Denon AVR-X4800H AVR Review

[ivo.f.doma]

The difference in SINAD between the 5100,5101 and 5102 are negligible, but those are specified as "typical", so a few dB better as measured on ASR is believable.

Various posters have speculated why Denon/Marantz could achieve measured SINAD a few dB (not 10, don't know you saw that) better than the typical 93 dB specified by TI. Regardless, it at least shows that D+M has done an excellent job in implemented this chip.

I admit, the difference in Amir's measurements is 8dB, not 10.

https://www.ti.com/lit/ds/symlink/pcm5102a.pdf?ts=1679558161169&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FPCM5102A

See page 4. There, Sinad is worse than measured by Amir.

 

[peng]

I admit, the difference in Amir's measurements is 8dB, not 10.

https://www.ti.com/lit/ds/symlink/pcm5102a.pdf?ts=1679558161169&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FPCM5102A

See page 4. There, Sinad is worse than measured by Amir.

93 vs 95, that's 2 dB, how did you get 8?

 

[ivo.f.doma]

93 vs 95, that's 2 dB, how did you get 8?

3800 vs 4800... 87 vs 95dB, there is an incomprehensible difference for me

 

[ban25]

Another concern of mine about 3800H/4800H is that they do not support native playback of DSD because PCM 5102A does not.
Does nobody mind this point?
I mind this point because I have SACDs.

My A14 MKII doesn't either. When moving from AKM to TI, we lost DSD but gained MQA. I listen to quite a bit of Tidal, so I'm happy to have native MQA unfolding. For DSD, I just use Roon / HQPlayer or if I'm lazy, just Roon.

 

[peng]

3800 vs 4800... 87 vs 95dB, there is an incomprehensible difference for me

Okay, I thought you were comparing the dac spec vs the 4800's measured SINAD.

 

[David_Alexander]

Okay, I thought you were comparing the dac spec vs the 4800's measured SINAD.

Yes, I was just looking at the back and forth and thinking there was a crossed wire.
A couple of questions.

1. How can the implementation of a chip actually improve it performance wrt SINAD?
2. What is different in the architecture of the X4000 vs X3000 that drives the 8dB difference.
3. What are the tolerances of the measurements? I.E., if one by chance measured low on the day and one up, could this accentuate a difference?

 

[peng]

Yes, I was just looking at the back and forth and thinking there was a crossed wire.
A couple of questions.

1. How can the implementation of a chip actually improve it performance wrt SINAD?
2. What is different in the architecture of the X4000 vs X3000 that drives the 8dB difference.
3. What are the tolerances of the measurements? I.E., if one by chance measured low on the day and one up, could this accentuate a difference?

I don't know but I can guess. Guessing doesn't help anyone though.

Tolerances of the chips are a real thing for sure so that could explain small differences in bench test results on the same dut. I cannot find any info on tolerances of the TI, or any dac chips, may be an IC specialist can shed some light on this, or you can contact TI for more info. They likely only respond to someone who has an account with them.

 

[techsamurai]

Yes, I was just looking at the back and forth and thinking there was a crossed wire.
A couple of questions.

1. How can the implementation of a chip actually improve it performance wrt SINAD?
2. What is different in the architecture of the X4000 vs X3000 that drives the 8dB difference.
3. What are the tolerances of the measurements? I.E., if one by chance measured low on the day and one up, could this accentuate a difference?

Are they using the same DACs?

 

[ban25]

Yes, I was just looking at the back and forth and thinking there was a crossed wire.
A couple of questions.

1. How can the implementation of a chip actually improve it performance wrt SINAD?
2. What is different in the architecture of the X4000 vs X3000 that drives the 8dB difference.
3. What are the tolerances of the measurements? I.E., if one by chance measured low on the day and one up, could this accentuate a difference?

The datasheet specifies an approximate THD+N of ~93 dB. It's probably rated conservatively. So, careful implementation, combined potentially with some IC binning could easily result in a difference of +/- 3 dB. Additionally, I read Amir's review to indicate that he changed his testing methodology in some way. That could have contributed to an overall higher result for the 4800 vs the 3800.

