Denon AVR-X3800H Review

[peng]

Great. If that’s the case it should be easy to prove versus just theorizing. I don’t want to base my purchases on theories.

Me neither.

 

[JBanville]

Okay I understand your point, but I think such measurements are very helpful to people who trust them. For example, there is no way I can base my choice of an AVR or AVP+Power amp on listening tests because if will be too time consuming even if I can find enough dealers who stock devices on my short list. Going by measurements such as Amir's I would say a few hours or one fully day will allow me to set up a comparison table using a spreadsheet. Then I can shorten my list to few that has SINAD, IMD, SNR/DNR/linearity, cross talk, output level vs SINAD, and the rest of it, that indicate the numbers are good enough for me to the point I wouldn't be able to "hear" the effects of the non SOTA numbers. Listening tests would have been almost useless to me, and I certainly don't want to have to trust someone's ears/brains.

I’m 100% behind measuring. I’m also 100% behind listening tests. My point is they should correlate to one another. If it measures poorly, it should sound worse/poor/etc. If not, then what’s the point of the measurement?

 

[peng]

If it measures poorly, it should sound worse/poor/etc. If not, then what’s the point of the measurement?

That, I have to disagree, but only because "sound worse/poor/etc.," seem subjective so to some, it may sound worse, to others it may sound better. Measurements is not subjective, you see the numbers and you decide what you prefer. That is not to say people should not do listening tests, I certainly do that, far too often actually, wasted lots of time doing them too.

 

[JBanville]

Okay I understand your point, but I think such measurements are very helpful to people who trust them. For example, there is no way I can base my choice of an AVR or AVP+Power amp on listening tests because if will be too time consuming even if I can find enough dealers who stock devices on my short list. Going by measurements such as Amir's I would say a few hours or one fully day will allow me to set up a comparison table using a spreadsheet. Then I can shorten my list to few that has SINAD, IMD, SNR/DNR/linearity, cross talk, output level vs SINAD, and the rest of it, that indicate the numbers are good enough for me to the point I wouldn't be able to "hear" the effects of the non SOTA numbers. Listening tests would have been almost useless to me, and I certainly don't want to have to trust someone's ears/brains.

But if the threshold of the tests is beyond what “you” could actually hear, wether it’s “noise” or distortion, etc, then what’s the relevance of the numbers?

 

[JBanville]

That, I have to disagree, but only because "sound worse/poor/etc.," seem subjective so to some, it may sound worse, to others it may sound better. Measurements is not subjective, you see the numbers and you decide what you prefer.

But the measurements MUST correlate to something we can perceive by listening or else what’s the point of the measurement? That’s my point. If it measures poorly then people should be able to hear this poor performance in a bind test, otherwise, what exactly is the point of the measurement?
It seems to me like a waste of time to review a piece of audio gear that you haven’t listened to first. Make your notes of what you hear, then measure. That way the “numbers” can’t influence what you heard, and then honestly write review explaining what you did or did not hear that correlates to what you measured.

 

[peng]

But the measurements MUST correlate to something we can perceive by listening or else what’s the point of the measurement? That’s my point. If it measures poorly then people should be able to hear this poor performance in a bind test.

There can be a number of points to measure even if the differences may not be discernible to some, or most people. For example:

- Unless you measure, you don't really know if the measurable differences are at least theoretically audible/discernible.
- If say SINAD of the output is >115 dB, then we know it is theoretically not even audible, let alone discernible by humans including those with excellent hearing and has great ability in discerning minute differences that most people don't have, but if SINAD is just >75 dB or even >90 dB, many could hear the noise in a relatively quiet room.
- Also for SINAD of less than say 70 dB, some people may be able to hear the harmonic distortions under some conditions if it is distortions dominated, in that case, one can look into the harmonic profile, such as examine the FFT carefully and if they believe in the 2nd harmonics being responsible for the often described as warm sound, then such measurements can be useful, for such individual's to prepare their short list for listening tests.
- If power amps perform well in power output into 4 ohms, and excel in reactive loads, then again, it helps for someone trying come up with a short list for subsequently tests.

