots we can see that the DAC seems to improve more than the ADC from having some clearance to 0dBFS.

For a better isolated test of the ADC performance in loopback style an IMD measurement, using 2 DAC channels mixed together passively, should work out well (on the to-do list).


For the sake of completeness, the previous plot with added spectrum for ADC and DAC both running at -10dBFS (+13dBu setting). ADC seems to dominate H2 whereas the H3 level seems to show some cancelling and H4 some constructive addition.
View attachment 30500

Overall, the Adi-2 Pro is impressive! Things can be tested down to -110dB distortion quite reliably when we run about -10dBFS levels for DAC and ADC, with no additional tricks.

Impressive- how did you introduce the offset?

 

[KSTR]

Simply by using 1000Hz (ADC), 1009Hz(DAC) and 1018Hz(both) test frequencies for each run (after checking that the distortion remains the same). And dBc nomalization was done by hand in the overlay panel. This way several distortion plots can be directly compared in one graph.

 

[KSTR]

FWIW, it seems I've found a phenomenon with the AK4490 (of my RME Adi-2 Pro FS) that might be some form of noise-floor modulation. Between around -30dBFS and -50dBFS I see an increase of harmonics and mirrored harmonics in the upper frequency range. If one were to factor in these in a wideband THD (sans noise) measurement we would have a mid-level THD rise (not unlike the ESS chips) even when the low components stayed completely the same, because at higher levels the HF components get lower and lower in level and get spread out in frequency range, and at lower levels they get shifted higher and higher in frequency, hence making less of a contribution. Actually, only a hint of H2 was visible, and always lower in level than this new dirt.

See attached plot, a loopback measurement (348kHz SR, 512k FFT, 32 averages). Shown is a 300Hz signal at levels from -34dBFS to -42dBFS. Traces have been separated by 10dB for visibility, center trace has no offset. ADC runs at 20dB gain to eliminate its influence (DAC level +24dBu, ADC level +4dBu). ADC is not the culprit, double checked with analog generator (Tek SG505). Things do *not* depend on sample rate. With other rates than 384kHz used here only the right side of the plots is truncated, otherwise no change. Frequency also doesn't matter much (and don't need to be bin-centered. If it's not a bin-center, FFT needs window which raises processing noise floor). With higher freq the spacing gets wider and the onset of the mirrored components is more readily visible (see below). Wider spacing means less total energy, therefore the effect is worse at lower frequencies.

The really interesting thing is that the error spectra get heavily shifted around in frequency, especially the start of the error components, with rather moderate level changes of 2dB.

Also, we can see at -42dBFS the needles are all equally spaced at 300Hz intervals. At the higher levels, as second train of needles appears in the right-hand area of the plots which are 300Hz-spaced, but not on 300Hz multiples -- looks like mirrored at the sampling frequency (clearly visible when zoomed in and/or using higher test freq).

Those tones I saw can actually go way below 20kHz with some combinations of frequency and level, so it doesn't seem AKM did some form of shaping to have those tones strictly above the audio band, or at least it is not fully effective.

Not shown, I made further observations: If the test signal is not a pure sine, components start to reduce quickly. Even when a second tone (of arbitrary frequency, but preferably lower and unrelated) is 60dB down to the main tone, the hash disappears. DC level send to the DAC matters as well (which puts the modulator into a different operating point), as soon as the DC is higher than ~ -30dBFS, the needles go away).

Of course, all the dirt is way below RMS noise floor, also we don't have 100% pure sine content in any music, so it's possibly completely inaudible... but actually we don't know...

(also posted on DIYaudio.com a few days back)

 

[MC_RME]

That 'phenomenon' is not limited to the AK4490, or AKM at all. I found it nearly everywhere. The level to see it varies, and of course it is easier to find with low noise DACs. On most chips noise covers these effects so no one ever noticed them. That said the better the DAC's SNR the easier these tones become visible. Try the AK4499, no averaging needed to see them clearly.

The reason behind it we can only speculate. A request at AKM went unanswered.

 

[Music1969]

Things do *not* depend on sample rate. With other rates than 384kHz used here only the right side of the plots is truncated, otherwise no change.

I assume you meant lower than PCM384kHz. But can you test with PCM768kHz?

See this link, to another link (I don't want to post the same thing in different threads):

https://www.audiosciencereview.com/...osing-a-dac-for-dsd-playback.7701/post-206834

 

[MC_RME]

Sample rate doesn't matter.

 

[KSTR]

That 'phenomen' is not limited to the AK4490, or AKM at all. I found it nearly everyhwere. The level to see it varies, and of course it is easier to find with low noise DACs. On most chips noise covers these effects so no one ever noticed them. That said the better the DAC's SNR the easier these tones become visible. Try the AK4499, no averaging needed to see them clearly.

The reason behind it we can only speculate. A request at AKM went unanswered.

I already suspected this could be a general issue with Delta-Sigma DACs.
I don't have any other quality DACs at home anymore and haven't measured the others (in a TASCAM Us144 MkII and a SMSL 1955+, plus some very old PCI cards for which I'd need to setup a historic PC win Win98 ;-). An during customer measurement I normally don't dig in that deep.

A pity that the new 4499 also shows this, thanks for the info.... which makes me think a way to mitigate this at least partly is summing channels with opposite DC offsets, or maybe just a small DC offset per channel when output coupling capacitors are used.
Someone over at DIYAudio.com com mentioned that in DSD mode the artifact are less pronounced, something I have to check as well.

For 99% of audio measurements those spurs don't matter anyway even though I managed to make them look like a big issue by choosing extreme scales of the plots, but it's not, of course.

 

[KSTR]

I assume you meant lower than PCM384kHz. But can you test with PCM768kHz?

See this link, to another link (I don't want to post the same thing in different threads):

https://www.audiosciencereview.com/...osing-a-dac-for-dsd-playback.7701/post-206834

REW currently only supports 384kHz but I could do 768kHz when recording offline with other tools.
I'm not sure that what is discussed in your link is related to the topic here.

 

[Music1969]

REW currently only supports 384kHz but I could do 768kHz when recording offline with other tools.

Thx. Might be interesting to see. Maybe not.

I'm not sure that what is discussed in your link is related to the topic here.

Better measured performance with ADI-2 with higher PCM sample rate...?

 

[KSTR]

I'll be on vacation the next 10 days (and completey offline to refresh), so things will have to wait...