 

[RF Air]

So since it is also in the sound system in Volvo do they charge customers to get a license?

Dirac sends Audio Specialist that work with the Manufacturers that are developing Dirac set-up technology for vehicle sound packaging. After that, I think the set-up is fixed as a one-time tuning for the vehicle. Someone else may have more information.

 

[Alice of Old Vincennes]

It's like the computer and software industry that is trying to protect proprietary products. They have intellectual property and can only support their business operations with a strategy to protect it by licensing the Manufacturer of Dirac product technology/equipment and offering fee for access by End User License. Defending the product is key to the survival of their business. It will take competition and new rival technology for the pricing to be challenged. I don't know if there is anything on the horizon that can do this, especially when they have 20 years of development, patents and experience with the Dirac products.

If they were not successful in maintaining proprietary access with their product, they would have failed. And if the product was not regarded as exceptional and leading technologically, there would be no demand and other products could prevail. They have to continue to sink investment into protecting and building the product technology to remain relevant to the industries they serve.

It appears they are becoming a market leader and with a growing footprint in the market, pricing may come down to broaden their reach and access to customers. Adding the Sound United products is a step in the right direction to lower costs and access for the future. Perhaps the sales/market strategy will change over time if the "next best thing" was offered to market by a competitor, forcing Dirac to compete for sales and technological advantage.

Distribution through Sound United will bring prices down.

 

[tobby]

1. How can the implementation of a chip actually improve it performance wrt SINAD?
2. What is different in the architecture of the X4000 vs X3000 that drives the 8dB difference.
3. What are the tolerances of the measurements? I.E., if one by chance measured low on the day and one up, could this accentuate a difference?

1.) You can't really improve SINAD performance if it's limited by the performance of the chip. But you can decrease it, if you are for example doing a bad PCB design.
2.) So maybe the PCB(s) of the X4800H are well designed, those of the X3800H not (less optimization). Or maybe it was just good/bad luck - a sample of the DAC which was above average in the X4800H and a sample below average in the X3800H. Or maybe they are binning the chips and use the better ones for the X4800H and the worse ones in the X3800H like they do in the market of CPUs and GPUs.
3.) That's a good question only amir may answer. We do know the measurements, but we don't know the standard deviation or variance.

Distribution through Sound United will bring prices down.

So you know about the calculations of the prices at Dirac and Denon and also about their internal contract? Interesting...

 

[tsammyc]

2.) So maybe the PCB(s) of the X4800H are well designed, those of the X3800H not (less optimization). Or maybe it was just good/bad luck - a sample of the DAC which was above average in the X4800H and a sample below average in the X3800H. Or maybe they are binning the chips and use the better ones for the X4800H and the worse ones in the X3800H like they do in the market of

I would lean to sample variation as Amir only tested one sample of each and that does not make for a general conclusion. In the 3800 review, it seems he did consult the Denon people and seemed to come away that they did not seem surprised by his 3800 results. Production in Vietnam may also be responsible for higher sample variation

 

[peng]

I would lean to sample variation as Amir only tested one sample of each and that does not make for a general conclusion. In the 3800 review, it seems he did consult the Denon people and seemed to come away that they did not seem surprised by his 3800 results. Production in Vietnam may also be responsible for higher sample variation

That could be true that the factory in Vietnam may not follow the QS procedures as strictly as the one in Japan but I doubt that because other places in Vietnam or China seems capable of producing AV devices that measured well above 100 dB SINAD.

There are many possible reasons but we will all be guessing. In reality the AVR-X3800H's results are actually very good, given that it has a low (relatively speaking only) SINAD DAC IC, take a look:

THD+N/-SINAD: Yes as pointed out, it is 8 dB lower in SINAD vs the 4800, and even more vs the 3700 AKM version, but if you examine the distortions, there are hardly any high order harmonics, it is dominated by the 2nd and the 3rd, so in terms of harmonic structure, this thing may even sound better than other devices that has SINAD >95 dB.