There can be more reasons, probably too many lol, but the main one is really the first one, that is, if you don't measure, you won't even have data to correlate to the things we perceive by listening anyway. Once you see the measurement data, then you have a better chance to eventually make an informed decision, especially after conducting listening tests. If you have unlimited time to find ways to do it all by listening, then okay may be measurement data are not even necessary for you.

 

[JBanville]

There can be a number of points to measure even if the differences may not be discernible to some, or most people. For example:

- Unless you measure, you don't really know if the measurable differences are at least theoretically audible/discernible.
- If say SINAD of the output is >115 dB, then we know it is theoretically not even audible, let alone discernible by humans including those with excellent hearing and has great ability in discerning minute differences that most people don't have, but if SINAD is just >75 dB or even >90 dB, many could hear the noise in a relatively quiet room.
- Also for SINAD of less than say 70 dB, some people may be able to hear the harmonic distortions under some conditions if it is distortions dominated, in that case, one can look into the harmonic profile, such as examine the FFT carefully and if they believe in the 2nd harmonics being responsible for the often described as warm sound, then such measurements can be useful, for such individual's to prepare their short list for listening tests.
- If power amps perform well in power output into 4 ohms, and excel in reactive loads, then again, it helps for someone trying come up with a short list for subsequently tests.

There can be more reasons, probably too many lol, but the main one is really the first one, that is, if you don't measure, you won't even have data to correlate to the things we perceive by listening anyway. Once you see the measurement data, then you have a better chance to eventually make an informed decision, especially after conducting listening tests. If you have unlimited time to find ways to do it all by listening, then okay may be measurement data are not even necessary for you.

My question is, if you (or say a group of average people with average hearing capabilities) can’t perceive the result of a “poor” measurement, what’s the point of the measurement?

 

[JBanville]

There can be a number of points to measure even if the differences may not be discernible to some, or most people. For example:

- Unless you measure, you don't really know if the measurable differences are at least theoretically audible/discernible.
- If say SINAD of the output is >115 dB, then we know it is theoretically not even audible, let alone discernible by humans including those with excellent hearing and has great ability in discerning minute differences that most people don't have, but if SINAD is just >75 dB or even >90 dB, many could hear the noise in a relatively quiet room.
- Also for SINAD of less than say 70 dB, some people may be able to hear the harmonic distortions under some conditions if it is distortions dominated, in that case, one can look into the harmonic profile, such as examine the FFT carefully and if they believe in the 2nd harmonics being responsible for the often described as warm sound, then such measurements can be useful, for such individual's to prepare their short list for listening tests.
- If power amps perform well in power output into 4 ohms, and excel in reactive loads, then again, it helps for someone trying come up with a short list for subsequently tests.

There can be more reasons, probably too many lol, but the main one is really the first one, that is, if you don't measure, you won't even have data to correlate to the things we perceive by listening anyway. Once you see the measurement data, then you have a better chance to eventually make an informed decision, especially after conducting listening tests. If you have unlimited time to find ways to do it all by listening, then okay may be measurement data are not even necessary for you.

I didn’t say don’t measure. I said listen first, make notes of what you hear, and then measure. This way the objective measurements don’t bias your subjective listening. Then explain how what you heard correlates to what you measured. If you didn’t actually hear an issue that involved a “poor” measurement, just say the issue wasn’t audible. Simple.

 

[peng]

My question is, if you (or say a group of average people with average hearing capabilities) can’t perceive the result of a “poor” measurement, what’s the point of the measurement?

I have already answered that question, but that's just my answer. That is, Unless you measure, you don't really know if the measurable differences are at least theoretically audible/discernible, and "if you don't measure, you won't even have data to correlate to the things we perceive by listening anyway".

If you know what you referred to as "poor" measurement, then you have already taken the measurement, otherwise how would you know it was poor, to begin with.

 

[JBanville]

I have already answered that question, but that's just my answer. That is, Unless you measure, you don't really know if the measurable differences are at least theoretically audible/discernible, and "if you don't measure, you won't even have data to correlate to the things we perceive by listening anyway".

