Denon AVR-X3800H Review

[oivavoi]

What IF... Robinson doesn't have a bias?

Either Robinson can't detect the distortion, which proves the measurement is meaningless, proving everyone's anxiety over the x3800H is also meaningless. ...or he can hear the distortion, validating the measurements and proving the x3800H is a product to avoid.

You're going to have to pick one and move on with your life...

I don't think this is correct. It's not either this or that. A few points:

a) measurements are not meaningless in any case, it's good to know about. Then the question is about interpreting those measurements
b) everybody has a bias. It's not only Robinson. We can't remove those biases when we listen even if we try. It's not even a conscious bias, it's that the brain is trying to fill it expects to hear. There is lots of information about this on ASR, just do a search

The new/original part in this thread: a lot of 'strange' posts about the SINAD number and audibility. All ~wrong by default since those two were proven again and again to be not correlated. But even 'more wrong' than usual here: some seem to think that distortion is just "one thing", some think that even SINAD 60 is not audible, some seem to think that all speakers have SINAD over 70 (my choice as the most 'new'), some think that SINAD X is generally more (or less) audible than SINAD Y... and many of those posts have a disturbing number of upvotes.

I would recommend everyone to read the above linked post very carefully, especially points 1,4,6.
Spoiler: no1 gives a much better number where you may start to assume that SINAD is no problem. A lot bigger than this AVR's 87 and even that number is just a sort of guess/assumption. Like pretty much every other connection that people make between SINAD and audibility.

Feel free to organize a blind test of two or three AVRs in pure direct mode powering speakers, and please report in if you can tell a difference between SINAD levels of 75, 85 and 95. If you are able to consistently tell a difference using music or movies as a test signals, I promise to increase my donation to ASR over the next months.

 

[lashto]

Feel free to organize a blind test of two or three AVRs in pure direct mode powering speakers, and please report in if you can tell a difference between SINAD levels of 75, 85 and 95. If you are able to consistently tell a difference using music or movies as a test signals, I promise to increase my donation to ASR over the next months.

Not even sure how to answer that...
Here's the shortest version: not correlated means NO!

A lot of research was done between 1950 (BBC engineers) and 2000's (Toole). They only found that the SINAD/THD numbers do not correlate well with either audibility or preference. You simply can not organize a test with THD numbers and expect conclusions about audibility. Same as you should not organize a test with apples and expect conclusions about helicopters. That is what "not correlated" means.

The SINAD number can tell you a lot of things. Audibility is just not one of them.
Everyone would like to have a test with audibility-relevant numbers. But we don't have any...

 

[Karls_speaker_works]

There are other possibilities, bias or not. For example (just one example), the x3800h might have been pushed to near, or even further beyond the amp's clipping point than what he did when listening to the Marantz.

This has been the case for nearly all detectable amplifier distortion in my experience. I wouldn't at all be surprised to find out the former was clipping.

 

[Roland]

This thread casts doubt on the link between measured amplifier performance and sound quality. Perhaps better measures are needed? The current measures only seem to be a guide to some esoteric engineering quality (but not reliability, features, usability, ergonomics, aesthetics, efficiency etc.).

 

[voodooless]

This thread casts doubt on the link between measured amplifier performance and sound quality. Perhaps better measures are needed?

No, it just says SINAD has little correlation with perceived differences. There are many other measurements left.

 

[lashto]

This thread casts doubt on the link between measured amplifier performance and sound quality. Perhaps better measures are needed? The current measures only seem to be a guide to some esoteric engineering quality (but not reliability, features, usability, ergonomics, aesthetics, efficiency etc.).

About the THD number and audibility/preference (or "sound quality" as you put it): it is not "this thread" but 50+ years of research. Very serious research. And it does not "cast doubt", it's a pretty loud and clear "no correlation".

Otherwise, I love your "esoteric engineering quality" quote .
Just read and learn. It will get easier. Somewhat. If it was totally super easy, even Denon would've got it right ..

 

[Bren Derlin]

you can do a lot better. Quoting you to yourself should be quite popular

Everytime when measurements are below expectations there is this big 'dance': the review/er is wrong, measurements do not tell everything, SINAD/etc are irrelevant etc..

The new/original part in this thread: a lot of 'strange' posts about the SINAD number and audibility. All ~wrong by default since those two were proven again and again to be not correlated. But even 'more wrong' than usual here: some seem to think that distortion is just "one thing", some think that even SINAD 60 is not audible, some seem to think that all speakers have SINAD over 70 (my choice as the most 'new'), some think that SINAD X is generally more (or less) audible than SINAD Y... and many of those posts have a disturbing number of upvotes.

Straw man argument. No one here is talking about an AVR with a SINAD measurement of 60.