Looking at the actual SINAD vs output, users who don't need a ton of power to listen to reference level, the DAC is capable of between 90 to 95 dB SINAD, and again, given the harmonics are dominated by the 2nd (highest) and the 3rd, this AVR, in a DBT, may sound very similar to a $10,000 separates pair.

SNR/DR: The pre out measured about the same as the AVR-X3700H (the one that has the AKM DAC IC)

SNR at 5 W output: about the same as the AVR-X4700H, less than 3 dB difference vs the X6700H and Marantz SR8015, almost 10 dB better than the Emotiva bass X A500, almost a few dB better than the Outlaw 5000 and Outlaw M2220

So, to summarize:

The AVR-X3800H that has Amir measured SINAD <90 dB, can sound just as good as a pair of much more expensive separates because:

- Harmonics structure/profile shows little high order harmonics
- SNR better than some popular separate power amps, keep in mind it is an avr so the noise is made up of not just the power amp but also the preamp/dac
- SINAD, in apples-apples comparison, is more than 10 dB better than the Marantz AV7705 (two measured on ASR)

I guess it is not difficult to make the case, if we are to based our assessment on objective ASR measurements, that the AVR-X3800H, while not as good the 4800, is a decent unit at its price point. May be owners of the Marantz AV7705, SR7015 should have complained more, yet on ASR, they seem to say their units sound great, so what's the beef of the AVR-X3800H!!

Harmonic profiles AVR-X3800H, and AVR, vs AV7705, a so call separate:

Note: I traded in both of my Denon AVR and Marantz AVP already, so this post is just to show while I am very disappointed about Denon and Marantz's choice of the DAC IC, I feel like some of posts on ASR might have misled their owners to think their AVRs will not "sound good" vs others who own comparable AVRs at similar, or much higher price points.

Facts are facts, anyone who is concerned about the audible effects of the higher harmonic distortions and noise, just compare the much lower SINAD of the AV7705 at 2 V. Okay the graph shows 4 V but that's only because Amir measured it on the XLRs, if he had measured it on the RCAs, it would have been 2 V, and the number would likely be even worse, however slightly:

See Marantz's much higher harmonics at above 3rd, yet their owners heard musical, warm sound, so don't worry too much about 3800's SINAD and enjoy the movies and music, that's the point I am trying to make.

That does not excuse D+M's use of the lower spec DAC IC (i.e. regression...). It is actually a shame, because their engineering team obviously are capable of implementing the PCM5102 to the point they managed to milk that chip dry! Imagine if they had used another inexpensive ES chip such as the ES9010K2M that offers >10 dB SINAD!! So, as good as the X3800H and X4800H, people should still press them to use a better one in their next model. The Masimo presenter did sort of drop a hint, in the FAQ session in his video presentation, that the AKM chip is again available....

Edit: Amir measured two samples of the AV7705, both shows basically the same SINAD.

Here's the one for the second sample, a refurbished one, on Aug 2021:

 

[Brambo67]

That could be true that the factory in Vietnam may not follow the QS procedures as strictly as the one in Japan but I doubt that because other places in Vietnam or China seems capable of producing AV devices that measured well above 100 dB SINAD.

There are many possible reasons but we will all be guessing. In reality the AVR-X3800H's results are actually very good, given that it has a low (relatively speaking only) SINAD DAC IC, take a look:

THD+N/-SINAD: Yes as pointed out, it is 8 dB lower in SINAD vs the 4800, and even more vs the 3700 AKM version, but if you examine the distortions, there are hardly any high order harmonics, it is dominated by the 2nd and the 3rd, so in terms of harmonic structure, this thing may even sound better than other devices that has SINAD >95 dB.

Looking at the actual SINAD vs output, users who don't need a ton of power to listen to reference level, the DAC is capable of between 90 to 95 dB SINAD, and again, given the harmonics are dominated by the 2nd (highest) and the 3rd, this AVR, in a DBT, may sound very similar to a $10,000 separates pair.

