1 Topic by pinox67 (edited by pinox67 2023-11-07 22:04:14)

  • From: Italy
  • Registered: 2019-01-31
  • Posts: 14

Topic: Strange Fequency Response

I own the RME ADI-2 Pro FS, which I have been using with satisfaction for several years to perform amplifiers or environmental measurements with REW and a MacBook Pro (Intel). Well, doing the usual calibration procedure I detected a strange behaviour. With these settings (screenshots of ADI2 Remote), Loopback set:

https://drive.google.com/file/d/1wwFo9H … sp=sharing
https://drive.google.com/file/d/1jTql1l … sp=sharing
https://drive.google.com/file/d/1tT7mTH … sp=sharing

I obtain, doing an usual measurement with a sweep with REW, this Frequency Response (384KHz sample rate):

https://drive.google.com/file/d/18gs3ON … sp=sharing

In purple the magnitude; in red the phase.
Even using other lower sample rates, without using Loopback, other Levels or Filter Types the result is the same.

Is my unit damaged?

2 Reply by KaiS (edited by KaiS 2023-11-06 10:44:54)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,127

Re: Strange Fequency Response

pinox67 wrote:

…I obtain, doing an usual measurement with a sweep with REW, this Frequency Response (384KHz sample rate):

https://drive.google.com/file/d/18gs3ON … sp=sharing

In purple the magnitude; in red the phase.

Something’s wrong with your measurement setup.

With the frequency response you get, the phase response can‘t be just a straight line.

• Sweep speed too fast, so values don’t settle?
• Round trip delay compensation set wrong?
• Accidental loopback PC input to output?
• Some sample rate conversion active?
• Sample rate / clock send-return mismatch?

KaiS's Website

3 Reply by pinox67 (edited by pinox67 2023-11-08 08:03:12)

  • From: Italy
  • Registered: 2019-01-31
  • Posts: 14

Re: Strange Fequency Response

Thanks for the hints.

Well, I created a physical loopback with a XLR cable, and tried different sweeps (1M, 2M or 4M samples), different sample rate (192KHz, 96KHz, 44.1KHz) and different I/O levels; No Time Ref., Set t=0 at IR peak, 1 measure, No calibration file. The result remains more or less the same:

https://drive.google.com/file/d/1X6Nm83 … sp=sharing

The phase here is not flat, as expected, but the magnitude (in green) has strange oscillations.
The output level is set to -1.8dBFS to avoid clipping, but the input level is -4dBFS (strange, it should be -1.8).

I measured more times the output level by XLR and TRS connections for a sine at 1KHz, 0dBFS; the strange results are:

                     XLR+/-   TRS
I/O +4 dBu:  1.2 V      1.2 V
I/O +13 dBu: 3.4 V     3.4 V
I/O +19 dBu: 6.9 V     6.9 V
I/O +24 dBu: 12.3 V   6.9 V

Between 19 and 24 dBu there is no switch using the TRS connection.
I can't understand if there are other external factors that can affect these results.

4 Reply by KaiS (edited by KaiS 2023-11-08 00:47:35)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,127

Re: Strange Fequency Response

At least the XLR vs TRS level has a simple explanation:

TRS can’t go higher than +19 dBu.
TRS is impedance balanced, meaning only the (+)/Tip is active, the (–)/Ring is passive with matching resistor to ground.

Therefore TRS doesn’t do the last Reference Level step to +24 dB.



The frequency response looks like some sample rate conversion is going on, with aliasing above 20 kHz.
Maybe the Mac’s system sound is doing it.
I’m not on Mac, but isn’t there a setting like Window’s “Exclusive Mode” available that disables SRC?

• Could you verify the frequency response with steady state sine waves, 1 kHz, 3 kHz, 10 kHz, 25 kHz and 30 kHz?

• If the result is consistent, maybe just create a calibration file..

• Here’s a VERY comprehensive page for REW setup:
https://www.roomeqwizard.com/help/help_ … dcard.html

KaiS's Website

5 Reply by pinox67 (edited by pinox67 2023-11-10 18:39:23)

  • From: Italy
  • Registered: 2019-01-31
  • Posts: 14

Re: Strange Fequency Response

I forgot the limitation of TRS output to 19 dBu, so it is ok for this aspect.