I would recommend everyone to read the above linked post very carefully, especially points 1,4,6.
Spoiler: no1 gives a much better number where you may start to assume that SINAD is no problem. A lot bigger than this AVR's 87 and even that number is just a sort of guess/assumption. Like pretty much every other connection that people make between SINAD and audibility.

The arguments being made to validate the measurements are literally invalidating themselves. Perhaps they should be paying attention to point #8:

From your link: 8. Please don't quote other uninformed/non-practicing opinions of people regarding audibility of distortion. This type of work requires listener training to represent all listeners. Using casual groups of people, as sadly done in some research, just produces wrong results. Witness how the masses are oblivious to compression artifacts whereas trained listeners can easily hear them.

You realize you're saying people need "trained hearing" to detect the distortion noise out one side of your mouth. Then out of the other side of the mouth, it's being suggested, "no one outside of a select few members of ASR have such training". ...so let's pretend that's true...

The Denon x3800H is likely not targeting the select few "trained listeners" with "elevated standards" as their consumer. The rest of us, who can't decipher such anomalies, because of our untrained listening capabilities, will be perfectly fine with a product like the Denon x3800H, Onkyo RZ50, or a Marantz Cinema 60.

It's bad enough there are boutique retailers who turn up their nose to people with a $5,000 Audio or A/V budget. In here we have the "you can't hear what we can hear" (highly elevated standards), and I promise you, not every ASR reader/member can hear at that elevated level.

...which circles back to: If you possess the golden ear of hearing, and the x3800H is going to be problematic for you, don't buy it. Get something else. It's a good bet the rest of the people looking for an AVR don't need to be frightened buying a product where they're never going to hear at such a problematic level. ...and if they do, return the receiver and pick something else.

 

[lashto]

Straw man argument. No one here is talking about an AVR with a SINAD measurement of 60.

Did not read the whole thread and cannot help with the "no one" part. But I did not write a single word about "an AVR with a SINAD measurement of 60". Such a thing does not even exist (AFAIK).
More generally, I did not write much in this thread, just quoted posts and research done by others.

From your link: 8.
...
You realize you're saying ...

Cannot help at all here. Those points 1-x are not mine, I just quoted amir's post. If you found errors, it's probably best to respond to that post in its original thread.

If you possess the golden ear of hearing, and the x3800H is going to be problematic for you

The x3800H is very problematic to many people, including me. But it does not hurt my ears, it mostly 'hurts' the part in between.
And no, I do not have golden ears. Actually, I do not have anything golden, not even golden socks...

 

[beagleman]

Not really, because I think beagleman 's point is not (or not just..) about the double down capability but also what he said in his last sentence "An amp great at low impedances, by its very design will lose performance at higher or normal impedance"

The AHB2 can almost double down into 4 ohms, but definitely not 2 Ohms.
As beagleman predicted, it "performance at higher impedance....", in this case, only 100 W into 8 Ohms, that is clearly not due to current capability but most likely voltage capability.

Take a look of the voltage requirements for the varying load impedance, based on the same current capability (Power = V^2/R, or I^2*R):

Voltage = square root of P*R,
I = square root of P/R

For 190 W, 4 ohms, Voltage required = square root 190*4 = 27.57 V, and Current = square root of 190/R = 6.892 A

Since it is capable of 6.892 A, the theoretical power output capability should be:

Power (W) = I^2*R = 380 W into 8 ohms, but the amp is only rated 100 W into 8 ohms. This amp may not perform as well when driving certain so called 8 ohm nominal, but with high peak impedance speakers than those nominal 4 ohms with low dip ones, when operating at or near its power output limit.

In short, if an amp is designed to double down (that is almost double down), it will have trouble to "double up". Hence a compromise, that need to be done for the targeted type of load for the targeted applications.

That's the nice thing about objective measurements, I most likely don't agree with @beagleman based on his/my subjective views/preference, but on the objective side, I have no choice but to agree with him on things he posted that are based on engineering principles.

Based on Benchmark's website information for the AHB2's power output for various load impedance:

CONTINUOUS AVERAGE OUTPUT POWER​

< 0.0003 % THD+N at full rated power, 20 Hz to 20 kHz

• 100 Watts per channel into 8 Ohms, both channels driven
• 130 Watts per channel into 6 Ohms, both channels driven
• 190 Watts per channel into 4 Ohms, both channels driven

Exactly and why I am bad at explaining things.....Great job!!

 

[hmt]

You'll need to factor in the listening distance, but yes, 98 dB speakers (can they manage that down low as well?), will be very loud with only 50W already. Now the question is: when are you ever going to listen that loud? Isn't the amps SINAD worse than the AVRs? In any case, I would not worry about it too much.

