Denon AVR-X4800H AVR Review

[peng]

Hmm, still trying to make sense of this tbh but I read it differently.

I think we can both agree that optimum SINAD is achieved at 1V and 0dB attenuation but that likely to be very loud. However, if the objective is to reduce the signal voltage by say 10 times, we could divide the digital signal by 10 in a dsp and let it output 0.1V and use 0dB on the volume IC, that would get us 0.002% THD + N...however if we choose instead to keep the DAC output around 1V and attenuate in the volume IC by 10dB (as I believe is the normal way unless something fancy or hybrid is happening) we get around 0.001 % THD + N i.e. half the distortion.

Your math is not correct. Don't forget voltage ratio in dB is in log scale, so -10 dB is not the same as 1/10. Even if we do it your way, if you calculate it correctly:
From Amir's measurements, at 0 dBFS, volume 82, the pre out voltage was about 2 V, so for 1 V, volume will be 76, with 0 dBFS, (voltage halved for every 6 dB drop)
If you reduce the volume IC output by 10 dB from 1 V, then the pre out voltage will be 1*(1/√10) = 0.316 V, that's not at all!

So, of course the THD+N would be lowered if you drop the pre out voltage to 0.316 V, compared with that for 1 V.

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

As I said before, it is better, simpler to just look at those two curves I posted, as they are the actual curve from the volume IC's data sheet; and you can see that if you lower the input signal level by 20 dB, from say,

1 V to about 0.1 V (your pick), THD+N drops from about 0.0002% to about 0.0023%, or SINAD of 97.7 dB

Now if want to keep the pre out voltage at 1 V, again, the volume will have to be at 76 (voltage halved for every 6 dB drop)
So, in order to drop the pre out voltage from 1 V to 0.1 V, the volume will have to be attenuated by 20 dB, from the curve, that is, from 76 to 56. In this case, you can see that for pre out voltage to drop from:

1 V to about 0.1 V, THD+N drops from 0.0002% to about 0.0013%, or SINAD of 93.98 dB.

To conclude:

A) If we keep the volume the same, that is 76, in order to get pre out voltage of 1 V, and vary the input signal from 0 dBFS to -20 dBFS, SINAD drops from 0.0002% to 0.0023%, or SINAD of 92.77 dB
B) If we keep the input signal at 0 dBFS, and vary the volume from 1 V to 0.1 V, that is -20 dB on the volume setting dial, SINAD would drop from 97.7 dB to 92.77 dB.

A) is the old method, and B) is the new method Amir used on measuring the 4800, that that means if Amir had used the same method, the 4800 would do a touch better under low level listening conditions, all else being equal. The new method, based on this example, using you choice of 1 V and 0.1 V, the old method would do better by 4.3 dB at the 0.1 V level. Again, that is opposite to what you seem to expect. We have opposing results!!

Regardless, since the AKM4700 has a better DAC, it is very likely that overall, it would still do better under all conditions including low level listening.

So our conclusions are different, quite the opposite.

Thats just the start though, now if we choose 0.01V DAC output vs 20dB attenuation we get 0.02% vs 0.003%. Almost an order of magnitude difference. Much better to attenuate than to reduce the incoming signal.

At 3 orders of magnitude the effect is even bigger at 0.001v@0dB = 0.2% vs 1V@-30dB = 0.01%. I read this as favouring use of the volume IC vs lowering the output in the DAC.

That of course is only considering the effect at the volume IC part of the chain. I would expect that a lower signal coming from the DAC would also have a much worse signal to noise ratio than a high signal which IMHO would further reduce SINAD

Again, to me, your calculations are not correct, so your numerical example do not represent an apples to apples comparisons.

