Measurement: The effect of a slow DAC filter on speaker distortion? (Marantz SR-7013, Revel C426Be)

[Ageve]

The measurements were done in-room. I've measured several times with the same result, and the distortion at 48 kHz was audible as well (weird overtones). I measured at 75dB SPL, 30 cm from the tweeter.

Equipment used:

Receiver: Marantz SR-7013
Power amp: Audiophonics MPA-M400ET (Purifi)
Speaker: Revel C426Be
Mic: MiniDSP UMIK-2
REW V5.30.5 on a Macbook M1 Pro

The amplifier was connected to left front preout (in order to use the left channel in REW).

The only difference between the two measurements were the sample rates used for the sweeps, 48 kHz and 96 kHz. The files were played back from a USB stick (front USB-port on SR-7013), in stereo mode (I tried pure direct as well but it made no difference).

48 kHz. 3.83% THD at 6.5 kHz.

96kHz. 0.078% THD at 6.5 kHz.

The amplifier used is the mono version of HPA-S400ET, reviewed by Amir:

Audiophonics HPA-S400ET Review (Stereo Amplifier)

This is a review and detailed measurements of the Audiophonics HPA-S400ET. It was sent to me by the company after member request and costs 1 490,00 € (US $1,638). I must say this is one of the best packaging of class D amplifiers (Purifi in this case) that I have seen! It is slick and has...

www.audiosciencereview.com

 

[MaxwellsEq]

the distortion at 48 kHz was audible as well (weird overtones)

So it can't be assumed to be an artefact of the UMIK+REW setup if you can hear it! What do you get if you measure at the speaker terminals and at the input to the power amplifier?

It needs further isolation to identify what is happening. If you can hear distortion, it's pretty loud, so it should be possible to isolate the issue. I'd try carefully feeding the output of a laptop into the power amplifier to see if you get the same effect

 

[kemmler3D]

How did you generate the sweeps? This is such a dramatic result that I'd almost assume there was some foldover happening in the software rather than the hardware... can you share the audio files here?

Definitely an interesting result, will be following as we get to the bottom of this one.

 

[Ageve]

So it can't be assumed to be an artefact of the UMIK+REW setup if you can hear it! What do you get if you measure at the speaker terminals and at the input to the power amplifier?

It needs further isolation to identify what is happening. If you can hear distortion, it's pretty loud, so it should be possible to isolate the issue. I'd try carefully feeding the output of a laptop into the power amplifier to see if you get the same effect

Exactly. Playing the 48 kHz sweep from a USB-stick in the Marantz generates the weird overtones that are absent when playing the 96 kHz sweep.

I tried connecting the power amplifier to a Denon AVR-4306, and there was no audible distortion when playing the same 48 kHz sweep. I'll do some measurements (on the Marantz preout as well) later.

How did you generate the sweeps? This is such a dramatic result that I'd almost assume there was some foldover happening in the software rather than the hardware... can you share the audio files here?

Definitely an interesting result, will be following as we get to the bottom of this one.

That's what I thought as well, since I'm using a unconventional* method to record them in REW (in order to play them back from a USB-stick in the Marantz, I'm just hitting start in REW, with the output from the Macbook muted).

To confirm that this wasn't causing the issues, I measured a active Dali Zensor 1 AX (SPDIF), both using a proper "live" sweep on my PC, and using the method above (recording it in REW on my Macbook while playing it back from my PC, to the Dali speaker). The results were nearly identical (low distortion at both 48 kHz and 96 kHz).

I also tried different sample rates for playback and recording (and playing the slightly longer 96kHz sweep while REW was recording at 48kHz to force sync errors (and the other way around), but I was unable to cause any problems. It's seems like REW doesn't care, at least not when it comes to distortion.

* The only reason for using this method, was to make sure that the problem was caused by the Marantz. If it's distorting while playing a file from USB-stick, any problems with the measurement setup can be ruled out (I hope).

I'll upload the sweeps tonight. They were generated in REW.

 

[Sokel]

You can also measure an external source using the "from file" function of REW along with the timing reference.
It make things really easy,it just waits for the reference to start recording.



You only have to generate an IR file sweep with the REW's generator to place in the box.

 

[Ageve]

You can also measure an external source using the "from file" function of REW along with the timing reference.
It make things really easy,it just waits for the reference to start recording.

Thanks. I did the following measurements using that method (75 dB, 30 cm). The sweeps are attached as well.

This time I'm using:

Speaker: Revel M16
Amplifier: Rotel RB-1080 (24 years old)
Mic: MiniDSP UMIK-2
REW V5.30.5 on a Macbook M1 Pro

It's in the same room. I was just too lazy to connect the center speaker as front again.

SR-7013, 96 kHz, USB-stick. THD at 5.88 kHz: 0.083%.

SR-7013, 48 kHz, USB-stick. THD at 5.88 kHz: 3.51%.

