Marantz 8805 review - very poor measurements

[PENG]

@PENG should chime in. He actually discussed this with Marantz engineers, and they don't dispute the measurements.

I just did, sorry for being late, i.e., long after ADTG launched his quick friendly attack already..

 

[PENG]

I would be a little "concerned" about the whole X3500/4500/6500/8500 series until we see new measurements.

Why would you start worrying about the higher noise originated from a tone at 35 kHz due to the Marantz choosen slow roll off DAC filter setting? I thought you DAC chips are so matured now they are no longer the weak link in a modern AVP/AVR. The Denons do have the same DAC and circuitry but not necessarily the same filter setting. One of my external DAC has 7 selectable filter settings, 4 for PCM, 3 for DSD. I couldn't hear any difference between them, and that's way before I read that 8805 review/measurements so I was biased, just bad hearing I guess. Still, as I always wonder, how come my imperfect hearing ability can hear differences between different headphones, speakers, and good, bad recordings, yet not so much between my DACs and amps.

 

[AcuDefTechGuy]

Why would you start worrying about the higher noise originated from a tone at 35 kHz due to the Marantz choosen slow roll off DAC filter setting? I thought you DAC chips are so matured now they are no longer the weak link in a modern AVP/AVR. The Denons do have the same DAC and circuitry but not necessarily the same filter setting. One of my external DAC has 7 selectable filter settings, 4 for PCM, 3 for DSD. I couldn't hear any difference between them, and that's way before I read that 8805 review/measurements so I was biased, just bad hearing I guess. Still, as I always wonder, how come my imperfect hearing ability can hear differences between different headphones, speakers, and good, bad recordings, yet not so much between my DACs and amps.

That’s why I put quotations around “concerned”, which means I’m not really concerned or worried, but a little irked about the “poor measurement review” until I see new SNR, Crosstalk, THD, FR measurements from AH to put this “poor measurement” to rest” .

I am not used to Amir’s measurements, so all this talk about “poor measurements” annoys me.

As for Frequency Range, I only care about 25Hz-18kHz, which is probably what I can actually hear. Don’t care about 22.5kHz and sure as heck not 35kHz.

I care exactly zero percent about DACs and know they all sound the same. So I don’t know what to make of the fact that Amir measures a ton of DAC. Does he think they all DAC sound different?

But this talk of “poor measurements” annoys me, and I would like to see the typical SNR, XTalk, THD, FR from AH to rid this bad “taste”.

So what I want to know is, how would you extrapolate Amir’s measurements to the typical S&V SNR, Crosstalk, FR, and THD?

Is it like the AV8802 which has “great measurements”:
SNR: 130dBA
Crosstalk: 92dB
FR: 20Hz-20kHz -0.04dB/-0.01dB
THD: 0.008%

 

[PENG]

That’s why I put quotations around “concerned”, which means I’m not really concerned or worried, but a little irked about the “poor measurement review” until I see new SNR, Crosstalk, THD, FR measurements from AH to put this “poor measurement” to rest” .

I am not used to Amir’s measurements, so all this talk about “poor measurements” annoys me.

As for Frequency Range, I only care about 25Hz-18kHz, which is probably what I can actually hear. Don’t care about 22.5kHz and sure as heck not 35kHz.

I care exactly zero percent about DACs and know they all sound the same. So I don’t know what to make of the fact that Amir measures a ton of DAC. Does he think they all DAC sound different?

But this talk of “poor measurements” annoys me, and I would like to see the typical SNR, XTalk, THD, FR from AH to rid this bad “taste”.

So what I want to know is, how would you extrapolate Amir’s measurements to the typical S&V SNR, Crosstalk, FR, and THD?

Is it like the AV8802 which has “great measurements”:
SNR: 130dBA
Crosstalk: 92dB
FR: 20Hz-20kHz -0.04dB/-0.01dB
THD: 0.008%

Agreed, may be you should let him know what some of us feel about his narrative part that is misleading to speed readers who may inadvertently (can you be that diplomatic though) easily take him out of context. When I shared the Marantz response with him, he even said he understood why Marantz did what they did, but felt they should give users the option.

