• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of Marantz AV8805 AV Processor

Joined
Mar 22, 2019
Messages
73
Likes
11
That's easy, it's called marketing. Can't sell the new and improved products if you don't have something new to offer. How else you gonna make MONEY.
Well, what can I say ... successful marketing since 1996 when Elgar was introduced for the first time, and it is still in demand today (look for the prices around)? I would call it as a “successful engineering” instead, but it is up to you how to call it.
 

audioBliss

Active Member
Joined
May 7, 2019
Messages
117
Likes
111
Just found this site! Amazing stuff here I must say! I've always been skeptical of Marantz after owning a few lower end units and being disappointed with the sound. It would be quite interesting to see some reviews of Arcam or Anthem. Something like the Arcam AVR390 or AVR550.
 
Joined
Mar 22, 2019
Messages
73
Likes
11
First of all the comparable Anthem item is AVM60, not AV receivers Arcam AVR390 or AVR550.
Second - don’t you want to correct your measurements, Amir? They are misleading and people read them making the wrong conclusions. Then they go to the end of the thread posting their impressions based on the wrong measurements and your wrong suggestions. If you call your forum “audiovideoscience” it shall be somehow scientifical, without blaming stuff for no reason.
 

audioBliss

Active Member
Joined
May 7, 2019
Messages
117
Likes
111
First of all the comparable Anthem item is AVM60, not AV receivers Arcam AVR390 or AVR550.
Second - don’t you want to correct your measurements, Amir? They are misleading and people read them making the wrong conclusions. Then they go to the end of the thread posting their impressions based on the wrong measurements and your wrong suggestions. If you call your forum “audiovideoscience” it shall be somehow scientifical, without blaming stuff for no reason.
I merely meant anything "high-end" would be interesting. Probably not a lot of people willing to send in their receivers or processors for that matter but it would still be interesting imo to see measurements of products from other brands that people generally hold in higher regards. What are the issues with the measurements?
 

Sal1950

Major Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
6,709
Likes
4,729
Location
Central Fl
I merely meant anything "high-end" would be interesting.
How much do you have to spend to qualify for "high end"?
I would think a $4,500.00 preamp should have fit the bill?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
Second - don’t you want to correct your measurements, Amir? They are misleading and people read them making the wrong conclusions.
Correct what exactly?
 
Joined
Mar 22, 2019
Messages
73
Likes
11
I merely meant anything "high-end" would be interesting.
We are discussing Marantz AV processor here with MSRP $4,500. Anthem AVM60 AV processor’s price tag is $3,000, which is way less, also this processor can not be called “a high end”. Their Hi End (actually it is an initial high end level) is Anthem D2v 3D with the MSRP $9,500. So AVM60 is perfectly fine for comparison with AV8805.
 

audioBliss

Active Member
Joined
May 7, 2019
Messages
117
Likes
111
I see I wasn't making myself clear enough. With my quotes I meant the likes of AVM60 and AV860 but that if that was not possible then it would imo also be interesting to compare to other AVRs from Arcam and Anthem. For instance the lowest end Arcam AVR 390 is a $2500 product but it has almost all the same parts as the AVR550 and AV860. Anything in that range is interesting because of that.
 
Joined
Mar 22, 2019
Messages
73
Likes
11
Correct what exactly?
You didn’t read my posts, did you?
1. You have performed measurements only at 4.12V RMS output level, which is an overkill for a processor with officially max rated output at 2.4V RMS (in the specification). It us unclear why you have decided to use specifically overloading level for the tests, there is no reasonable cause.
Reviewer from HomeTheatreHiFi, who gave Marantz 8802 “Best of Award for the year 2015” (Marantz 8805 just has some additional features, the design is pretty much similar), performed the measurements properly, just before the max output level at 2.05V, and the level of distortions was WAY lower comparing to your measurements. In addition, they have performed measurements at 5V, which are comparable with yours. But it is not a standard mode for the processor.
2. You made some wrong conclusions regarding quality of the processor and based on the overloading level, for example:
Sad because this would be poor rating even for a $99 desktop DAC let alone a $5,000 processor.
THD+N is dominated by the third harmonic which means that it will peak above the noise floor of 16 bit CD content (96 dB dynamic range). So we lack transparency there let alone for high-resolution content.
but has too much distortion as the signal gets louder:
Considering that I kept the output level at 4 volts, i.e. well below max, the rising distortion is worrisome.
Correction: “way ABOVE max”
Next test, THD+N versus frequency was a shocker: 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?
There is no “heck”, it seems the processor clips earlier at frequencies above 5kHz, but again the level is completely wrong.
Thinking this may be an aliasing issue, I ran my white noise test where we look at how much filtering we get at 22.05 kHz. Theory demands that we get infinite amount of signal reduction at that frequency. This is what we get instead:
why would anyone expect to see a cut off at 22.05 kHz (44.1 kHz sampling rate) if 192 kHz sampling rate was used? It is obvious looking at the 90 (probable 96 kHz) graph. What you see is a slow roll off at 24 kHz, because the Audyssey processing (or all processing) was performed at 48 kHz. This article explains why: https://www.audioholics.com/room-acoustics/audyssey-room-eq-interview
I searched the manual for any DAC filter setting but did not find anything. This is flat out broken. I get wanting slow roll off but this smells like incorrectly programmed filter setting.
Correction : nothing is “broken”, nothing “smells”.
We get essentially the same performance. Then again, what we have here is limited by high harmonic distortion which is likely in the analog domain post the DAC so improvements elsewhere may not show up.
Then again, 4.13 V RMS vs 2.4V max per specification.
Fortunately there is nothing drastically wrong here, sans the DAC filtering. That aspect needs to be reported to Marantz as hopefully can be fixed with a firmware update.
Firmware update for what? To see 22.05 kHz roll off at 192kHz sampling rate or update the processing to 192 kHz?
I would say buy the Marantz AV8805 because it has the features it has not because you think it will provide reference quality audio performance. It will not.
Again, wrong conclusions based on the wrong measurement results.
Correction : see the review from HomeThetreHiFi, compare Marantz 8805 with other AV processor which supports more or less similar features (Audyssey processing requires a lot of processing power), and make new conclusions.

