Denon avr-x3900h preamp measurements

[napfkuchen]

I think everyone's waiting for Amir's review of the 3900 with bated breath!

As a (former) owner of the 3700X/3800X, I have absolutely zero desire to upgrade. For that to happen, the 3900X would need to offer extensive new features that make a genuine perceptible difference. The 3800X combined with DiracART and better speakers completely outperforms my previous 3700X setup, despite the slightly lower SINAD.
The only thing of interest (for me) would be Dolby's new Atmos Channel Expander. But nothing is known yet about its performance compared to, say, Auro3D or DTS Neural:X. For stereo music, I mostly use Auro3D because it expands the soundstage (especially vertically) without compromising spatial precision.wah

 

[EWL5]

As a (former) owner of the 3700X/3800X, I have absolutely zero desire to upgrade. For that to happen, the 3900X would need to offer extensive new features that make a genuine perceptible difference. The 3800X combined with DiracART and better speakers completely outperforms my previous 3700X setup, despite the slightly lower SINAD.
The only thing of interest (for me) would be Dolby's new Atmos Channel Expander. But nothing is known yet about its performance compared to, say, Auro3D or DTS Neural:X. For stereo music, I mostly use Auro3D because it expands the soundstage (especially vertically) without compromising spatial precision.wah

+1
Not much to tempt 99% of 3800 owners.

 

[Toni71]

Dolby Atmos Channel Expander. I do wonder if this function will be available on the 'older' flagships. AV10.

 

[Marc v E]

As a (former) owner of the 3700X/3800X, I have absolutely zero desire to upgrade. For that to happen, the 3900X would need to offer extensive new features that make a genuine perceptible difference. The 3800X combined with DiracART and better speakers completely outperforms my previous 3700X setup, despite the slightly lower SINAD.
The only thing of interest (for me) would be Dolby's new Atmos Channel Expander. But nothing is known yet about its performance compared to, say, Auro3D or DTS Neural:X. For stereo music, I mostly use Auro3D because it expands the soundstage (especially vertically) without compromising spatial precision.wah

Thank you for posting this.
It's very helpful, especially for people new to measurements of equipment like dacs, avrs and amplifiers.

Adding a sub to a system and having a dsp to take care of high/low passing plus equalisation makes such a huge difference. That's where the great gains are ime.

Nothing wrong with choosing the best sinad of a device you need. But first we need to know what kind of device we need to get best results imo.

 

[peng]

+1
Not much to tempt 99% of 3800 owners.

I would guess 99.5%, because it is hardly an upgrade anyway. Going to the 6900 would be an upgrade, or even the 4900 might be considered an upgrade, though not as clear cut.

 

[Marc v E]

I would guess 99.5%, because it is hardly an upgrade anyway. Going to the 6900 would be an upgrade, or even the 4900 might be considered an upgrade, though not as clear cut.

Now that would be an interesting test.
My educated guess is we wouldn't be able to tell them apart, in a blind test.
Adding more subs than 1, adding dirac, art yes, probably.
But upgrading an avr, I highly doubt I could hear a difference.

(Upgrading for more or better connections, having 4 sub outs in stead of one or two, a stable amplifier under load etc does make sense to me.)

 

[peng]

Now that would be an interesting test.
My educated guess is we wouldn't be able to tell them apart, in a blind test.

That is highly likely. One would do well if one considers upgrade when they want the added features, functionality, aesthetic, and/or better audio specifications, build quality. If the goal is mainly for audible sound quality improvements, especially if/when DSP are not in the equation, one should do their research well prior.

 

[alex.schroedsen]

New measurements based on the feedback I received here.

Thank you @NTK for the heads up on the FFT window. I was not able to find the setting within Rightmark to make that change, but I will keep working on it.

Thank you also to @VintageFlanker who tipped me off on the mono mode setting. I previously had the volume on the Cosmos ADCiso set to 100 percent. The ADCiso does not use that volume control conventionally. The ADCiso switches to stereo mode in the bottom 50 percent of volume control and mono in the top 50 percent. Thanks to @staticV3 for info related to ADCiso mono-vs-stereo modes in another thread.

