Denon AVR-X4700H 2020 AVR Review

[zelig]

It isn’t a given a company will even do that these days, as NAD has shown.

Go NAD!

 

[peng]

If you ever consider sending the X4400H to Amir, I would gladly contribute to the shipping costs so we can build up the Denon database

Good idea, and I would if I could, but I have a jungle of wire hooked up to it, it'll take a lot of time removing everything, reconnecting the AV8801 temporarily when waiting for its return and then repeat the process. It may be better if we can find 20 people to share the cost of an AVR-X4500H that is still available at AC4L for $1,000 for Amir to use it as a reference AVR, one that he can measure everything one can think of.

The X4500H for all intents and purposes is identical to the X4400H, and SR7012 (likely 7013 too but I have not seen the SM for that one so I would be guessing), except the HDAM.

 

[peng]

Exactly...
Multichannel music vs Stereo music is as simple as this,
Stereo = watching the orchestra/band/artist from the seats
Multichannel music = Actually being on stage IN the orchestra!

Anyone that's not a believer in multi c. audio needs to simply listen to the Hans Zimmer Live in Prague bluray, try it in 2.1 then go to surround. It's night and day and I wouldn't have it any other way.

I think it is a very subjective thing though don't you think? For sure you would be right about the night and day difference for some people but there will also be no shortage of those who prefer just two channels. I have quite a few 5.1/7.1 SACD and BR concerts (including "live" ones) and while I also enjoy multichannel, I ultimately prefer 2 channel, don't remember when was the last time I use my 105 with my stockpile of discs these days. It may be because my relative small room that cannot take full advantage, but then 7.1.4 works well for me in the same room, so go figure..

Since you mentioned the Hans Aimmer specifically, I may try that one and see what happens, though ouch, $61 on Amazon.ca..so let me think..

 

[peng]

Not quite
Reference is suppose to be 85db with 105dB peaks for mainspeakers and 115dB peaks for subs.
http://www.acousticfrontiers.com/2013314thx-reference-level/

Many AVR companies internally lower the volume of receiver generated test tones to 75dB to avoid damaging consumer speakers, but that is internally accounted and in the the final result you are supposed to be seeing 105dB peaks from the loud speakers with your cinema content at reference.

We did some testing on receivers auto calibration routines in 2014, and frankly the results were terribly unstandardized based on the bundled auto EQ mics - even with flagship AVRs.

Basically, don’t trust a receivers autoeq process to get you anywhere near a standardized volume and EQ.


https://www.avsforum.com/forum/90-r...mparison-g2g-november-8-2014-kansas-city.html

This is the is the “calibrated” reference comparison of a handful of receivers with the mic positions taken at standardized positions. It was/is pathetic.


post 199 and subsequent in the avsforum link:View attachment 70299

The explanation of above graph from test results:

“The Madness
So what's the deal? Why was it madness? Well - here are the post calibration frequency responses from each entry. No funny business, just absurdity. The helpers and I set the mic in the same standardized positions for each system (unless specific places were actually required, IE Anthem, DIRAC, Yamaha) and the starting position for the initial calibration was the exact same spot for ever processor. Stitch1 loaned a drum kit with a bunch of high hat stands (used as mic stands) - to ensure our mic capture positions weren't different from processor to processor. In theory, after calibration each processor should be close to the same SPL at least, if not generally reasonably close to a flatter frequency plot - RIGHT?? I mean that's the point of these systems -- RIGHT? To get the AVRs to a reference volume and try to flatten frequency response while doing so - so that each user's system in different rooms and different speaker setups has a similar audio experience?!?!?! Well, with eight different systems here is what was captured by omnimic for each as post calibration results. We followed instructions to let each auto processor optimize the room. The ONLY change we allowed post calibration was setting speakers to small and crossover to 80hz when the processor/AVR allowed. To capture the post calibration frequency response plots shown here I simply turned each AVR to -12dB on the main volume knob and played track 2 of the omnimic disk from the HTPC to the processor. The results are ridiculous. But that is the tested state of variance in these processors.”

That's informative and very interesting results, thank you for sharing. Since I am familiar with D+M's so I focused on the Denon curve and from what I could see if you run a tone similar to Deno's internally generated one, it would likely be close to the targeted 75 dB but if you are using a tone at the standard level it should be closer to 85 dB. I have checked my AVR-X4400H, AV8801 and a couple more before those two and was always getting close the 85 dB at MV=0 post Audyssey. So I wonder why you guys were getting much lower levels at "0".