If you know what you referred to as "poor" measurement, then you have already taken the measurement, otherwise how would you know it was poor, to begin with.

I guess I didn’t understand your answer. Could you try again please? Maybe something brief.
My question is: what is the *relevance* of a measurement result in a piece of audio gear
that someone has deemed to be “poor”, but that is not perceived by you in a listening test?

 

[JBanville]

This is a review and detailed measurements of the Denon AVR-X3800H 9.4 Home Theater Audio/Video Receiver. It was kindly drop shipped to me by a member and costs US $1,699.
View attachment 239676
Not much to say about the look of the unit as it is pretty much similar/same as last generations. Here is the back panel:
View attachment 239677

I had one operational problem with the unit where the Setup button would not bring up the menus anymore. A power cycle fixed it. I was happy to see solid extruded aluminum heatsink. Despite that, the top front of the of the unit (above the heatsink) got quite hot to the touch. I highly recommend ample airflow in that area. As usual, there is a fan there but it would not come on to cool the unit.

I grabbed a preliminary set of measurements from the DAC section of the 3800H and ran it by the company. Within typical margin of error, the measurements were the same as company's own.

Denon AVR-X3800H DAC Measurements
The DAC measurements and usage are made better by a global "preamplifier" mode which shuts down the amps and keeps them from overloading the unit during testing. I started my tests using Toslink optical to remove any possibility of ground loops:
View attachment 239678

I was fair bit disappointed to see the high distortion spikes which caused the unit to underperform the AVR-X3700H by some 10 dB. This naturally knocks down the positioning of the unit by good bit in our SINAD ranking:
View attachment 239679
The culprit seems to be the optimal output shifting fair bit lower in voltage:
View attachment 239680

With amps shut off, the 3700H was happy to go 2.0 volts.

So fairly large degradation here. Company explanation was that this was good enough for "99%" of the external amps out there. Be that as it may, it still doesn't explain the drop in performance at the peak level (101 dB for 3700 vs 95 for 3800).

Connecting HDMI causes a rise in noise floor but SINAD didn't change since it is dominated by distortion:
View attachment 239682

Fortunately dynamic range is unchanged from last generation:
View attachment 239683

IMD is naturally worse due to earlier saturation:
View attachment 239684

Linearity has suffered a bit:

View attachment 239685

Multitone shows the degradation we would expect:
View attachment 239686

Due to different DAC chip used (?), the filter performance is a bit worse as well:
View attachment 239687


View attachment 239688

This naturally increases the noise+distortion in our wideband measurements:

View attachment 239689

Quite poor result.

Jitter is bad but so was the last generation:
View attachment 239690

Denon AVR-X3800H Amplifier Measurements:
Let's start with analog input with and without pure mode:
View attachment 239691
This is good results.

Crosstalk is very similar to last generation and in need of some improvement:
View attachment 239692

Dashboard performance is a bit worse than last generation but not as bad as the hit in the DAC section:
View attachment 239693

View attachment 239694
HDMI input raises the noise floor:
View attachment 239695

Fortunately noise performance is the same as before:
View attachment 239696
Multitone has power supply noise in lower frequencies and rising distortion in highs:
View attachment 239701

Rest of the measurements are essentially identical to last generation AVR:
View attachment 239697

View attachment 239698

View attachment 239699

View attachment 239700

All in all, the amplifier is as performant as before.

Conclusions
I had high hopes going into this review thinking the company had seen the advantage its superior objective measurements in the past and would try to capitalize on them. Sadly, the reverse seems to be true with the DAC section taking a large step backward. Considering that the 3800 costs $500 more than 3700H, this is very surprising to me. Yes, inflation has a lot to do with that but surely the eye needed to be focused on making sure they at least met the same level of performance as last generation.

The good news is that the amplifier seems to be same design as last generation and has only taken a small hit.

Denon had been my "goto" recommendation for AVRs and even AVPs. When anyone asked me about either, I would just say "get a Denon AVR." While subjectively the performance of this new generation may be similar, I can't accept the regression in objective measured performance.