Thank you to everyone for your thanks, feedback and patience with my first thread! Your tips are helping me to improve any content I post in the future.

My settings in FlexASIO have also been updated to move to a WDM-KS Kernel Streaming backend. This should guarantee bit-perfect signal path from RightMark to the HDMI driver for streaming to the Denon x3900h and bit perfect signal path from the ADCiso back to Rightmark. 32-bit signal path has also been utilized from the ADCiso into Rightmark, even though all the specs fit well within a 24-bit envelope, the results when using 32-bit input were a few decibels improved over 24. The ADCiso is using its Linear phase Apodizing input filter. I also tried the Minimum Phase filter and there was no noticeable difference.

​

​

​

RightMark Audio Analyzer test report (48 kHz input)​

​

​

​

Testing device	Denon x3900h preamp full range xover 82.5v 3248 r2
Sampling mode	32-bit, 48 kHz
Interface
Testing chain	External loopback (line-out - line-in)
RMAA Version	6.4.5
20 Hz - 20 kHz filter	ON
Normalize amplitude	ON
Level change	-12.0 dB / -11.9 dB
Mono mode	OFF
Calibration singal, Hz	1000
Polarity	inverted/inverted

​

Summary​

​

​

​

​

Frequency response (from 40 Hz to 15 kHz), dB	+0.04, -0.23​	Very good​
Noise level, dB (A)	-113.1​	Excellent​
Dynamic range, dB (A)	112.6​	Excellent​
THD, %	0.00087​	Excellent​
THD + Noise, dB (A)	-84.4​	Good​
IMD + Noise, %	0.015​	Very good​
Stereo crosstalk, dB	-85.9​	Excellent​
IMD at 10 kHz, %	0.00111​	Excellent​
General performance	​	Very good​

​

Frequency response (48 kHz input)​

​

​

​

​

​

Left​	Right​
From 20 Hz to 20 kHz, dB	-0.85, +0.04​	-0.72, +0.16​
From 40 Hz to 15 kHz, dB	-0.23, +0.04​	-0.10, +0.16​

​

Noise level (48 kHz input)​

​

​

​

​

​

Left​	Right​
RMS power, dB	-112.5​	-111.9​
RMS power (A-weighted), dB	-113.4​	-112.8​
Peak level, dB FS	-61.7​	-61.6​
DC offset, %	+0.0​	-0.0​

​

Dynamic range (48 kHz input)​

​

​

​

​

​

Left​	Right​
Dynamic range, dB	+112.3​	+111.7​
Dynamic range (A-weighted), dB	+112.8​	+112.3​
DC offset, %	-0.00​	-0.01​

​

THD + Noise (at -3 dB FS) (48 kHz input)​

​

​

​

​

​

Left​	Right​
THD, %	0.00092​	0.00081​
THD + Noise, %	0.00546​	0.00545​
THD + Noise (A-weighted), %	0.00606​	0.00603​

​

Intermodulation distortion (48 kHz input)​

​

​

​

​

​

Left​	Right​
IMD + Noise, %	0.01451​	0.01450​
IMD + Noise (A-weighted), %	0.00383​	0.00382​

​

Stereo crosstalk (48 kHz input)​

​

​

​

​

​

Left​	Right​
Crosstalk at 100 Hz, dB	-63​	-103​
Crosstalk at 1000 Hz, dB	-70​	-100​
Crosstalk at 10000 Hz, dB	-70​	-89​

​

IMD (swept tones) (48 kHz input)​

​

​

​

​

​

Left​	Right​
IMD + Noise at 5000 Hz,	0.00134​	0.00123​
IMD + Noise at 10000 Hz,	0.00085​	0.00090​
IMD + Noise at 15000 Hz,	0.00111​	0.00120​