Regarding the 8802's, if you give one to Amir to measure, I bet he would put it through the similar battery of tests, or find another one to reveal similarly "poor" performance lol. Imo the guy is super technical/objective oriented, and would probably pay little attention to the "audibility" thing. I wonder if he has tested any tube amps but I am going to find out just to see what he would have said about their performance.

By the way, the great 8802 was still using the LSI chip that Dr. Rich said was a real bottleneck as they don't perform better than all but the least expensive DACs. I posted the link to his article a few times before but may be still worth repeating what he said about the 8801:

"A key takeaway: circuit quality in the direct mode (stereo or 7.1) is almost always invariant to AVR prices in the range of $400 to $2,000. As examples, the $250 Yamaha RX-V367 and Marantz AV8801 ($3000) use the same Renesas LSI chip (R2A15220FP). With the LSI analog chip in these products, the sound of the direct mode is relatively constant, although a more robust power supplies, addition a quality output buffer and enhanced DC blocking capacitor quality can make small differences. "
https://hometheaterhifi.com/technical/technical-reviews/options-by-supplier-and-price/

That almost sounded like you talking..

Coming from a PhD in EE, that was eye opening in a sense, that even such a low cost chip (Large scale integrated chip) preamp/vol control could result in such great measurements, yet people still worry so much about DACs and swear they were hearing different kind of sound(s) even just by changing the filer setting between sharp, slow roll off etc. He was wrong (imo) about integrated amps though, but I think it was only because he didn't realize there were so many entry level integrated amps that were like what you and I called "Receiver derived", such as Yamaha low end ones.

 

[AcuDefTechGuy]

Agreed, may be you should let him know what some of us feel about his narrative part that is misleading to speed readers who may inadvertently (can you be that diplomatic though) easily take him out of context. When I shared the Marantz response with him, he even said he understood why Marantz did what they did, but felt they should give users the option.

Regarding the 8802's, if you give one to Amir to measure, I bet he would put it through the similar battery of tests, or find another one to reveal similarly "poor" performance lol. Imo the guy is super technical/objective oriented, and would probably pay little attention to the "audibility" thing. I wonder if he has tested any tube amps but I am going to find out just to see what he would have said about their performance.

By the way, the great 8802 was still using the LSI chip that Dr. Rich said was a real bottleneck as they don't perform better than all but the least expensive DACs. I posted the link to his article a few times before but may be still worth repeating what he said about the 8801:

"A key takeaway: circuit quality in the direct mode (stereo or 7.1) is almost always invariant to AVR prices in the range of $400 to $2,000. As examples, the $250 Yamaha RX-V367 and Marantz AV8801 ($3000) use the same Renesas LSI chip (R2A15220FP). With the LSI analog chip in these products, the sound of the direct mode is relatively constant, although a more robust power supplies, addition a quality output buffer and enhanced DC blocking capacitor quality can make small differences. "
https://hometheaterhifi.com/technical/technical-reviews/options-by-supplier-and-price/

That almost sounded like you talking..

Coming from a PhD in EE, that was eye opening in a sense, that even such a low cost chip (Large scale integrated chip) preamp/vol control could result in such great measurements, yet people still worry so much about DACs and swear they were hearing different kind of sound(s) even just by changing the filer setting between sharp, slow roll off etc. He was wrong (imo) about integrated amps though, but I think it was only because he didn't realize there were so many entry level integrated amps that were like what you and I called "Receiver derived", such as Yamaha low end ones.

Yeah the fascination with DAC-sound-signature is probably as bad as the infatuation with Amp-sound-signature.

As long as I can remember, dating back to The Audio Critic, most reviews are relatively similar with the usual 4 specs: SNR, Crosstalk, THD, and FR.