The relevant formula used in my numerical example are included in the following linked calculators:

Decibels to Voltage Gain and Loss convert calculation conversion amplification amplifier electronics - sengpielaudio Sengpiel Berlin

Gain and loss - damping and amplification in dB voltage conversioncalculation amplification amplifier electronics - field parameter - Eberhard Sengpiel sengpielaudio

www.sengpielaudio.com

THD to dB - convert percent % to decibels dB percentage voltage % vs per cent converter THD+N total harmonic distortions calculation signal distortion factor attenuation in dB to distortion factor k in percent decibel damping - sengpielaudio Sengpiel

Calculate THD and distortions decibels dB to percentage % per cent percent voltage % vs converter and dB signal distortion THD+N total harmonic distortions factor attenuation damping - Eberhard Sengpiel sengpielaudio

www.sengpielaudio.com

 

[peng]

Better yes, but not more representative. And I also have a problem with that, that the measurement methodology was changed in favor of 4800. Only in this one case! It is then poorly comparable. I'm very much looking forward to what method Amir will use to measure the nearest non-Harman AVR...

I believe the opposite is true, but the difference is very small either way, really negligible. Feel free to check my calculations and interpretation. If proven wrong, I will edit or delete the post.

 

[jpqpi]

I believe the opposite is true, but the difference is very small either way, really negligible. Feel free to check my calculations and interpretation. If proven wrong, I will edit or delete the post.

Think it's me that will be deleting/editing my post tbh More used to thinking of power gain than voltage gain where I believe 10dB gain = a factor of 10...when I get more time I will look over all this with fresh eyes and thinking in terms of root 10 i.e. a factor of 3.16

One thing I suspect though is that the volume 82.5 figure that Amir mentions doesn't refer to decibels...I believe it is just a number chosen somewhat arbitrarily by Denon to represent perceived volume. Far from certain on that though.

Thank you for your input, and I have to say I love this forum as I learn so much... every day is a school day here!!

 

[peng]

One thing I suspect though is that the volume 82.5 figure that Amir mentions doesn't refer to decibels...I believe it is just a number chosen somewhat arbitrarily by Denon to represent perceived volume. Far from certain on that though

Yes, 82.5 may be a random choice by Denon, but I can assure you if you turn it up 6 dB, you get double the voltage, 12 dB, get 4X, 18 dB, 8X, and 20 dB will get you 10X. Conversely, if you lower it by 20 dB, you will get 1/10 the voltage. That's for voltage, obviously for power, every 3 dB would represents a factor of 2X, or 0.5 for -3 dB. I have actually taken measurements and can confirm that's the way it works. So the way it is scaled, the number does corresponds to dB.

By the way, I am not sure if I posted a link to the datasheet of the chip:

https://www.nisshinbo-microdevices.co.jp/en/pdf/datasheet/NJU72343_E.pdf

 

[OCA]

Not aware of any internal 32khz sample rate in any DAC. I am curious though, could you elaborate a little more and include where you found your info for this?

If you hack the MullEQ Editor app calibartion file (.ady) in a json editor, you can see 3 sets of IDENTICAL filters for 3 different sample rates. Obviously, the filter resolution is capped by the lowest one:

 

[OCA]

Exactly my question too. According to one of the founder of Audyssey, D+M used 48 kHz, due to the limitation of the processing capability of their gear, never heard of the 32 kHz thing.

Why do you think highest manual graphic equalizer band frequency is 16kHz?

 

[ivo.f.doma]

I believe the opposite is true, but the difference is very small either way, really negligible. Feel free to check my calculations and interpretation. If proven wrong, I will edit or delete the post.

According to your calculations, it really looks the same...
It's just a shame that he didn't measure it both ways the first time and we could have been clear about the future. That is, even those less technically proficient.
Thanks for explaining the calculations!

 

[Rockman2]

Well....not to happy with my 4800 so far. It has had HDMI hand shake issue with my Dish Network Sat receiver and my #1 sub output is not working. Denon had me do a factory reset. This did not work. I am outside the exchange widow so now the unit needs to be sent in for service/replacement. So I have to deal with that.