RightMark Audio Analyzer test report (44.1 kHz input)​

​

​

​

​

​

​

​

Testing device	Denon x3900h preamp full range xover 82_5v 3244 r3
Sampling mode	32-bit, 44 kHz
Interface
Testing chain	External loopback (line-out - line-in)
RMAA Version	6.4.5
20 Hz - 20 kHz filter	ON
Normalize amplitude	ON
Level change	-12.0 dB / -11.9 dB
Mono mode	OFF
Calibration singal, Hz	1000
Polarity	inverted/inverted

​

Summary​

​

​

​

​

​

​

​

Frequency response (from 40 Hz to 15 kHz), dB	+0.92, -3.41​	Poor​
Noise level, dB (A)	-112.6​	Excellent​
Dynamic range, dB (A)	104.5​	Excellent​
THD, %	0.00086​	Excellent​
THD + Noise, dB (A)	-54.6​	Poor​
IMD + Noise, %	0.298​	Average​
Stereo crosstalk, dB	-92.3​	Excellent​
IMD at 10 kHz, %	0.00168​	Excellent​
General performance	​	Very good​

​

Frequency response (44.1 kHz input) (yes, this is correct, reproduced consistently 3 times)​

​

​

​

​

​

Left​	Right​
From 20 Hz to 20 kHz, dB	-10.70, +0.92​	-10.57, +1.05​
From 40 Hz to 15 kHz, dB	-3.41, +0.92​	-3.28, +1.05​

​

Noise level (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
RMS power, dB	-111.9​	-111.4​
RMS power (A-weighted), dB	-112.8​	-112.4​
Peak level, dB FS	-61.4​	-61.3​
DC offset, %	+0.0​	-0.0​

​

Dynamic range (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
Dynamic range, dB	+105.2​	+105.0​
Dynamic range (A-weighted), dB	+104.6​	+104.4​
DC offset, %	+0.00​	-0.01​

​

THD + Noise (at -3 dB FS) (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
THD, %	0.00092​	0.00080​
THD + Noise, %	0.16920​	0.16920​
THD + Noise (A-weighted), %	0.18718​	0.18718​

​

Intermodulation distortion (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
IMD + Noise, %	0.29849​	0.29847​
IMD + Noise (A-weighted), %	0.07633​	0.07630​

​

Stereo crosstalk (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
Crosstalk at 100 Hz, dB	-75​	-107​
Crosstalk at 1000 Hz, dB	-83​	-100​
Crosstalk at 10000 Hz, dB	-82​	-89​

​

IMD (swept tones) (44.1 kHz input)​

​

​

​

​

​

Left​	Right​
IMD + Noise at 5000 Hz,	0.00138​	0.00126​
IMD + Noise at 10000 Hz,	0.00096​	0.00096​
IMD + Noise at 15000 Hz,	0.00275​	0.00280​

​

 

[peng]

New measurements based on the feedback I received here.

Thank you @NTK for the heads up on the FFT window. I was not able to find the setting within Rightmark to make that change, but I will keep working on it.

Thank you also to @VintageFlanker who tipped me off on the mono mode setting. I previously had the volume on the Cosmos ADCiso set to 100 percent. The ADCiso does not use that volume control conventionally. The ADCiso switches to stereo mode in the bottom 50 percent of volume control and mono in the top 50 percent. Thanks to @staticV3 for info related to ADCiso mono-vs-stereo modes in another thread.

Thank you to everyone for your thanks, feedback and patience with my first thread! Your tips are helping me to improve any content I post in the future.

My settings in FlexASIO have also been updated to move to a WDM-KS Kernel Streaming backend. This should guarantee bit-perfect signal path from RightMark to the HDMI driver for streaming to the Denon x3900h and bit perfect signal path from the ADCiso back to Rightmark. 32-bit signal path has also been utilized from the ADCiso into Rightmark, even though all the specs fit well within a 24-bit envelope, the results when using 32-bit input were a few decibels improved over 24. The ADCiso is using its Linear phase Apodizing input filter. I also tried the Minimum Phase filter and there was no noticeable difference.