Dali Zensor 1 AX playing the same 48 kHz sweep (SPDIF from PC). Recorded on my Macbook, just like above ("from file").

edit: Another measurement using the same 48 kHz sweep (75 dB, 30cm):

Source: Apple TV 4K
Receiver: Denon AVR-4306
Power amp: Rotel RB-1590
Speaker: Revel F208

 

[Sokel]

So,three different speakers,2 different amps (more if we takes the active's internal ones)* and it all comes down to SR-7013, 48 kHz playback that gives such an absurd result.
That's odd.

*Edit:more after the edit

 

[kemmler3D]

Very odd indeed. This marantz receiver is way too recent to have this kind of problem unless (as OP suggests in the title) the DAC filter is simply wrong for lower sampling rates.

@Ageve if you simply record the sweep as audio, and post the audio file for both, we could take a look and see if it's aliasing from the DAC or what.

 

[Sokel]

Very odd indeed. This marantz receiver is way too recent to have this kind of problem unless (as OP suggests in the title) the DAC filter is simply wrong for lower sampling rates.

@Ageve if you simply record the sweep as audio, and post the audio file for both, we could take a look and see if it's aliasing from the DAC or what.

He already posted it in the attachments,above my post,and the result is weird.

 

[Ageve]

Very odd indeed. This marantz receiver is way too recent to have this kind of problem unless (as OP suggests in the title) the DAC filter is simply wrong for lower sampling rates.

It's probably using the same filter as SR-7015, measured by Amir:

Marantz SR7015 8K AVR Review

This is a review and detailed measurements of the Marantz SR7015 "8K" 9.2 channel Home Theater Audio/Video Receiver (AVR). It was purchased new and kindly sent to me by a member. The SR7015 costs US $2,300 from Amazon including Prime shipping. Not much to say about the design of the SR7015 as...

www.audiosciencereview.com

 

[Sokel]

It's probably using the same filter as SR-7015, measured by Amir:

Marantz SR7015 8K AVR Review

This is a review and detailed measurements of the Marantz SR7015 "8K" 9.2 channel Home Theater Audio/Video Receiver (AVR). It was purchased new and kindly sent to me by a member. The SR7015 costs US $2,300 from Amazon including Prime shipping. Not much to say about the design of the SR7015 as...

www.audiosciencereview.com

That explains it:

Very poor choice of filter,indeed.

 

[kemmler3D]

There's definitely something going on in the 48khz sweep that isn't in the 96khz one. Apologies for the janky screenshot, just zoomed in on the apparent problem area. It doesn't quite look like classic foldover to me, but maybe more like high frequency noise or something? Definitely odd.

 

[Ageve]

There's definitely something going on in the 48khz sweep that isn't in the 96khz one. Apologies for the janky screenshot, just zoomed in on the apparent problem area. It doesn't quite look like classic foldover to me, but maybe more like high frequency noise or something? Definitely odd.

It does look a bit odd.

This is the spectral view in Adobe Audition:

48 kHz sweep:

48 kHz sweep when using Denon AVR-4306 (Revel F208), recorded at 96 kHz:

48 kHz sweep when using SR-7013 (Revel M16), recorded at 96 kHz:

 

[Sokel]

It does look a bit odd.

This is the spectral view in Adobe Audition:

48 kHz sweep:
View attachment 348066

48 kHz sweep when using Denon AVR-4306 (Revel F208), recorded at 96 kHz:
View attachment 348064

48 kHz sweep when using SR-7013 (Revel M16), recorded at 96 kHz:
View attachment 348065

Even the timing references are odd.
This registers as broken in my book,Marantz needs to address it.

 

[Ageve]

I did a few recordings today, to find out if the distortion is audible on music, not just on test signals.

I have attached two short clips. They were recorded using:

SR-7013
Audiophonics MPA-M400ET amplifier
Revel C426Be
MiniDSP UMIK-2

The files were played back from a USB-stick in the Marantz. One of them was 44.1kHz, and the other one resampled to 96 kHz.

My results:

foo_abx 2.1 report
foobar2000 v1.6.4
2024-02-08 22:24:13

File A: klipp1.wav
SHA1: e446635ac9f91777461ab9479fb1a5c578c59ee9
File B: klipp2.wav
SHA1: b6ad876cec719afdb3aa2eb846d9006eaf24cd03

Output:
Default : Digital Audio (S/PDIF) (High Definition Audio Device)
Crossfading: NO

22:24:13 : Test started.
22:25:16 : 01/01
22:25:53 : 02/02
22:26:21 : 03/03
22:27:12 : 04/04
22:28:01 : 05/05
22:29:18 : 06/06
22:31:05 : 07/07
22:31:59 : 08/08
22:33:12 : 09/09
22:33:50 : 10/10
22:35:24 : 11/11
22:36:06 : 12/12
22:37:21 : 13/13
22:38:00 : 14/14
22:38:41 : 15/15
22:40:33 : 16/16
22:40:33 : Test finished.