Do most people measure with 2.83V? And Amir measured at 4V?

 

[PENG]

Yeah the fascination with DAC-sound-signature is probably as bad as the infatuation with Amp-sound-signature.

As long as I can remember, dating back to The Audio Critic, most reviews are relatively similar with the usual 4 specs: SNR, Crosstalk, THD, and FR.

Do most people measure with 2.83V? And Amir measured at 4V?

I think most measure at lower voltage. 4V isn't bad because he use the XLR output. He also used input voltage higher than others, just going by memory now, will have to double check.

We should encourage him as there aren't many doing measurements any more. In fact, aside from him and Gene, I can't think of anyone else. Audiovision is like S&V, but not as good.

 

[AcuDefTechGuy]

I think most measure at lower voltage. 4V isn't bad because he use the XLR output. He also used input voltage higher than others, just going by memory now, will have to double check.

We should encourage him as there aren't many doing measurements any more. In fact, aside from him and Gene, I can't think of anyone else. Audiovision is like S&V, but not as good.

Yeah, we should encourage Amir to continue doing measurements. But someone needs to figure out how to extrapolate his measurements to what AH does or at least what S&V did.

2.83V, 1 watt, 20Hz-20kHz, 0.1% and 1% THD. SNR, Crosstalk, FR, THD.

I know it’s his world and his time and money, but we can’t compare apples to apples if he does one thing and everyone else does another.

 

[PENG]

Yeah, we should encourage Amir to continue doing measurements. But someone needs to figure out how to extrapolate his measurements to what AH does or at least what S&V did.

2.83V, 1 watt, 20Hz-20kHz, 0.1% and 1% THD. SNR, Crosstalk, FR, THD.

I know it’s his world and his time and money, but we can’t compare apples to apples if he does one thing and everyone else does another.

What you quoted S&V was obviously for integrated amps, receivers, AVRs, not AVP.

Regardless, you are right, until they get their act together, it is very difficult to compare such review measurements done by S&V,, AH, Hometheaterhifi and Audiosciencreview.com in terms of apples to apples.

S&V, Hometheaterfifi.com both tested the AV8802A, and if you try to compare their result with Audiosciencereview.com's results on the updated/upgraded AV8805 (good luck..) you have to be very careful because for the most part they are not apples to apples. For example, S&V seemed to have used the multichannel analog inputs, while hometheaterhifi.com used both analog inputs and HDMI input. Audiosciencereview.com used S/PDIF and HDMI inputs. As well, they were all over the map in terms RCA unbalanced vs XLR balanced, or both for different tests, too confusing for me to follow but you are welcome to try.

Regardless, they all show THD+N in within the audio band being 0.00XYZ %, so one may ask a legit question, i.e., who cares? Amir also commented that pure direct made no difference to the measured results! Try telling the golden ears group that and see how they react..

Edit: Before someone argue that it depends on the types and orders of the hamonics, let me add that the reason why I said "who cares?", was because if we're talking about THD in the third decimal places, then whether the distortions were mainly crossover distortion (not practically possible), and/or high order harmonics like 5th and up, also practically not possible for well designed popular class AB and D amps.

 

[PENG]

I am more used to seeing the typical SNR, Crosstalk, FR, THD.

Can anyone interpret the measurements that Amir did into what we would normally see?

He did not do XT and FR on the Marantz, but I suspect Amir has no doubt about the AV8805's performance in those two aspects. You would have to ask him if you want to be sure.

He did measure SNR and THD and commented that those bad stuffs found were not audible.

I noticed that some of the reviews (like Yamaha MX-A50 got SNR of 96dB at 5W 22.5kHz BW) show SNR. I assume it is Unweighted since it didn't say "dBA" or "A-weighted"?

I would bet it is not A-weighted but I would also bet he had the AES17 band pass filter on like the one he did on the AV8805 (111 dB). He should have stated the conditions clearly and be consistent about SNR (WX-A50) vs DNR (AV8805). For now, I assume they are directly comparable.