View attachment 537527

RightMark Audio Analyzer test report​

​

​

​

​

​

Testing device	Denon x3900h preamp full range xover 82.5v 3248 r2
Sampling mode	32-bit, 48 kHz
Interface
Testing chain	External loopback (line-out - line-in)
RMAA Version	6.4.5
20 Hz - 20 kHz filter	ON
Normalize amplitude	ON
Level change	-12.0 dB / -11.9 dB
Mono mode	OFF
Calibration singal, Hz	1000
Polarity	inverted/inverted

​

Summary​

​

​

​

​

​

Frequency response (from 40 Hz to 15 kHz), dB	+0.04, -0.23​	Very good​
Noise level, dB (A)	-113.1​	Excellent​
Dynamic range, dB (A)	112.6​	Excellent​
THD, %	0.00087​	Excellent​
THD + Noise, dB (A)	-84.4​	Good​
IMD + Noise, %	0.015​	Very good​
Stereo crosstalk, dB	-85.9​	Excellent​
IMD at 10 kHz, %	0.00111​	Excellent​
General performance	​	Very good​

​

Frequency response​

​

​

​

​

View attachment 537530
​

	Left​	Right​
From 20 Hz to 20 kHz, dB	-0.85, +0.04​	-0.72, +0.16​
From 40 Hz to 15 kHz, dB	-0.23, +0.04​	-0.10, +0.16​

​

Noise level​

​

​

​

​

View attachment 537531
​

	Left​	Right​
RMS power, dB	-112.5​	-111.9​
RMS power (A-weighted), dB	-113.4​	-112.8​
Peak level, dB FS	-61.7​	-61.6​
DC offset, %	+0.0​	-0.0​

​

Dynamic range​

​

​

​

​

View attachment 537532
​

	Left​	Right​
Dynamic range, dB	+112.3​	+111.7​
Dynamic range (A-weighted), dB	+112.8​	+112.3​
DC offset, %	-0.00​	-0.01​

​

THD + Noise (at -3 dB FS)​

​

​

​

​

View attachment 537533
​

	Left​	Right​
THD, %	0.00092​	0.00081​
THD + Noise, %	0.00546​	0.00545​
THD + Noise (A-weighted), %	0.00606​	0.00603​

​

Intermodulation distortion​

​

​

​

​

View attachment 537534
​

	Left​	Right​
IMD + Noise, %	0.01451​	0.01450​
IMD + Noise (A-weighted), %	0.00383​	0.00382​

​

Stereo crosstalk​

​

​

​

​

View attachment 537535
​

	Left​	Right​
Crosstalk at 100 Hz, dB	-63​	-103​
Crosstalk at 1000 Hz, dB	-70​	-100​
Crosstalk at 10000 Hz, dB	-70​	-89​

​

IMD (swept tones)​

​

​

​

​

View attachment 537536
​

	Left​	Right​
IMD + Noise at 5000 Hz,	0.00134​	0.00123​
IMD + Noise at 10000 Hz,	0.00085​	0.00090​
IMD + Noise at 15000 Hz,	0.00111​	0.00120​

​

RightMark Audio Analyzer test report​

​

​

​

​

​

Testing device	Denon x3900h preamp full range xover 82_5v 3244 r3
Sampling mode	32-bit, 44 kHz
Interface
Testing chain	External loopback (line-out - line-in)
RMAA Version	6.4.5
20 Hz - 20 kHz filter	ON
Normalize amplitude	ON
Level change	-12.0 dB / -11.9 dB
Mono mode	OFF
Calibration singal, Hz	1000
Polarity	inverted/inverted