After more trouble shooting found out the sub1cable was bad. Called Martin Logan as I thought I had a dead sub and they had me disconnect the sub cable from the AVR and tap it with my finger. They said if the sub thumped the cable and sub were good. Well it did thump so I proceeded with more trouble shooting and got down to doing a factory reset. Luckily, I thought I would try a different sub cable before sending the unit back. That fixed the issue. Also doing the factory reset fixed the Hand Shake issue with the Dish receiver. I now have picture on the TV when using the AVR. When the Hand shake issue was discussed with Denon one of their tier 2 support people, he said it would be better to run the Sat and DVD direct to the TV and then back to the AVR using Earc. He claimed this would eliminate the hand shake issue. I had called Denon support when buying HDMI cables and they had said the opposit was better and to go from the Sat Reciver and DVD into the AVR first so that is what I did.

Any thoughts on which way is best to run the HDMI cables? AVR or TV first?

 

[peng]

Why do you think highest manual graphic equalizer band frequency is 16kHz?

I don't know exactly how to interpret that hacked file. It is under "referenceCurveFilter", and Audyssey had explained (don't remember where I read it) that the filter resolutions vary within the audio frequency range. Audyssey did say MultEQ could apply RC from 10 to 24,000 Hz It is easy to agree that if Denon resamples everything to 32 kHz, then correction to frequency above 16 kHz would not be possible.

Not knowing how to interpret that file and how the sampling frequency relate to the coefficient32 kHz, I am in no position to agree or disagree with your point: "Obviously, the filter resolution is capped by the lowest one". To me, I just don't know if that is true or false and of course you may be right. Or maybe there are 3 sets, and only one is used depending on the Audyssey versions. There are 3 versions currently in use for the entry level, mid level and higher level AVRs, i.e. MultEQ, MultEQ XT and MultEQ XT32, again, I just don't know. It is a question for Denon or Marantz, but that is only if you can reach their highest level of customer support. The first or even second line support is not going know, and the highest level will have the answer but likely will not tell us. And if you ask Audyssey, they will (90% certain) refer you to Denon or Marantz because they are not responsible for the Editor App.

 

[peng]

According to your calculations, it really looks the same...
It's just a shame that he didn't measure it both ways the first time and we could have been clear about the future. That is, even those less technically proficient.
Thanks for explaining the calculations!

Fully agreed, but I also appreciate Amir's workload, so he has to strike a good balance, and that means do the essentials but skip the nice to do kind of stuff. It takes a lot of time to review/measure electronic gear, especially AVRs/AVPs, and Amir measures more than one gear a week, sometimes one a day.

 

[OCA]

I don't know exactly how to interpret that hacked file. It is under "referenceCurveFilter", and Audyssey had explained (don't remember where I read it) that the filter resolutions vary within the audio frequency range. Audyssey did say MultEQ could apply RC from 10 to 24,000 Hz It is easy to agree that if Denon resamples everything to 32 kHz, then correction to frequency above 16 kHz would not be possible.

Not knowing how to interpret that file and how the sampling frequency relate to the coefficient32 kHz, I am in no position to agree or disagree with your point: "Obviously, the filter resolution is capped by the lowest one". To me, I just don't know if that is true or false and of course you may be right. Or maybe there are 3 sets, and only one is used depending on the Audyssey versions. There are 3 versions currently in use for the entry level, mid level and higher level AVRs, i.e. MultEQ, MultEQ XT and MultEQ XT32, again, I just don't know. It is a question for Denon or Marantz, but that is only if you can reach their highest level of customer support. The first or even second line support is not going know, and the highest level will have the answer but likely will not tell us. And if you ask Audyssey, they will (90% certain) refer you to Denon or Marantz because they are not responsible for the Editor App.