 

[LTig]

It is not the same sample rate anymore.

You can play the 48 kHz samplerate file using Roon or whatever and record the output with 192 kHz samplerate. See my 2nd example behind the spoiler. What would be the problem with this setup?

 

[SimpleTheater]

Good idea, and I would if I could, but I have a jungle of wire hooked up to it, it'll take a lot of time removing everything, reconnecting the AV8801 temporarily when waiting for its return and then repeat the process. It may be better if we can find 20 people to share the cost of an AVR-X4500H that is still available at AC4L for $1,000 for Amir to use it as a reference AVR, one that he can measure everything one can think of.

That's a great idea. @amirm can you setup a GoFundMe account for this? I guess I could do it, as long as people trust me to actually send you the money.

 

[peng]

That's a great idea. @amirm can you setup a GoFundMe account for this? I guess I could do it, as long as people trust me to actually send you the money.

Thanks, by the way I picked the X4500H because its service manual is available (under $20) and it has the proven (meaning its been out for a while now) HDMI 2.0 HDCP2.3 4K HDMI board so it most likely would (or at least should) perform as good or better than the AVR-X3600H, unless the vol chip's specs tolerance is in fact a factor as I surmised, or the 3600 really has some unique feature such as better tolerance resistor and stuff that someone mentioned, but again I highly doubt Denon would have done such a thing so quietly.

If this becomes real, we should order two, assuming Amir would be willing to take the time to measure two samples to see how significant the variance in measurements are, and return one after the measurements. Not sure if that is ethical, I hope it is because we do intend to keep one.

 

[peng]

Searched again and found the sm for the 3600, for USD 14.95, but not sure if the site is legit and authorized..

 

[polmuaddib]

I posted a new thread to test Denon 4500 preouts agianst a dedicated stereo preamp in home theater forum if you wanna check it out.

 

[RichB]

Not quite
Reference is suppose to be 85db with 105dB peaks for mainspeakers and 115dB peaks for subs.
http://www.acousticfrontiers.com/2013314thx-reference-level/

Many AVR companies internally lower the volume of receiver generated test tones to 75dB to avoid damaging consumer speakers, but that is internally accounted and in the the final result you are supposed to be seeing 105dB peaks from the loud speakers with your cinema content at reference.

We did some testing on receivers auto calibration routines in 2014, and frankly the results were terribly unstandardized based on the bundled auto EQ mics - even with flagship AVRs.

Basically, don’t trust a receivers autoeq process to get you anywhere near a standardized volume and EQ.


https://www.avsforum.com/forum/90-r...mparison-g2g-november-8-2014-kansas-city.html

This is the is the “calibrated” reference comparison of a handful of receivers with the mic positions taken at standardized positions. It was/is pathetic.


post 199 and subsequent in the avsforum link:View attachment 70299

The explanation of above graph from test results:

“The Madness
So what's the deal? Why was it madness? Well - here are the post calibration frequency responses from each entry. No funny business, just absurdity. The helpers and I set the mic in the same standardized positions for each system (unless specific places were actually required, IE Anthem, DIRAC, Yamaha) and the starting position for the initial calibration was the exact same spot for ever processor. Stitch1 loaned a drum kit with a bunch of high hat stands (used as mic stands) - to ensure our mic capture positions weren't different from processor to processor. In theory, after calibration each processor should be close to the same SPL at least, if not generally reasonably close to a flatter frequency plot - RIGHT?? I mean that's the point of these systems -- RIGHT? To get the AVRs to a reference volume and try to flatten frequency response while doing so - so that each user's system in different rooms and different speaker setups has a similar audio experience?!?!?! Well, with eight different systems here is what was captured by omnimic for each as post calibration results. We followed instructions to let each auto processor optimize the room. The ONLY change we allowed post calibration was setting speakers to small and crossover to 80hz when the processor/AVR allowed. To capture the post calibration frequency response plots shown here I simply turned each AVR to -12dB on the main volume knob and played track 2 of the omnimic disk from the HTPC to the processor. The results are ridiculous. But that is the tested state of variance in these processors.”

It always was madness.
The next time I go to the cinema I'll take an SPL meter.
Just using my phone, the loudest part of the experience was the trailers which did not exceed 97 dB.
At Showcase cinemas, I am fairly sure they do not use reference.