​

Summary​

​

​

​

​

​

Frequency response (from 40 Hz to 15 kHz), dB	+0.92, -3.41​	Poor​
Noise level, dB (A)	-112.6​	Excellent​
Dynamic range, dB (A)	104.5​	Excellent​
THD, %	0.00086​	Excellent​
THD + Noise, dB (A)	-54.6​	Poor​
IMD + Noise, %	0.298​	Average​
Stereo crosstalk, dB	-92.3​	Excellent​
IMD at 10 kHz, %	0.00168​	Excellent​
General performance	​	Very good​

​

Frequency response (refer to the first post for 44.1khz frequency response, we'll just consider this an example of the occasional odd performance of this unit)​

​

​

View attachment 537537​

​

	Left​	Right​
From 20 Hz to 20 kHz, dB	-10.70, +0.92​	-10.57, +1.05​
From 40 Hz to 15 kHz, dB	-3.41, +0.92​	-3.28, +1.05​

​

Noise level​

​

​

​

​

View attachment 537538
​

	Left​	Right​
RMS power, dB	-111.9​	-111.4​
RMS power (A-weighted), dB	-112.8​	-112.4​
Peak level, dB FS	-61.4​	-61.3​
DC offset, %	+0.0​	-0.0​

​

Dynamic range​

​

​

​

​

View attachment 537539
​

	Left​	Right​
Dynamic range, dB	+105.2​	+105.0​
Dynamic range (A-weighted), dB	+104.6​	+104.4​
DC offset, %	+0.00​	-0.01​

​

THD + Noise (at -3 dB FS)​

​

​

​

​

View attachment 537540
​

	Left​	Right​
THD, %	0.00092​	0.00080​
THD + Noise, %	0.16920​	0.16920​
THD + Noise (A-weighted), %	0.18718​	0.18718​

​

Intermodulation distortion​

​

​

​

​

View attachment 537541
​

	Left​	Right​
IMD + Noise, %	0.29849​	0.29847​
IMD + Noise (A-weighted), %	0.07633​	0.07630​

​

Stereo crosstalk​

​

​

​

​

View attachment 537542
​

	Left​	Right​
Crosstalk at 100 Hz, dB	-75​	-107​
Crosstalk at 1000 Hz, dB	-83​	-100​
Crosstalk at 10000 Hz, dB	-82​	-89​

​

IMD (swept tones)​

​

​

​

​

View attachment 537543
​

	Left​	Right​
IMD + Noise at 5000 Hz,	0.00138​	0.00126​
IMD + Noise at 10000 Hz,	0.00096​	0.00096​
IMD + Noise at 15000 Hz,	0.00275​	0.00280​

​

Can you explain why THD and noise are good yet THD+N is so poor?

 

[alex.schroedsen]

Can you explain why THD and noise are good yet THD+N is so poor?

That is a curiosity I have, but cannot explain. I know that the test runs multiple different signals one after the other. It could be that the segment of the test had some soft of odd buffering issue, like I was experiencing when trying to use WASAPI exclusive.

I ran these tests dozens of times this morning and could never get a THD+N beyond about -85dB, but as you mentioned, the stats are good individually. The 44.1khz THD+N was in the mid 50s. The 44.1khz tests had the same behavior, with THD looking good and Noise looking good as well. I'll do another round of tests, probably next weekend.

[EDIT] I ran the test three more times tonight and the funky frequency response graph printed the same three times in a row. The 44.1khz measurements have have proven reproducible.

 

[NTK]

Do you have another device you can measure? Those FFT spectra look strange, but it is possible that they were the actual outputs of the X3900H. It would be nice to check another device to see if those strange looking spectra stay.

 

[alex.schroedsen]

Do you have another device you can measure? Those FFT spectra look strange, but it is possible that they were the actual outputs of the X3900H. It would be nice to check another device to see if those strange looking spectra stay.