What is the SNR 22.5kHz BW @ 5 watts for the Marantz AV8805?

It can't be 45dB (when we factor in the THD beyond 5kHz),can it?

About 110 dB, the 45 dB only applied when he increased BW to 90 kHz, otherwise that huge "noises" spike (Marantz calledd it quantization aliasing noise )" in the neighborhood of 34 kHz (input signal) would otherwise not be visible when the input signal > 5 kHz. Again, according to Marantz, that isn't a "bug", it is their "sound policy", to select slow roll-off filter on the DAC.

Is "THD" really > 1% above 10kHz?

No, but THD+N is, if you use BW > around 34 kHz Practically speaking, we rarely see DACs specifying 20 - >30 kHz, let alone amplifiers.

 

[AcuDefTechGuy]

He did not do XT and FR on the Marantz, but I suspect Amir has no doubt about the AV8805's performance in those two aspects. You would have to ask him if you want to be sure.

He did measure SNR and THD and commented that those bad stuffs found were not audible.



I would bet it is not A-weighted but I would also bet he had the AES17 band pass filter on like the one he did on the AV8805 (111 dB). He should have stated the conditions clearly and be consistent about SNR (WX-A50) vs DNR (AV8805). For now, I assume they are directly comparable.



About 110 dB, the 45 dB only applied when he increased BW to 90 kHz, otherwise that huge "noises" spike (Marantz calledd it quantization aliasing noise )" in the neighborhood of 34 kHz (input signal) would otherwise not be visible when the input signal > 5 kHz. Again, according to Marantz, that isn't a "bug", it is their "sound policy", to select slow roll-off filter on the DAC.

No, but THD+N is, if you use BW > around 34 kHz Practically speaking, we rarely see DACs specifying 20 - >30 kHz, let alone amplifiers.

SNR of 110dB Unweighted is excellent.

So Amir basically measured above parameters/frequencies that are inaudible?

If he had measured just 20Hz-20kHz bandwidth, not 35kHz or 90kHz, he wouldn't see any "poor measurements".

Is there any rationale for measuring above 20kHz?

 

[Irvrobinson]

Yeah, we should encourage Amir to continue doing measurements. But someone needs to figure out how to extrapolate his measurements to what AH does or at least what S&V did.

2.83V, 1 watt, 20Hz-20kHz, 0.1% and 1% THD. SNR, Crosstalk, FR, THD.

I know it’s his world and his time and money, but we can’t compare apples to apples if he does one thing and everyone else does another.

I think your post brings up a concern that I've had and stated for years: measured data takes technical knowledge to interpret and place in context. I like seeing different sites using different test conditions because the different conditions can be revealing in ways that only one set of tests wouldn't be. (In the database computing world, for example, for years database software and systems vendors published benchmark results only for tests specified by the Transaction Processing Council (e.g. TPC-H, TPC-C, etc.),and vendors put optimizations in their products specifically for those tests, so over time they became much less useful for predicting actual enterprise application performance differences.) You seem to be looking for one site to validate another's results, and to be able to compare test results from different products and different testers. The challenge is that unless every test site used the same brand and model of test equipment, and tested under standard conditions, you still wouldn't get what you want. And you do seem to like figures of merit, which are IMO of limited usefulness for anything but casual discussion. To me, the most interesting data on AH or ASR are the broad spectrum analyses like this one:

And AH, ASR, and Stereophile have published them. In the infamous ASR 8805 review the measured data makes it clear where the issues really were, in the ultrasonic range, and everyone can decide for themselves how they apply the data to a buying decision. Amir published his opinion of the test results, but it was just an opinion. Marantz engineers clearly optimized for different results than Amir was looking for. Not every tester is going to have the same objectives or aim at the same audience. AH seems to place more emphasis on education and inclusion of non-experts, while ASR aims specifically at engineers and technologists.

 

[PENG]

SNR of 110dB Unweighted is excellent.