As suggested, I've checked the levels of a pure sine in steady-state.
With loopback XLR connection, 24 dBu I/O levels, 192KHz sample rate, sine peak at -1 dBFS, and without a calibration file, the  (strange) levels are the same of those detected by the swipe (the 1 dBFS of difference is because the level was set -2 dBFS):

30KHz: -2.86 dBFS
25KHz: -3.60 dBFS
10KHz: -3.06 dBFS
3KHz:  -3.30 dBFS
1KHz:  -3.24 dBFS

If I run the calibration, the result is the following:

https://drive.google.com/file/d/1N0kueE … sp=sharing

Redoing the measure of Frequency Response with this calibration file, it changes only above 30KHz (flat instead of being decreasing).

Changing the speed/length of the sweep (10Hz-90KHz), the result is more or less the same. However, when the sweep is with 1M samples (or longer), in the ADI2 display there is the message "DC detected on Line Out".

In the Mac, the only settings are about the sample rate in the Audio Midi Setup, that shall match that in REW.
I still can't figure out where the problem is.

6 Reply by KaiS (edited by KaiS 2023-11-11 08:53:28)

  • KaiS
  • DAW Mastermind
  • Offline
  • From: Germany
  • Registered: 2020-02-05
  • Posts: 3,127

Re: Strange Fequency Response

The message "DC detected on Line Out" and therefore DC protection can be disabled:
I/O / Line Out / Settings / Dig. DC Protection: OFF



Interesting that REW’s calibration doesn’t do perfect linearization for the whole range beyond 20 Hz - 20 kHz, that’s unusual.
Maybe some manual editing of the calibration file needs to be done.
Typically not too complicated.


Currently I have no measurement soft prepared to measure outside the 5 Hz to 24 kHz range (SR 48 kHz) and no time to dig deeper.
So I can’t counter-check your results, sorry.


In the manual page 85 the target curves for AD-DA at various sample rates are shown, specifically the 48 kHz curve shows some minor ripple, as far as visible in the scale and with the multitude of curves overlaid.
@MC could supply plots with higher resolution maybe.

So I don’t think your ADI-2 is broken.
To be sure, do a full factory reset (see below).


Having to calibrate an analog or digital measurement instrument is standard.
Finally, once calibrated, you have a powerful measurement rig, so …



Reset:
While powering up, push and hold the VOL button (NOT the VOLUME dial) and Encoder 1, until you see the message: “Reset done“.
User Setups and EQ Presets are not deleted.

Or a full factory reset:
While powering up, push and hold the VOL button (NOT the VOLUME dial) and Encoder 1 and Encoder 2, until you see the message: “Reset done“.
All user Setups and EQ Presets are deleted.
Take notes before, or store them through the Remote App, if you want to keep some.

KaiS's Website

7 Reply by pinox67 (edited by pinox67 2023-11-22 23:18:56)

  • From: Italy
  • Registered: 2019-01-31
  • Posts: 14

Re: Strange Fequency Response

As suggested, I've done the full factory reset.
In the past I've done hundreds of measurements with ADI2, environmental (with QuadMic) and of preamps and power amps.

Talking also with J. Mulcahy (REW's author) I discovered that in the Pref. panel it was set (hidden) the .cal file of the mic. My mistake!
So, the strange frequency response above are due to the (unwanted) calibration of the mic.
Now the response, without any .cal, with XLR loopback, seems correct, except when the sample frequency is 352KHz:

https://drive.google.com/file/d/1rd9TVb … sp=sharing

There are oscillation above 80KHz (filter is Sharp for I/O). At 192KHz or 768KHz it seems instead correct.

Moreover, the "DC Detected in Line Out" alert is still shown in the ADI2 display (in any sample rate), even if it is set to OFF/None for both I/O.
The curious thing is that also doing the "Check levels..." in REW Preference panel, that uses a noise signal, this alert still appears. If I enable DC Filter and DC protection, the ADI2 deactivates the Line (still in Loopback). Still strange...

8 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,849

Re: Strange Fequency Response

There is no Mute with 'Filter' set.

352 kHz FR: measurements like this (log swept sine?) are always an error in the measurement software. You can have such results even with a 30.000 USD Audio Precision system, when the parameters for the sensitve test signal are not set correctly. This never happens using the old (but time consuming) stepped sine sweep.

The DC warning most probably is right. Noise not bandwidth limited includes heavy rumble above 0 Hz. You could then disable the DC protection, use lower levels to measure (for FR measure at -10 dBFS), or generate the test signal with corner frequency of at least 10 Hz.

Regards
Matthias Carstens
RME

MC's Website