At a local Dolby cinema, and it was considerably louder than my usual Showcase cinema.
My family and found the whole experience unpleasant. We are not going back.

I calibrate the relative speaker levels using internal tones and test disks.
For each source, the volume is adjusted for dialog volume then let it ride.
Usually, there is a 3 to 6 dB acceptable window, after which the content gets so loud it is distracts.

The variability in program material is up to 20 dB, varying between -34 for night viewing, and 13 for sources that require that for intelligibility.
This weekend we watched the Star Trek reboots (1 and 2). I had the level set to -16 that was intentionally loud.
The sword fight on the drill platform made me wince. I don't remember doing this in the theater.
Mixes are different as well as the distance between speakers. Home feels like near-field in comparison.

In stereo with my processor set to -31 produces 86 dB at my listening position.
Two speakers so, then -27 is THX reference. I have never and will never listen to a properly mastered movie at -7.
I have listened to HT -36 to -13 at different times.
Applying an additional -7 dB of attenuation for THX translates my typical listening levels between the range of -27 to -6. Still beyond what I find tolerable.
Pursing THX reference numerology is worthless.

- Rich

 

[Archaea]

That's informative and very interesting results, thank you for sharing. Since I am familiar with D+M's so I focused on the Denon curve and from what I could see if you run a tone similar to Deno's internally generated one, it would likely be close to the targeted 75 dB but if you are using a tone at the standard level it should be closer to 85 dB. I have checked my AVR-X4400H, AV8801 and a couple more before those two and was always getting close the 85 dB at MV=0 post Audyssey. So I wonder why you guys were getting much lower levels at "0".

Do note the Denon and Sherbourne (Trinnov) units were right at about the same SPL levels. Which was interesting. The SPL levels sort of fell into groups, but definitely not all the same.

We let each AVR auto EQ routine do its thing, and then I turned the main volume to -12dB from its reference setting on each unit (just to protect the ears of the setup/helper guys in the back of the room while doing post calibration sweeps) and so that chart is a recorded a frequency response sweep of the track 2 omnimic disk at -12dB on all processors post auto EQ routine.

Even if I use the software to match up the overlays at the 800-1000hz range, the corrected FR response were still all over the map!
Pic here:

One would have hoped all of those units would have been around 73dB (85dB - 12dB) (+- cheap bundled mic variance at 800 to 1000hz (typical frequency band used to measure mainspeaker SPL). They weren’t.

As you can tell the results are laughable in the context that people assume once they’ve run autoeq they are getting a fairly equitable experience to anyone else. On top of this, we know Audyssey Mics have a +-2.5 to 3dB allowable tolerance, and I’m sure other company’s bundled mics have the same loose standards. That’s a six dB swing possible in allowed mic manufacturing alone!!!
Its nearly impossible to comprehend the nearly 20dB deltas we measured in the post calibration results of these products.

Amazingly bad really.


https://www.avsforum.com/forum/90-r...vember-8-2014-kansas-city-7.html#post28900602
Post 199 is the setup config
Post 200-207 documents each setups specifics
Post 208 is some followup analysis

When we realized the volumes were so very different post autoeq calibration we realized we couldn't do a listening test like that, so we manually calibrated the main volume using the Disney + calibration disk test tones on each AVR to 72dB on the front left speaker for our subjective listening test.

This Pic is more like what we actually auditioned them at after manually level matching, I used the omnimic software to show the same basic thing we did with the Disney + disk here: You can see when you manually adjust for SPL the results aren't horrendous, but still far from a ideal standard. This info is in post 208. Even manually level matched at 1Khz You still have 20dB swings in the bass section below 100hz, and 5-10 db swings top to bottom everywhere else.

^
In the picture's legend ( in front of processor name) you can see the dB offsets I had to use in omnimic in order to get the different units to roughly line up at 1Khz.


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


The subs FR response is particularly strange. We had 4 sealed 18's, and they were at a good starting place to begin with! How the autoeq routeines screwed that up is another small mystery. I.E. What was the Denon 4520CI doing to the already good subwoofer response? It put a terrible dip at the 70hz area, as did the Anthem (right in that sweet kick drum range)... derp...
Dirac clearly did the best in our test with Sub EQ.