To test this theory, I ran a test on a Matrix Audio Mini-i Pro 2 (single ended output). The odd spectrum spread seems to be a product of the Denon x3900h indeed. You can see the results on the Matrix Mini-i Pro 2 below​

​

​

First, the FlexASIO settings for this test.​

RightMark Audio Analyzer test report (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

Testing device	Matrix Mini-i Pro 2
Sampling mode	32-bit, 48 kHz
Interface
Testing chain	External loopback (line-out - line-in)
RMAA Version	6.4.5
20 Hz - 20 kHz filter	ON
Normalize amplitude	ON
Level change	-1.6 dB / -1.6 dB
Mono mode	OFF
Calibration singal, Hz	1000
Polarity	correct/correct

​

Summary (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Frequency response (from 40 Hz to 15 kHz), dB	+0.06, +0.02​	Excellent​
Noise level, dB (A)	-122.5​	Excellent​
Dynamic range, dB (A)	122.2​	Excellent​
THD, %	0.00054​	Excellent​
THD + Noise, dB (A)	-103.5​	Excellent​
IMD + Noise, %	0.00111​	Excellent​
Stereo crosstalk, dB	-116.4​	Excellent​
IMD at 10 kHz, %	0.00219​	Excellent​
General performance	​	Very good​

​

Frequency response (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
From 20 Hz to 20 kHz, dB	-0.07, +0.03​	-0.03, +0.07​
From 40 Hz to 15 kHz, dB	-0.03, +0.02​	+0.02, +0.06​

​

Noise level (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
RMS power, dB	-118.3​	-120.1​
RMS power (A-weighted), dB	-122.4​	-122.6​
Peak level, dB FS	-84.5​	-84.4​
DC offset, %	+0.0​	-0.0​

​

Dynamic range (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
Dynamic range, dB	+118.5​	+120.5​
Dynamic range (A-weighted), dB	+122.1​	+122.3​
DC offset, %	+0.00​	-0.00​

​

THD + Noise (at -3 dB FS) (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
THD, %	0.00054​	0.00054​
THD + Noise, %	0.00068​	0.00065​
THD + Noise (A-weighted), %	0.00067​	0.00066​

​

Intermodulation distortion (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
IMD + Noise, %	0.00114​	0.00109​
IMD + Noise (A-weighted), %	0.00080​	0.00079​

​

Stereo crosstalk (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
Crosstalk at 100 Hz, dB	-113​	-114​
Crosstalk at 1000 Hz, dB	-119​	-112​
Crosstalk at 10000 Hz, dB	-110​	-110​

​

IMD (swept tones) (Matrix Mini-i Pro 2 CONTROL MEASUREMENT, NOT DENON x3900h)​

​

​

​

​

​

​

Left​	Right​
IMD + Noise at 5000 Hz,	0.00240​	0.00243​
IMD + Noise at 10000 Hz,	0.00158​	0.00159​
IMD + Noise at 15000 Hz,	0.00259​	0.00257​

​

 

[NTK]

To test this theory, I ran a test on a Matrix Audio Mini-i Pro 2 (single ended output). The odd spectrum spread seems to be a product of the Denon x3900h indeed. You can see the results on the Matrix Mini-i Pro 2 below​

​

Thanks! You are awesome! It seems Denon might have a broken X3900H.

 

[alex.schroedsen]

Thanks for the heads up. I'll try to find a solution tomorrow.

Hello, I have not yet found a solution to readability in dark mode in this site. The main idea I've had is that I have attached the Rightmark sav files to the post. You can download and view the results in Rightmark Audio Analyzer. I have also uploaded updated measurements, implementing the advice members here have given me. I can attach PDF files that are rendered in dark mode if that would be helpful.

 

[EWL5]

Thanks! You are awesome! It seems Denon might have a broken X3900H.

I will not discount the possibility that some OEMs may present a product that will do well for certain tests they know are coming. Let's wait for Amir's testing to make that call as he is a seasoned tester.

Even if the outcome is the same SINAD as the 3800, we can all be disappointed but that means the 3800 becomes even more of a bargain these days than ever before (grab them while you can)!