For information: input levels for power amplifiers:
Bryston 4b sensitivity at 1.4V for 250W at 8 ohm (it has switchable gain 23dB or 29dB).
Onkyo 5501 sensitivity 1V balanced, 2V unbalanced
Proceed AMP 5 sensitivity 2.24V for full rated output (balanced), 1.12V for full rated output (single-ended or unbalanced)

And so on ...
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
why would anyone expect to see a cut off at 22.05 kHz (44.1 kHz sampling rate) if 192 kHz sampling rate was used? It is obvious looking at the 90 (probable 96 kHz) graph.
You are misreading the measurement. Sampling rate is 44.1 kHz as clearly stated on the graph:



I just pulled up the project file that I saved. Here are the settings:

1557275060080.png


As you see, sample rate is clearly 44.1 kHz.

The input rate of AP is set to 624 kHz or bandwidth of 250 kHz. Out of this, I chose to plot 90 kHz as you see in the graph. Since analog data is being captured from the unit, I could set the input sampling rate to anything I want, independent of how the DAC's sample rate is being used.

As you say yourself, if the DAC was set to 192 kHz, then this would be fine (although then would be rolling off too early). But not when it is set to 44.1 kHz.

You agree this is a bug now, yes?
 
Joined
Mar 22, 2019
Messages
73
Likes
11
You are misreading the measurement. Sampling rate is 44.1 kHz as clearly stated on the graph:
And how come the frequency range of the window is 90 kHz, if it is a digital processing? To get that range you need 180 kHz sampling rate per Nyquist theorem, right?
Sorry, I see it now. You have measured the analog output using sampling rate 624 kHz for ADC in the analyzer and filtering everything above 90 kHz.
But it is not a bug if you will get that the output was at 24kHz frequency range ... They probably used ARC ...
 
Last edited:
Joined
Mar 22, 2019
Messages
73
Likes
11
As you say yourself, if the DAC was set to 192 kHz, then this would be fine (although then would be rolling off too early). But not when it is set to 44.1 kHz.

You agree this is a bug now, yes?
Ok I will tell you a secret - to do the Audyssey processing at 48kHz the AV processor downsamples (and probably upsamples in the case of 44.1kHz) the input signal to 48 kHz, no matter what was the sampling rate at the input. I am not sure what is the sampling rate with Audyssey turned off, but I suspect it is still 48kHz, or may be 96kHz.
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
And how come the frequency range of the window is 90 kHz, if it is a digital processing? To get that range you need 180 kHz sampling rate per Nyquist theorem, right?
Sorry, I see it now. You have measured the analog output using sampling rate 180 kHz for ADC in the analyzer.
But it is not a bug if you will get that the output was at 24kHz frequency range ... They probably used ARC ...
Again, the sampling rate of the Audio Precision input was 624 kHz with an available bandwidth of 250 kHz. I have just chosen to show the bandwidth up to 90 kHz in the graph. I could increase that to 300 kHz if I wanted (with reduced resolution):

1557276517793.png


You see the 250 kHz roll off now. Since our interest is in much lower frequencies, I choose to not show such a high bandwidth.
 
Joined
Mar 22, 2019
Messages
73
Likes
11
You see the 250 kHz roll off now. Since our interest is in much lower frequencies, I choose to not show such a high bandwidth.
Thanks, I understood already. Logically the FR bandwidth shall be limited at 22.05 kHz with pretty sharp roll off if you use the input sampling frequency 44.1kHz.
The reason why you’ve got higher FR with slow roll off could be something like Pioneer’s Legato Link http://www.soundfountain.com/amb/cd25pioneer.html - take a look at the last graph on the bottom of the table with graphs (resulting characteristic), doesn’t it look similar (except of the sharp deep, which could be masked in our case by upsampling or using other methods)? It is not a bug (a programming issue performed unintentionally), I would call it “a trick” ... they do it intentionally.