It is with much sadness that I cannot recommend the Denon AVR-X3800H.

----------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

But how did it sound?

 

[peng]

I guess I didn’t understand your answer. Could you try again please? Maybe something brief.
My question is: what is the *relevance* of a measurement result in a piece of audio gear
that someone has deemed to be “poor”, but that is not perceived by you in a listening test?

Okay, I'll try again..

Of course in that specific scenario there would not likely be relevance except in marginal cases. I guess I didn't understand why you would even ask when the answer would have been so obvious so to play safe I answered the way I did, but that obviously did not work as intended.

The answer unfortunately could not be as brief as you may wish, because it almost always depends.., and I did cited a few examples for when it might still be relevance in some borderline cases. Sorry to say this, but if you need read what I wrote you would understand my point in the first place. I did read your question carefully and I understood your point the first time.

 

[JBanville]

Okay, I'll try again..

Of course in that specific scenario there would not likely be relevance except in marginal cases. I guess I didn't understand why you would even ask when the answer would have been so obvious so to play safe I answered the way I did, but that obviously did not work as intended.

The answer unfortunately could not be as brief as you may wish, because it almost always depends.., and I did cited a few examples for when it might still be relevance in some borderline cases. Sorry to say this, but if you need read what I wrote you would understand my point in the first place. I did read your question carefully and I understood your point the first time.

Sorry, but that didn’t answer my question. It’s a very simple question. If a “poor” measurement is not audible, to you or a random group of average people with average hearing, how is it “marginal”? Either you/they hear it or you don’t.
And if you don’t hear it, why is it even an issue?

 

[G-Can]

I guess I didn’t understand your answer. Could you try again please? Maybe something brief.
My question is: what is the *relevance* of a measurement result in a piece of audio gear
that someone has deemed to be “poor”, but that is not perceived by you in a listening test?

so it took me few days to see what kind of crew amhangs around here. You know an audiophile crew, eccentrics obsessed with testi go and listening gear?

Well this is a free if measuring and testing frequencies.

Same apples/different packaging. Both sides extremists.

Making purchasing decision on measurements is like deciding how a meal tastes based on ingredients used without actually tasting it.

 

[JBanville]

so it took me few days to see what kind of crew amhangs around here. You know an audiophile crew, eccentrics obsessed with testi go and listening gear?

Well this is a free if measuring and testing frequencies.

Same apples/different packaging. Both sides extremists.

Making purchasing decision on measurements is like deciding how a meal tastes based on ingredients used without actually tasting it.

Could you imagine doing speaker “reviews” and not even listening to the speaker, but basing your assessment on measurements?! Crazy.

 

[G-Can]

Could you imagine doing speaker “reviews” and not even listening to the speaker, but basing your assessment on measurements?! Crazy.

well my impression of this forum is that, that's exactly what you should do

Bud, I'll let you know my impressions how it sounds in my environment, hopefully within few days. Waiting to be hooked up.

 

[popej]

Making purchasing decision on measurements is like deciding how a meal tastes based on ingredients used without actually tasting it.

IMHO it is opposite. We know, what ingredients are and we know it taste good. We only check, how diligent was the chief.

 

[JBanville]

IMHO it is opposite. We know, what ingredients are and we know it taste good. We only check, how diligent was the chief.

But if the food is good and you can’t taste anything “off” or wrong, what does it matter if the chef used 1/4 tsp of a spice versus the 2/3 tsp that the recipe called for?
Are you going to say that since the spice quantity wasn’t “right”, you can’t recommend eating the food?

 

[G-Can]

But if the food is good and you can’t taste anything “off” or wrong, what does it matter if the chef used 1/4 tsp of a spice versus the 2/3 tsp that the recipe called for?
Are you going to say that since the spice quantity wasn’t “right”, you can’t recommend eating the food?

That is exactly my approach to hifi world. Above certain price point, all sounds good, it just has different flavours to it

 

[JBanville]

The title of this thread should be changed from “Denon AVR-X3800H review” to “Denon AVR-X3800H measurements”. There was no actual review.