Not unweighted, but band pass limited using the AES17 filter. If "A" weighted, that 111 dB may improved to 130, like what you quoted for the AV8802, that S&V measured.

So Amir basically measured above parameters/frequencies that are inaudible?

Its not the first time he has done that..

Didn't you notice what he said about that graph (highlighted for you)?

What the heck? The moment we go past 5 kHz, the distortion+noise shoots sky high. It is off the chart literally post 10 kHz. What is going on here? We are using 90 kHz bandwidth so lots of things may be going on in ultrasonic range that is not visible here. Let's tease them out using high-resolution FFT spectrum:

https://www.audiosciencereview.com/forum/index.php?threads/review-and-measurements-of-marantz-av8805-av-processor.6926/ :

We now have our answer but not why. We have a tone around 34 kHz that is just -45 dB down from our main signal. If we were to run the dashboard using the same 10 kHz tone and 90 kHz bandwidth, the SINAD would drop to just 45 dB! The level of that peak fortunately is proportional to the signal. Since in real content there is not a ton of amplitude at 10 kHz and higher, practical impact is not as high but let's dig in to find out what is going on.

If he had measured just 20Hz-20kHz bandwidth, not 35kHz or 90kHz, he wouldn't see any "poor measurements".

That's my understanding.

Is there any rationale for measuring above 20kHz?

Yes, because people like me would want to see (not/can't hear) what the heck is going on up there.

Many "typical" (something like the Lyndorf so called 100% digital may be one of the exceptions) class D amps have output filters to attenuate the ultrasonic frequencies/switching noise too, but you know that already right?

 

[AcuDefTechGuy]

you do seem to like figures of merit, which are IMO of limited usefulness for anything but casual discussion. To me, the most interesting data on AH or ASR are the broad spectrum analyses

Are the differences in Broad Spectrum Analyses more "audible" than the typical SNR, Crosstalk, FR, THD figures?

Like you said - especially when you measure above the audible range, which is probably above 18kHz for many adults.

 

[AcuDefTechGuy]

1. If "A" weighted, that 111 would likely be improve to 130 (SNR).

2. "We have a tone around 34 kHz that is just -45 dB down from our main signal. If we were to run the dashboard using the same 10 kHz tone and 90 kHz bandwidth"


3. All "typical" class D amps have output filters to filter out the ultrasonic frequencies/switching noise too, but you know that already right?

1. So extrapolating Amir's SNR to Sound and Vision Magazine's standards, the AV8805's SNR is 130dBA, which is superb. And the lab results are "awesome", not "poor" if we are going with S&V lab standard.

2. Yeah, I guess since I am not an "Engineer or Technologist", I don't really care for any frequencies I can't hear. I think it's weird when I see speaker frequencies that read "20Hz - 40kHz". I guess I don't care for anything I can't SEE either, but I do love it when the display reads "4K" or soon "8K".

3. No, I did not know that most class-D amps filter out frequencies above 20kHz, but that's all good to me.

 

[Irvrobinson]

Are the differences in Broad Spectrum Analyses more "audible" than the typical SNR, Crosstalk, FR, THD figures?

Like you said - especially when you measure above the audible range, which is probably above 18kHz for many adults.

It's not about audibility per se, it's about where the artifacts are in the frequency spectrum and what the magnitudes are. And knowing what the magnitudes mean. When you see equipment with harmonic distortion spiking at most frequencies at not more than 100db below the fundamental, as in the Yamaha graph I copied from an AH review, that is an indicator of a low-distortion implementation. Just an indicator, but a very rich one.

 

[AcuDefTechGuy]

I do see Irv's point about companies trying to "optimize" these specs (SNR, Crosstalk, FR, THD).

But who's to say what's the best measurement specs?

What's more audible? Spectrum Analysis vs SNR/Crosstalk/THD/FR?