----------
Total: 16/16
p-value: 0 (0%)

-- signature --
effda48a8d9f633b04c5e5e6d6ce3335ad8ec4b9


Recording from 44.1 kHz file:

Recording from 96 kHz file:

edit: Two more files attached. The treble is brighter and more distorted when playing back 44.1 kHz.

 

[kemmler3D]

It does look a bit odd.

This is the spectral view in Adobe Audition:

48 kHz sweep:
View attachment 348066

48 kHz sweep when using Denon AVR-4306 (Revel F208), recorded at 96 kHz:
View attachment 348064

48 kHz sweep when using SR-7013 (Revel M16), recorded at 96 kHz:
View attachment 348065

That last graph appears to show classic harmonic distortion AND foldover. I must have been looking at the wrong file before, I guess maybe it was the raw sweep?

This is simply textbook broken DAC implementation. The lines going downward are aliasing, and we also have pretty dramatic amounts of THD on top of that.

One or the other would be bad, both is unforgivable, if I'm reading this right.

 

[Ageve]

That last graph appears to show classic harmonic distortion AND foldover. I must have been looking at the wrong file before, I guess maybe it was the raw sweep?

This is simply textbook broken DAC implementation. The lines going downward are aliasing, and we also have pretty dramatic amounts of THD on top of that.

One or the other would be bad, both is unforgivable, if I'm reading this right.

Yep, I just couldn't accept the weird measurement results, so I kept on testing, and I've made some progress.

I ended up resetting the receiver to factory defaults, and it got rid of the high THD, but the foldover remains (as expected with the slow DAC filter). The high THD was probably caused by a software bug. All processing (Audyssey, EQ etc) was disabled during all of the tests.

48 kHz before resetting (M16, stereo mode, recorded at 96 kHz):

48 kHz after resetting (M16, stereo mode, recorded at 96 kHz):

96 kHz after resetting (M16, stereo mode, no resampling by SR-7013, recorded at 96 kHz):

48 kHz after resetting (M16, direct mode, recorded at 192 kHz):

96 kHz after resetting (M16, direct mode, no resampling by SR-7013, recorded at 192 kHz):

Denon AVR-4306 as a comparison:

44.1 kHz, Apple TV 4K, F208 (Apple TV resamples everything to 48 kHz, but it uses Core Audio from MacOS, so it's not causing any issues. Recorded at 192 kHz):

48 kHz, Apple TV 4K, AVR-4306, F208 (recorded at 192 kHz):

96 kHz, Apple TV 4K, AVR-4306, F208 (resampled to 48 kHz by Apple TV. Recorded at 192 kHz):

New REW measurements:

48 kHz before resetting SR-7013 (stereo mode, RB-1080, M16):

48 kHz after resetting SR-7013 (stereo mode, RB-1080, M16):

So I guess the problem is solved (the high THD at least). I still don't know why a reset fixed the problem though, so I don't really trust this receiver.

edit: the 3rd harmonic in the midrange is caused by the M16 bass/midrange driver, not the SR-7013.

 

[kemmler3D]

A reset changing THD is pretty much unacceptable already, lol.

And having that much foldover... I didn't quite realize how bad these slow filters really are. Clown town.

Thanks for posting all of this. Very informative and honestly I'd sell this receiver and buy something else as soon as I could.

 

[Ageve]

I did a few preout measurements today.

Soundcard:
Asus Essence ST PCI (ASIO 24-bit 96 kHz)

I used a 96 kHz 0-40000 Hz sweep and downsampled it for the 48 kHz test (to get 20 kHz bandwidth for the distortion measurements). Perhaps there's a better way to do it, but it didn't cause any issues.

The sweeps were played back from a USB stick in SR-7013.

SR-7013, 48kHz, frequency response (-2.5 dB at 20 kHz, caused by the slow filter):

SR-7013, 48 kHz. 14.5% distortion at 15.4 kHz (-16.8 dB). It might look weird, but it matches the ASR review of SR-7015 (1% distortion at ~12 kHz):

SR-7013, 96kHz, frequency response:

SR-7013, 96 kHz. 0.0084% distortion at 15.4 kHz (-81.5 dB):

Soundcard loopback as a reference, measured at 192 kHz. 0.00077% distortion at 15.4 kHz (-102 dB):

Frequency response of soundcard (loopback, 192 kHz, -0.3 dB at 90 kHz):

Recording of the 48 kHz sweep used above (SR-7013):

48 kHz sweep resampled to 96 kHz again, to make sure the resampling to/from 48 kHz wasn't causing any issues.

And finally, here's a short recording (The Shadows - Theme From 'Missing'):

44.1 kHz (Compared to original file in Deltawave):

Resampled to 96 kHz (Compared to original file in Deltawave): ***

Difference between original file and 44.1 kHz playback (Highpass at 1 kHz to filter out small differences in bass response):

Difference between original file and 96 kHz playback (44.1 kHz resampled to 96 kHz. Highpass at 1 kHz to filter out small differences in bass response):

*** The spike is caused by yet another problem with this receiver. The preout starts clipping when the volume reaches 80. I used this volume to ge the best SNR in REW.