The question here would be also if the noise floor is low enough. 98 db sensitivity LS and a 29 db gain amp are gonna be pretty hissy unless you sit far away.

I have one little note to make for the NAD 758 as a broken example. Even here it depends on the use case. The SINAD is not so low because that DAC is bad. It is because the amps are clipping and bleeding distortion into the pre outs. At least at 1V SINAD is good with 88 db. So when paired with some amp with a sensitivity of 1.2V one will never encounter this high distortion (the amp in the NAD is pretty mediocre however).

 

[GalZohar]

@voodooless Thank you for the quick reply.

Distance to the sweet spot is 9 feet away from LCR and I have multiple subs so my crossover is 80Hz for all channels. I generally listen to movies at -10 and sometimes reference level. (Top Gun) So that would be what... 75dB to 85dB?

Reference level is peaks of 105db for speaker channels and 115db for LFE, where "normal" sound levels will be 85db for speakers and 95db for LFE.

For 9 feet you should subtract about 6-9db - Not sure about exact formula, you'll probably have to measure for your room, but 9.5db is what you would need to subtract "outside" (without any room gain / walls).


Also, keep in mind that speakers that declare a very high sensitivity are usually lying bigtime... Look for measurements for those speakers to see the true sensitivity and the frequencies in which it applies.

 

[lashto]

... (but not reliability, features, usability, ergonomics, aesthetics, efficiency etc.).

Here is a kind of 'metric' which should be much easier to understand/follow and covers most of your extra criteria. It is not exactly a precise/sure/safe metric but should be good enough as a rule-of-thumb (also very 'healthy'):

If a device's review thread is very long and very disputed, it's probably much better to avoid it.


Why would you risk your money & time on a device surrounded by controversy !?

• Some of the disputed points may be bullshit or very unclear (to you), but the controversy itself could be a big red flag.
• There are many controversy/headache-free alternatives.
• Fixed/updated devices are sometimes released in such 'hot' cases.
• In this case, D&M will have new AVRs next year (the latest). In the case of DACs, it's almost sure that a 'better' one will be tested next week...

Spoiler: a sort of P.S.

There is already way too much confusion/misunderstandings/emotion/etc in this thread. Not only on ASR but ~everywhere: audioholics/avsforum/etc. It's enough to make the entire D&M 2023 lineup 'swim' in controversy. Forever.
So again, why would you want that 'noise' in your room/life?!

It's also very healthy to not get involved much in such threads. And I should follow my own 'metrics' more often

 

[HarmonicTHD]

Expected after the chip change but nonetheless disappointing especially as Denon had alternatives for DAC chips for the X3800. The X3800 is pretty much a second gen X3700 with a (very) long shot promise for Dirac. Not buying.

Certainly will hold on to my first gen (AKM) X3700 as long as possible.

 

[beagleman]

LOL - there is no substitute for torque (in cars), and current (in amps)

There may be no substitute for either ,but there is also no nothing even remotely similar about the two.

High current is good at times. If one has a poorly designed speaker for instance, there can be dips down to very low impedances, where higher current will be of benefit.

When one has a speaker that its design takes into account, being within a normal impedance range, a high current amp will by its very design, be a compromise in offering less output at 6-16 ohms.

And, FYI, many speakers may be rated at say 6-8 ohms, and often have large regions that are in the 8-20 ohm range, in parts of the frequency range.

"Doubling down" with every drop in half of impedance is ONLY possible in amps that intentionally start with fairly low power outputs at normal impedances.

In amplifier cases, it is the "No free lunch" thing that applies.
If one has problematic speaker impedances that are very low, YES you need a high current amp at times, but where the impedance rises or in a speaker at normal impedance rating, you tend to lose some sound output was my only point.

 

[peng]

There is always a compromise. Many of these amps will have significantly more distortion when driving 2R loads:
View attachment 241948
The Krell is already shit in 8R, 2R will be horrible.
View attachment 241949
Even the new Hypex amp has a bit lower distortion, but much less so:
View attachment 241952
That’s still 0.00025% @20W into 2R compared to the 0.04% of the Krell at 8R or 0.025% of the 333 into 2R.

I think by now the original points made might have been lost somewhat, so let me summarize just a couple of the key points mentioned so far by the two gentlemen who started this whole topic of current vs (or not) voltage thing.

beagleman

Optimize for 8 ohms and it strains to do 4 ohm loads.
Optimize for 4 ohms, and it can not reach its full potential for 8 ohm loads.

dlaloum

Yes, the amp has to be designed to handle substantial current - but that should not negatively impact performance at higher impedances!

but this does not imply that an amp engineered to provide 40V @2ohm, will be unable to provide the same 40V@8ohm ...although it DOES imply that it will need to be more heavily engineered... and that this engineering, will be redundant when driving 8 ohm or 4 ohm loads - but there is nothing there that implies a reduced performance for an amp engineered that way, when driving "easier" loads.