P.S. I have a question though - why on your measurements with this FR spectrum is 8802 while you supposedly tested 8805, as it is seems on the other graphs?
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
1. You have performed measurements only at 4.12V RMS output level, which is an overkill for a processor with officially max rated output at 2.4V RMS (in the specification). It us unclear why you have decided to use specifically overloading level for the tests, there is no reasonable cause.
For digital audio products, the nominal output for unbalanced output is 2 volts RMS and 4 Volts for balanced. This is the output level from just about every DAC with balanced output I have tested such as this SMSL SU-8:


This is how the Marantz did again:



As we see, using almost identical output voltage, the $250 SMSL has 20 dB lower distortion and noise. That is a massive gap in performance.

You complain that I am overdriving the Marantz. Not so. They are under-driving it to get their spec. Just like any amplifier, higher outputs can cause more distortion. That the Marantz picks a low output level is a cheat. I certainly can't use their SINAD at half the output level and compare it to the rest of the DACs I have tested at double the output voltage.

The Marantz is a multi-thousand dollar processor. You want to give me an excuse for why it can't match let alone outperform the $250 SMSL SU-8? Is the balanced output too hard for them to implement with proper 4 volt output as opposed to every no-name DAC manufacturer?

Think of my metrics as 0 to 60 miles/hour for car acceleration. I don't care if the car manufacturer decides to spec 0 to 40 m/h. I test at 0 to 60 as I have always done.

If you want to say their specs are right, go ahead. But that is not material. What is material is how well it can replicate the performance of a stand-alone DAC. And in that regard, it is not even coming close.

Note that by lowering the output voltage, these external DACs can actually produce better performance too. Here is an example showing this:



Using the nominal output of 2 volts, the SINAD is 108 dB. Note however that if I reduce the output by -2 dB, then it jumps up to 112 dB.

So no, it is not proper or fair to other products to lower our standard output voltage to match some company's spec. They need to have the engineering know-how to produce exceptional performance at 4 volt output using XLR. If they cannot as is the case with this Marantz, then they need to be dinged. And dinged they were. :)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
There is no “heck”, it seems the processor clips earlier at frequencies above 5kHz, but again the level is completely wrong.
Your assumption is incorrect. I dug into this in the review. First, the offending measurement:



The red line is a $99 DAC. Not matter how I run the Marantz, it better not produce the graph in blue. Going along anyway, this is one of the few measurements I have which uses wide bandwidth of 90 kHz. The reason for this is that we want to capture the harmonic distortion of 20 kHz which would be at 40 kHz, 60 kHz, 80 kHz and so on. With 90 kHz bandwidth, we capture the most significant contributors to THD+N.

To see what contributed to the sky high THD+N after 5 kHz, I analyzed the spectrum of the Marantz output using a 10 kHz tone:



As I note on the graph, the reason the THD+N is awful is not because of harmonic distortion but rather, the massive peak around 34 kHz as indicated by the arrow. THD+N is a catch-all measurement, including the energy of everything other than our test tone at 10 kHz. As such, it includes this ultrasonic tone whether it is harmonic distortion or not.

An ideal DAC would show nothing but our 10 kHz tone. Here we have both harmonic distortion and spurious tones. The latter is there because this is a complex device with video and other functionality. The engineering is poor, not providing proper isolation between these components resulting in these other tones. Measurements should have been performed and causes identified and removed.

In summary, your assertion that this is output buffer saturting is not correct. The design simply lacks hygiene and that is what the measurements capture. For this kind of money, I better get squeaky clean output, not what we see.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
We can see lack of proper isolation in jitter test:



Here is our cheap $250 Balanced DAC in the same test:


(sample rates are different).

This leaves no doubt that the fidelity of the DAC in the Marantz was not a focus of the design.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
24,230
Likes
46,039
Location
Seattle Area
Correction : see the review from HomeThetreHiFi, compare Marantz 8805 with other AV processor which supports more or less similar features (Audyssey processing requires a lot of processing power), and make new conclusions.
I can't find a review of Marantz AV8805 in there. They only have AV8802 which may be a different design. Regardless, they say right in the review: https://hometheaterhifi.com/reviews/receiver-processor/processors/marantz-av8802-processor-review/

1557288998473.png


You see in black and white that 2 volts is not proper for XLR output. And why they tested at 5 volt which is even higher than what I tested.

Bryston 4b sensitivity at 1.4V for 250W at 8 ohm (it has switchable gain 23dB or 29dB).
Not correct. From the manual:

1557289307143.png


You need 3.46 volts for full power in low gain mode, not 1.4 volt. Even in high gain you won't achieve full power at 1.4 volts (1.73 volts is needed).

Regardless, there are no standards for power amplifier gain. Since XLR is a professional standard, let's go to venerable rane.com for verification: https://www.rane.com/note135.html

1557289974671.png


4 dBu = 1.2 volts. 20 dBu = 7.7 volts (RMS).

So as you see, 4 volts RMS that I test at is not at all extreme.

Bottom line: my reviews don't cater to spec verification. They follow standardized norms and apples vs apples comparison. We can't grade on a curve especially when we are talking such an expensive processor.
 
Top Bottom