After everything is said, I would conclude that if we were trying to compare apples to apples (like comparing lab results to S&V magazine),then the lab measurements of the Marantz AV8805 is "Excellent".

So it just depends what "lab standards" and auditory frequencies you want to focus on or your point of view.

 

[PENG]

@AcuDefTechGuy , just for you, I searched out the SNR of the AVR-X5200W that Gene measured:
https://www.audioholics.com/av-receiver-reviews/denon-avr-x5200w/measurements

He said "With the receiver in Direct mode bypassing the DSP and 100mVin, and 1 Vrms output, the AVR-X5200W exhibited an excellent SNR of 95dB with the AES17 filter engaged. With no filter, it dropped down to 88dB, which is still an excellent figure. "

And below is what he said about the AVR derived (armor jacket on) Yamaha A-S801:
https://www.audioholics.com/amplifier-reviews/yamaha-a-s801-amplifier-review/yamaha-a-s801-measurements

"With a 200mVrms input, I measured > 89dB at 1 watt output and >93dB with A-wt filter engaged. This is a great result and demonstrates why I felt the noise floor on this product was completely inaudible."

Mind you the Yamaha, being an integrated amp, has a CD direct input that improves SNR to 99 dB, A-wt.

Note that even both were measured by AH/Gene, you still have to be careful if you want to compare the results apples to apples. For example, the Denon AVR's 95 dB SNR vs the Yamaha's integrated amp were measured using different input voltage, 100 mV for the Denon and 200 mV for the Yamaha. So that puts the Yamaha at advantage.

 

[AcuDefTechGuy]

It's not about audibility per se, it's about where the artifacts are in the frequency spectrum and what the magnitudes are. And knowing what the magnitudes mean. When you see equipment with harmonic distortion spiking at most frequencies at not more than 100db below the fundamental, as in the Yamaha graph I copied from an AH review, that is an indicator of a low-distortion implementation. Just an indicator, but a very rich one.

So....for non-engineers/technologists, what's the takeaway message?

If none of these measurements (spectrum analysis, SNR, etc.) are really audible, who's to say what's more important?

Inaudible Harmonic distortion or inaudible low-distortion implementation is more important than inaudible SNR, Crosstalk, FR, and THD?

 

[AcuDefTechGuy]

@AcuDefTechGuy , just for you, I searched out the SNR of the AVR-X5200W that Gene measured:
https://www.audioholics.com/av-receiver-reviews/denon-avr-x5200w/measurements

He said "With the receiver in Direct mode bypassing the DSP and 100mVin, and 1 Vrms output, the AVR-X5200W exhibited an excellent SNR of 95dB with the AES17 filter engaged. With no filter, it dropped down to 88dB, which is still an excellent figure. "

And below is what he said about the AVR derived (armor jacket on) Yamaha A-S801:
https://www.audioholics.com/amplifier-reviews/yamaha-a-s801-amplifier-review/yamaha-a-s801-measurements

"With a 200mVrms input, I measured > 89dB at 1 watt output and >93dB with A-wt filter engaged. This is a great result and demonstrates why I felt the noise floor on this product was completely inaudible."

Mind you the Yamaha, being an integrated amp, has a CD direct input that improves SNR to 99 dB, A-wt.

Note that even both were measured by AH/Gene, you still have to be careful if you want to compare the results apples to apples. For example, the Denon AVR's 95 dB SNR vs the Yamaha's integrated amp were measured using different input voltage, 100 mV for the Denon and 200 mV for the Yamaha. So that puts the Yamaha at advantage.

That's why we need engineers and technologists to interpret some of these results.

Now we just need AH to measure every AVR, AVP, and Amp so I don't have to go reading other sites.

 

[PENG]

3. No, I did not know that most class-D amps filter out frequencies above 20kHz, but that's all good to me.

Not really, class D amp output filters filter the switching noises that are at very high frequencies so their output low pass filters "cut-off" point would likely be much higher than 20 kHz or even 30 kHz. I know know much about class D amps, we can always ask google.