If we read the above carefully, we would have been much more agreeable as there really isn't much if at all to disagree with or about. I highlighted 3 key words that, I thought might have contributed to the perceived need to debate a little, but that would be only in terms of on strictly speaking basis. Regardless, that's all part of the fun in this hobby, that includes to learn, and even argue on forums lol..

 

[peng]

LOL - there is no substitute for torque (in cars), and current (in amps)

No one disagrees with the need of current in amps, others are just saying there is a need for voltage as well and of course ideally an amp should have both high voltage and high current capability. But that would, as you mentioned before, involved some "heavily engineered...", and obviously relatively expensive parts.

I just randomly googled the impedance graph of one of the often mentioned as a difficult load, see those high impedance peaks beagleman talked about?
Now we have both examples, one from you about the impedance dip to near 2 ohms and this one with peaks to higher than 20 ohms.

The Quad amp you mentioned before may not struggle with your low impedance speaker under some conditions but it would struggle with speakers that have such impedance peaks under certain conditions of use.

Again, I agree with you that there are amps designed to maintain excellent performance regardless of the impedance peaks and dips. That's in theory, in practice I am not ware (they probably do exist) of any power amp that are truly designed to perform equally well with speakers that have relatively high impedance peaks and low dips in the frequency range that have a lot of contents. McIntosh use output autotransormers (they call them autoformers) for impedance matching purposes but the user have to manually move the connections if they switch from 8 ohm to 4 ohm speakers, so it is still not ideal for speakers that have both high impedance peaks and low impedance dips I mentioned. That's just all theory, in practice also as you mentioned before, no one need to worry about sacrificing performance as there is always the option to just buy more "power". So if one needs 100 W/160 W 8/4 ohm amp, then just go for a 400/600 W 8/4 ohm amp and all those wonderful 20 ohm peaks and 2 ohm dips will be forgiven!!

By the way, for a relatively low cost solution, bridging a pair of AHB2 can also be a good option for those who has such speakers.

Bowers & Wilkins 802 D3 Diamond loudspeaker Measurements

Sidebar 3: Measurements

www.stereophile.com

 

[amper42]

Denon's owner MASI just tagged the 52 week low this morning at 110.85 from a high of $305.21. If they can't hold that level it's possible financial issues may appear. From layoffs to asset sales may be next? Customer satisfaction would be a good goal to initiate at this point.

 

[-Matt-]

Distance to the sweet spot is 9 feet away from LCR and I have multiple subs so my crossover is 80Hz for all channels. I generally listen to movies at -10 and sometimes reference level. (Top Gun) So that would be what... 75dB to 85dB?

Reference level is peaks of 105db for speaker channels and 115db for LFE, where "normal" sound levels will be 85db for speakers and 95db for LFE.

For 9 feet you should subtract about 6-9db - Not sure about exact formula, you'll probably have to measure for your room, but 9.5db is what you would need to subtract "outside" (without any room gain / walls).

Assuming that the AVR is calibrated (i.e. by Audyssey) then shouldn't the stated levels already be correct at the listening position?

If so, there shouldn't be any need to subtract dBs for distance. In fact, you could probably add the 6-9 dB if you wanted to estimate the level at speakers.

Edit: Nevermind, just re-read the original question and you were working forward from speaker specs, rather than backward from listening level. (And not necessarily for an AVR).

 

[gew55]

It's really a pity that the measuring results are hum-ho.
But I bought this AVR to replace my (broken) UMC-1 Emotiva. So I use this AVR in pre-amp mode. And to me it sounds amazingly accurate. I have a lot of DTS-HD Master Audio music Blu-ray's and these sound much better than what I was used to at the umc-1. More space and better placement of instruments
And finally I can listen to my SACD recordings in DSD surround
And for stereo CD and high-res FLAC I use a s.m.s.l su-9 DAC through the pure direct mode and that's perfect for me
So here, a happy user of the X3800h
Waiting with high hopes for the dirac live upgrade

 

[GalZohar]

Assuming that the AVR is calibrated (i.e. by Audyssey) then shouldn't the stated levels already be correct at the listening position?

If so, there shouldn't be any need to subtract dBs for distance. In fact, you could probably add the 6-9 dB if you wanted to estimate the level at speakers.

Edit: Nevermind, just re-read the original question and you were working forward from speaker specs, rather than backward from listening level. (And not necessarily for an AVR).

Yes, calibration will set the level to try to be reference level at the listening position, but it doesn't take into consideration whether your speakers and amplifiers are actually able to take you there. It will try to play at reference and either fry something or greatly distort if some parts are not up to the task.