I'm sorry about the multiple edits, I wrote this initially on my phone, but transferred to PC to finish, and it needed a lot of wordsmithing and corrections. I'll probably drop the topic from here as it's not particularly relevant to the x4700h discussion - just an interesting sidebar based on a couple posts I saw.

 

[RichB]

Do note the Denon and Sherbourne (Trinnov) units were right at about the same SPL levels. Which was interesting. The SPL levels sort of fell into groups, but definitely not all the same.

We let each AVR auto EQ routine do its thing, and then I turned the main volume to -12dB from its reference setting on each unit (just to protect the ears of the setup/helper guys in the back of the room while doing post calibration sweeps) and so that chart is a recorded a frequency response sweep of the track 2 omnimic disk at -12dB on all processors post auto EQ routine.

Even if I use the software to match up the overlays at the 800-1000hz range, the corrected FR response were still all over the map!
Pic here:

One would have hoped all of those units would have been around 73dB (85dB - 12dB) (+- cheap bundled mic variance at 800 to 1000hz (typical frequency band used to measure mainspeaker SPL). They weren’t.

As you can tell the results are laughable in the context that people assume once they’ve run autoeq they are getting a fairly equitable experience to anyone else. On top of this, we know Audyssey Mics have a +-2.5 to 3dB allowable tolerance, and I’m sure other company’s bundled mics have the same loose standards. That’s a six dB swing possible in allowed mic manufacturing alone!!!
Its nearly impossible to comprehend the nearly 20dB deltas we measured in the post calibration results of these products.

Amazingly bad really.

Post 199 is the setup config
Post 200-207 documents each setups specifics
Post 208 is some followup analysis

This is excellent information. REQ without follow-up measurements is clearly a crapshoot.

Many run REQ until they like the result thinking they have achieved that the subjective assessment is an objective result.
Stepping back, it is clear that running something until your like it is just fine, but not the epitome of science.
Not only do supplied MICs wildly vary but small changes in position can cause huge swings.
It may be the authors of REQ know this, so after measurements would be difficult to correlate above Schroeder.
After all, measurements can be performed to generate filters, they can be run after.

- Rich

 

[Baff]

Pic here:
...
Post 199 is the setup config
Post 200-207 documents each setups specifics
Post 208 is some followup analysis

I think you forgot to add some links?

 

[GAmbrose]

First poster.

Will anyone be sending Amirm a 6700H to test against the 4700H?

I'd be interested to know the difference. The 6700H is assembled in Japan I believe?

 

[Caligari]

Jitter I have been running. Multitone most of the time is not possible because it is encoded at 192 kHz. Most Toslink/Coax interfaces stop at 96 kHz so the test won't run.

Are you going to rest DIRAC?

 

[Chromatischism]

Not quite
Reference is suppose to be 85db with 105dB peaks for mainspeakers and 115dB peaks for subs.
http://www.acousticfrontiers.com/2013314thx-reference-level/

That is correct - I used the wrong numbers. I referenced that site, even, don't know why I did that. You'll see in a subsequent post I got it right.

 

[peng]

Do note the Denon and Sherbourne (Trinnov) units were right at about the same SPL levels. Which was interesting. The SPL levels sort of fell into groups, but definitely not all the same.

We let each AVR auto EQ routine do its thing, and then I turned the main volume to -12dB from its reference setting on each unit (just to protect the ears of the setup/helper guys in the back of the room while doing post calibration sweeps) and so that chart is a recorded a frequency response sweep of the track 2 omnimic disk at -12dB on all processors post auto EQ routine.

Even if I use the software to match up the overlays at the 800-1000hz range, the corrected FR response were still all over the map!
Pic here:

One would have hoped all of those units would have been around 73dB (85dB - 12dB) (+- cheap bundled mic variance at 800 to 1000hz (typical frequency band used to measure mainspeaker SPL). They weren’t.

As you can tell the results are laughable in the context that people assume once they’ve run autoeq they are getting a fairly equitable experience to anyone else. On top of this, we know Audyssey Mics have a +-2.5 to 3dB allowable tolerance, and I’m sure other company’s bundled mics have the same loose standards. That’s a six dB swing possible in allowed mic manufacturing alone!!!
Its nearly impossible to comprehend the nearly 20dB deltas we measured in the post calibration results of these products.