 

[alex.schroedsen]

When Amir review/measure the 3900, I hope he will include the one that he typically (almost always but not every time) would do, that is the SINAD vs frequency.
The AVR-X4800H and Cinema 40 did very bad relative to the AVR-X3700H, X3600H that use the AK4458 DAC IC, the rest of the parts such as the volume IC, opamps remain the same so the DAC is the one that really made a big difference, though as mentioned many time by other members, audibility may not be an issue.

I am using the Cinema 40 for reference, the numbers/graphs are almost the same as the AVR-X4800H but it costs a lot more and is clearly better build so imo it is a better reference:

View attachment 537241

For comparison, here's the AVR-X3700 that uses the AK4458 DAC chip:

View attachment 537242

We can see the significant deterioration going from the AVR-X3700H to the Cinema 40 in SINAD even in test frequencies at below 2,000 Hz, comparing the red curves.

Again, hopefully the AVR-X3900, Cinema 51/41 (the model numbers are assumed as nothing has been announced yet), will bring the SINAD vs Frequency back to where the Denon 3000/4000 AVRs performance level.

Hi Peng, thanks for digging into this and helping with figuring things out. I had the ADC in MONO mode, which was causing the horrible stereo crosstalk. The initial measurement also had WASAPI automatic resampling enabled. This made the 44.1khz frequency response look waaay better than it is when the receiver is fed 44.1khz audio without first resampling to 48khz (I am guessing windows was resampling to 48khz). We also found out that the odd frequency spread in the THD+N graphs is generated by the Denon x3900h. The Matrix Mini-i Pro 2 DAC produced a more flat THD+N graph when measured in the same system with the same software setup. Thanks again for bringing your knowledge and experience. I'm having a blast learning here.

 

[alex.schroedsen]

I will not discount the possibility that some OEMs may present a product that will do well for certain tests they know are coming. Let's wait for Amir's testing to make that call as he is a seasoned tester.

Even if the outcome is the same SINAD as the 3800, we can all be disappointed but that means the 3800 becomes even more of a bargain these days than ever before (grab them while you can)!

I should have mentioned that the Denon x3900h has a quite low voltage output. With the Cosmos ADCiso set to 1.77 volt input mode, the test signal was still about -12dB in volume. This is at 82.5 volume level. As soon as you exceed volume level 85 or so, the noise graph goes wild with spikes across the 20-20k range, when you get up to volume 95, some of the noise spikes are as high as -18dB. The optimal volume level is definitely the 82.5 setting that Amir has used in his previous Denon reviews. With the odd behavior at 44.1khz in frequency response, this is definitely a device designed with 48khz operation in mind.

 

[peng]

Thanks! You are awesome! It seems Denon might have a broken X3900H.

it still did not explain why the unit had much better THD and N than THD+N. No doubt SINAD should be lower but shouldn’t be that much lower when the THD and N individually are so good. No previously measured Denon had SINAD that low, and if this one is broken then why only broken in SINAD?

 

[alex.schroedsen]

Thanks! You are awesome! It seems Denon might have a broken X3900H.

You're welcome! Thank you, I am glad to help solve this mystery before we enter the busy work week. Thanks for bringing the ideas to test. The spectrum spread is a bit foreboding for the x3900h. Given this discovery, I'm definitely going to test my PCM5102-bearing Denon x3700h to see how it compares.

 

[alex.schroedsen]

it still did not explain why the unit had much better THD and N than THD+N. No doubt SINAD should be lower but shouldn’t be that much lower when the THD and N individually are so good. No previously measured Denon had SINAD that low, and if this one is broken then why only broken in SINAD?

I am curious about this. I do notice that the THD+N on the control sample of the Matrix Mini-i Pro 2 is also subject to the better individual THD and N, but a lower combined THD+N (THD is about -105dB, noise is -122 and THD+N is -103dB, so not nearly as pronounced). I'll see if I can find another audio measurement software (I'm interested in any suggestions you may have) and also see if there is a problem with the impedance being low as mentioned in Amir's review of the E1DA Cosmos ADCiso.