SNR/DR: The pre out measured about the same as the AVR-X3700H (the one that has the AKM DAC IC)

SNR at 5 W output: about the same as the AVR-X4700H, less than 3 dB difference vs the X6700H and Marantz SR8015, almost 10 dB better than the Emotiva bass X A500, almost a few dB better than the Outlaw 5000 and Outlaw M2220

So, to summarize:

The AVR-X3800H that has Amir measured SINAD <90 dB, can sound just as good as a pair of much more expensive separates because:

- Harmonics structure/profile shows little high order harmonics
- SNR better than some popular separate power amps, keep in mind it is an avr so the noise is made up of not just the power amp but also the preamp/dac
- SINAD, in apples-apples comparison, is more than 10 dB better than the Marantz AV7705 (two measured on ASR)

I guess it is not difficult to make the case, if we are to based our assessment on objective ASR measurements, that the AVR-X3800H, while not as good the 4800, is a decent unit at its price point. May be owners of the Marantz AV7705, SR7015 should have complained more, yet on ASR, they seem to say their units sound great, so what's the beef of the AVR-X3800H!!

Harmonic profiles AVR-X3800H, and AVR, vs AV7705, a so call separate:

Note: I traded in both of my Denon AVR and Marantz AVP already, so this post is just to show while I am very disappointed about Denon and Marantz's choice of the DAC IC, I feel like some of posts on ASR might have misled their owners to think their AVRs will not "sound good" vs others who own comparable AVRs at similar, or much higher price points.

Facts are facts, anyone who is concerned about the audible effects of the higher harmonic distortions and noise, just compare the much lower SINAD of the AV7705 at 2 V. Okay the graph shows 4 V but that's only because Amir measured it on the XLRs, if he had measured it on the RCAs, it would have been 2 V, and the number would likely be even worse, however slightly:

See Marantz's much higher harmonics at above 3rd, yet their owners heard musical, warm sound, so don't worry too much about 3800's SINAD and enjoy the movies and music, that's the point I am trying to make.

That does not excuse D+M's use of the lower spec DAC IC (i.e. regression...). It is actually a shame, because their engineering team obviously are capable of implementing the PCM5102 to the point they managed to milk that chip dry! Imagine if they had used another inexpensive ES chip such as the ES9010K2M that offers >10 dB SINAD!! So, as good as the X3800H and X4800H, people should still press them to use a better one in their next model. The Masimo presenter did sort of drop a hint, in the FAQ session in his video presentation, that the AKM chip is again available....

Edit: Amir measured two samples of the AV7705, both shows basically the same SINAD.

Here's the one for the second sample, a refurbished one, on Aug 2021:

Very well said. I wouldn't be surprised at all if next iterations will include higher spec'd DAC's. For now personally I'll focus on optimizing my living room ;-)

 

[ivo.f.doma]

The datasheet specifies an approximate THD+N of ~93 dB. It's probably rated conservatively. So, careful implementation, combined potentially with some IC binning could easily result in a difference of +/- 3 dB. Additionally, I read Amir's review to indicate that he changed his testing methodology in some way. That could have contributed to an overall higher result for the 4800 vs the 3800.

That's exactly what I was pointing out, it seems to me that the measurement system was changed just to make the 4800 measure better. And I can't get it out of my head...
The single measurement of 4800 is not comparable to all others!

 

[ivo.f.doma]

That's exactly what I was pointing out, it seems to me that the measurement system was changed just to make the 4800 measure better. And I can't get it out of my head...
The single measurement of 4800 is not comparable to all others!

But again, it is true that the amplifiers on the 3800 measured better than the "individual" ones on the 4800.

 

[sask15]

Where did he say he was changing his methods for testing and that it could be favourable towards the 4800? Maybe I missed it. Thanks

 

[enricoclaudio]

For those in need of help with Dirac, here is the video recently posted by Masimo covering Dirac calibration process for D&M receivers and processors:

 

[peng]

That's exactly what I was pointing out, it seems to me that the measurement system was changed just to make the 4800 measure better. And I can't get it out of my head...
The single measurement of 4800 is not comparable to all others!

The changed method was for the SINAD vs output graph only. The SINAD at 2 V test was done the same way, no change.