Amazingly bad really.


https://www.avsforum.com/forum/90-r...vember-8-2014-kansas-city-7.html#post28900602
Post 199 is the setup config
Post 200-207 documents each setups specifics
Post 208 is some followup analysis

When we realized the volumes were so very different post autoeq calibration we realized we couldn't do a listening test like that, so we manually calibrated the main volume using the Disney + calibration disk test tones on each AVR to 72dB on the front left speaker for our subjective listening test.

This Pic is more like what we actually auditioned them at after manually level matching, I used the omnimic software to show the same basic thing we did with the Disney + disk here: You can see when you manually adjust for SPL the results aren't horrendous, but still far from a ideal standard. This info is in post 208. Even manually level matched at 1Khz You still have 20dB swings in the bass section below 100hz, and 5-10 db swings top to bottom everywhere else.

View attachment 70321
^
In the picture's legend ( in front of processor name) you can see the dB offsets I had to use in omnimic in order to get the different units to roughly line up at 1Khz.


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


The subs FR response is particularly strange. We had 4 sealed 18's, and they were at a good starting place to begin with! How the autoeq routeines screwed that up is another small mystery. I.E. What was the Denon 4520CI doing to the already good subwoofer response? It put a terrible dip at the 70hz area, as did the Anthem (right in that sweet kick drum range)... derp...
Dirac clearly did the best in our test with Sub EQ.
View attachment 70323

I'm sorry about the multiple edits, I wrote this initially on my phone, but transferred to PC to finish, and it needed a lot of wordsmithing and corrections. I'll probably drop the topic from here as it's not particularly relevant to the x4700h discussion - just an interesting sidebar based on a couple posts I saw.

I can only compare my AV8801 and AVR-X4400H, but I can tell you the curves were quite similar. Both could manage within +/- 3 to 3.5 dB post Audyssey from 20 to 120 Hz or higher and +/- 1 to 1.3 dB after manual adjustments, using the App for the Denon and PEQ/sub crawling with the Marantz (not compatible with the App).

Here's one of my latest plot:

I know we are talking about apple and orange, but if you can get it that flat, then you can be make they can't, and won't vary much regardless of which AVR is used, assuming we are comparing only models that come with XT32, or only models with YPAO, or AARC, Dirac Live etc.

 

[Standing in the rain]

I can understand your point of view. The language you used to describe the product wasn't an olive branch. Be honest with yourself. The words you chose weren't meant to invite a collaboration to improve the product. I can admit I was wrong about your intentions and motivations. You just want a better engineered AVR. But what if your reviews actually do more harm than good and more companies go under. What will the little guy be able to afford then? Just a thought.

Topping !!!

 

[amirm]

I can understand your point of view. The language you used to describe the product wasn't an olive branch.

Since someone just brought this up, I am not in olive branch business. The review is already quite helpful to manufacturers, providing data and measurements that they probably did not have. Post review, I am more than willing to help explain things, run more tests, review other samples, etc. as I have done countless times. But that doesn't mean I write the review to cater to the needs of the manufacturer. Or make them feel good despite poor performance of the device otherwise.

The review itself is as a service to the membership. I test what they send to me and report it for others to also read. My obligation is to the membership and owner of the device. It is not to manufacturers.

People who get loaners as a way of getting all test products are in a different boat and constraints. They are not as free as I can be with my testing and language I use. If people don't like what I do, there are plenty of other reviewers who cater to manufacturers in the way you suggest.

 

[Dj7675]

Since someone just brought this up, I am not in olive branch business. The review is already quite helpful to manufacturers, providing data and measurements that they probably did not have. Post review, I am more than willing to help explain things, run more tests, review other samples, etc. as I have done countless times. But that doesn't mean I write the review to cater to the needs of the manufacturer. Or make them feel good despite poor performance of the device otherwise.

The review itself is as a service to the membership. I test what they send to me and report it for others to also read. My obligation is to the membership and owner of the device. It is not to manufacturers.

People who get loaners as a way of getting all test products are in a different boat and constraints. They are not as free as I can be with my testing and language I use. If people don't like what I do, there are plenty of other reviewers who cater to manufacturers in the way you suggest.

This is what makes this place unique and so valuable. Gear that members send in that want to know how it measures. Free from manufacturer pressure of free samples and/or advertising. The results are what they are. The best manufacturers investigate the found issues and comment on them, fix them (if fixable via firmware etc), or learn from what was found to make the next generation of product better. Best case consumers get better performing products... worst case manufacturers are identified that show they are not interested in the results. Look forward to Denon's response.