This is from an XT32 correction ady. You're ignoring the fact that the convolution data is digit by digit identical in everyone of them otherwise one could argue that they are applying different filters for different sampling rates. This can only mean everything is being upsampled from a 32kHz base sampling rate. There's a second set of filters for the Flat curve btw.

 

[peng]

This is from an XT32 correction ady. You're ignoring the fact that the convolution data is digit by digit identical in everyone of them otherwise one could argue that they are applying different filters for different sampling rates. This can only mean everything is being upsampled from a 32kHz base sampling rate. There's a second set of filters for the Flat curve btw.

I am not ignoring anything at all at this point. I have already said "I am in no position to agree or disagree with your point..". As well, I also said you may be right. Just not 100% sure you are right. Also, just because the ady file is for XT32, it does not mean it wouldn't include stuff that pertains to the other versions. I have to wonder, if it is capped by the coefficient32kHz, then why were 44.1 and 48 kHz ones there at all? And what exactly does coefficient32kHz mean? I am familiar with Fourier transform, but not familiar with FIR filter implementations. I do know in general term, software and hardware, are both often subject to standardization in various similar products. And again, for the 3rd time, I did say you may be right. So, if you are so confident, then believe what you believe as factual, and there is no need to convince me. I am not trying to contradict you in any way, but remain skeptical, and speculative as this point, not knowing enough facts.

 

[jpqpi]

I wonder if @amirm would be kind enough do a special write up on reducing volume in the digital domain vs the analogue domain wrt SINAD concerns. That might give us something to work off generally

 

[jpqpi]

Why do you think highest manual graphic equalizer band frequency is 16kHz?

The clue is in the title Graphic equalizer "Band". Band being a group of frequencies having a lowest and highest frequency but described by a central frequency and a "bandwidth" which let's you know how far above and below the central frequency it goes. Bands often double in value from 1khz upwards (and halve in value from there downwards) I.e. 31hz, 62hz, 125hz, 250hz, 500hz, 1khz, 2khz, 4k, 8k, 16k..... so I'm not all surprised to see 16khz as a last band in most graphic equalizers as 32k wouldn't make sense (way out of our audible range)

 

[Jack B]

Still thinking about springing for the 4800 but am wondering about the versions of Dirac that are currently available. I have two SVS subs, both ported, but different models, and not equidistant from the MLP. The subs' amplifiers allow for shifting phase (that feature has proved useful), and Audyssey XT32 on my current Marantz allows for compensation for distance from each sub. What I am wondering is this: Do any of the versions of Dirac (currently available or anticipated for the Denon 4800) have a feature that will compensate for two subs not being equidistant from the MLP?

Or should I, if using Dirac, just treat the two subs as one, and depend on the phase adjustment feature in the subs' amplifiers to optimize them? Thanks much!

 

[drelldrell]

Still thinking about springing for the 4800 but am wondering about the versions of Dirac that are currently available. I have two SVS subs, both ported, but different models, and not equidistant from the MLP. The subs' amplifiers allow for shifting phase (that feature has proved useful), and Audyssey XT32 on my current Marantz allows for compensation for distance from each sub. What I am wondering is this: Do any of the versions of Dirac (currently available or anticipated for the Denon 4800) have a feature that will compensate for two subs not being equidistant from the MLP?

Or should I, if using Dirac, just treat the two subs as one, and depend on the phase adjustment feature in the subs' amplifiers to optimize them? Thanks much!

I have the X4800H with two SVS subs that have been in different positions including non-equidistant. Dirac results have been very good. Go to the links for tips - especially the mic calibration and speaker settings.

My understanding is Dirac adjusts phase and delay (distance) as needed. I connected the subs to the AVR separately, not with a splitter. I purchased full bandwidth correction.

Dirac Live v2 & v3

https://www.avsforum.com/threads/dirac-live.3137722/page-839

 

[Karls_speaker_works]

I wonder if @amirm would be kind enough do a special write up on reducing volume in the digital domain vs the analogue domain wrt SINAD concerns. That might give us something to work off generally

Might be helpful. I rated both X3800H and X4800H as "Fine" based on the reviews. I now own the X3800H and I stand by my vote (I can expand on this if anybody cares about my opinion as to why I consider the sound to be neutral (and zero apparent "noise") and how I am happy that it drives my difficult to drive Martin Logan Sequel's (and 5 additional speakers) with ease.) The real issue to me is ... logically ... I would expect the two receivers to be rated similarly, yet that is not the case (my engineering brain wants to explode reading all of these interpretations of the two methods used to measure SINAD and how I believe it influenced people's opinions.) I can't shake the feeling that flack from the review of the X3800H had an influence. I apologize in advance if such a statement ruffles some feathers, it's just how I "feel".

 

[Brambo67]

Might be helpful. I rated both X3800H and X4800H as "Fine" based on the reviews. I now own the X3800H and I stand by my vote (I can expand on this if anybody cares about my opinion as to why I consider the sound to be neutral (and zero apparent "noise") and how I am happy that it drives my difficult to drive Martin Logan Sequel's (and 5 additional speakers) with ease.) The real issue to me is ... logically ... I would expect the two receivers to be rated similarly, yet that is not the case (my engineering brain wants to explode reading all of these interpretations of the two methods used to measure SINAD and how I believe it influenced people's opinions.) I can't shake the feeling that flack from the review of the X3800H had an influence. I apologize in advance if such a statement ruffles some feathers, it's just how I "feel".

I think they are 'about' similar;
- same DAC's
- different amp lay-out but that shouldn't be an issue
- same components, different hands assembling them (Japanese vs. Vietnamese) will not make a difference (see 6700 v 4700)
- different way of presenting measurements....

Hey what really has changed? These new Denons, particularly now prices are coming down, are great! The older ones also even though missing some interesting features the new ones have.

On this forum people gather looking for the details of the details. Often not owning a piece of equipment but no shy of giving their opinions. That's with me but I don't value it. People like 'Peng' strike the right balance in my opinion. Yes; every manufacturer should be wanting to always make 'better' equipment, not worse. But he says it very well. This is all outside the human spectrum of being capable of hearing. I mentioned it before; in my setup I have a TI 4700 with two separate DAC's (Argon and Topping) These are never used any more... It is so much simpler to stream from Apple Music on my iPhone to the DAC in the 4700. I can't here the difference even if I wanted to. So just enjoy the 3800!

 

[tesseractASR]

Might be helpful. I rated both X3800H and X4800H as "Fine" based on the reviews. I now own the X3800H and I stand by my vote (I can expand on this if anybody cares about my opinion as to why I consider the sound to be neutral (and zero apparent "noise") and how I am happy that it drives my difficult to drive Martin Logan Sequel's (and 5 additional speakers) with ease.) The real issue to me is ... logically ... I would expect the two receivers to be rated similarly, yet that is not the case (my engineering brain wants to explode reading all of these interpretations of the two methods used to measure SINAD and how I believe it influenced people's opinions.) I can't shake the feeling that flack from the review of the X3800H had an influence. I apologize in advance if such a statement ruffles some feathers, it's just how I "feel".

I tried out the 3800 and 4800. Just sent the 3800 back yesterday, but absolutely hated to, it's a great unit. I got the 4800 as a refurb for only $500 more, so decided to keep it as it has a couple of extra features mentioned at the end of the review that I like.

I am also looking at buying 3 Martin Logan ESL panels across the front. The 3800 was going to power that and would do just fine in all but the most demanding situations. You can rest easy knowing you got a great AVR that is very flexible, sounds good and is well-built.

 

[Jack B]

I think I read somewhere that there are two ways to use only the pre-outs of the receiver: One way simply ignores the receiver's speaker out terminals and connects to the pre-outs; the other way actually shuts off the receiver's internal power amplifier channels, which I would like to do. Please enlighten me? Thanks.