• 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
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
I am grateful for the review by Amir and all the illuminating discussion, including links to reviews of apparently similar performing Marantz processors. I've read every single page here and many of the linked articles. I've loved this ASR website for quite awhile now but have been going through life and email changes so this might be my first post (can't remember if I joined or posted under a previous email). Amir's discussion of EQ in other pages is also extremely important and I mostly agree with everything I've read so far. I've been using manual "room correction" for a decade myself and wouldn't yet have it any other way (though multieq with the editor app looked like a possibility).

Of the issues Amir identified, by far the most shocking to me is the steep rise of distortion above 5kHz, reaching 1% just above 10khz.

I wish Amir had also tried the rated output level of 2.4V. I strongly suspect that would have "fixed" (or covered up) the shocking problem. One need only look at the Secrets review of the 8802(a), reportedly identical in audio output circuitry and specs, where they did measure 10kHz at 2 volts, obtaining a more acceptible 0.002% THD. THEN, they also reported 0.0035% at 5 volts, but only for 24/96 which is moving those pesky aliases up and out of the way. Very curiously, Secrets did not measure 10kHz at 5 volts with 44.1kHz sampling, which leaves just enough room for the shocking issue Amir reports to have been missed (or simply not reported).

If the source of the shocking rise in THD is merely the slower than slow reconstruction filter, why are there no visible aliasing products until you get over 2 volts, and then they take off like mad? It seems to me that aliasing products should increase more like linearly with input level.

I have noticed distortion products increasing like mad in another very different AV product I measured, an Integra Research RDC7. Everything looks fine (well, THD about 0.005% IIRC) until you get to 1V unbalanced and 2V balanced. Then, distortion products take off, and for that reason I was not able to do a direct measurement with my [email protected] card and RMAA at 4V balanced, where distortion exceeded 1%. At first I thought the RDC7 was broken, then I got another one, and after it looked the same I decided to try lower levels. I was pleased to see the distortion essentially disappear at lower levels. But I do not like preamps without distortion free headroom. Preamplifiers should be able to reach an amplifier's peak power, not just rated power, without adding any distortion. Amplifiers typically have 3dB or more headroom above rated power. This is why a 4V requirement is not at all a bad idea.

But meanwhile I wonder if some sort of soft clipping has become the standard for surround processors. I wonder if such a thing might even be required by proprietary standard, I know not which, there are many, such as THX, CSA, ETL, Dolby, DTS, Audyssey, and they may be contingent.

If so, it might be that the huge rise in distortion above 5kHz is caused not just by the slow reconstruction filter, but that in combination with a required (?) soft limiting curve. Just an idea that occurred to me tonight.

[updated to list variety of possible proprietary standards which might require soft clipping and/or certain output level limits]
 
Last edited:

RichB

Senior Member
Joined
May 24, 2019
Messages
447
Likes
454
Location
Massachusetts
With its signal switching capabilities all things are possible in the AV8805. Testing of these various signal flows is even included in the service manual. Unfortunately based on the AV8805 user manual, all 2ch (not 7.1 external inputs) playback goes through the attenuation/ADC/DAC chain.

ADC/DAC Signal Chain

Specifically this chain is -8.5 dB attenuation of the analog signal using an active opamp circuit with NJR (NJM2115) opamps and +/- 2.5 rail voltages(!) and less than great specifications. This gain reduction is to fit the signal within the 1V capabilities of the just OK AKM (AK5358B) ADC. After the ADC the signal goes to the NJR CMOS switches and CMOS volume control. This is no way to treat a nice analog signal.

2ch Playback

In the AV8805 user manual Page 240.

o "Select the method for ... 2-channel Playback direct and stereo playback modes - No distinction between digital and analog is given

(in other locations in the manual, direct and pure direct are given as the same in this area)

o Speaker size can be set

o SW Mode seems indicate digital processing

o Distance can be set which is hard to do in the analog domain

Page 50 of the manual is does not provide a clear answer, it just shows levels with digital or ADC signals inputs. Note the -8.5dB mentioned above.

The fact that this signal flow is not made clear in the user or service manuals also suggests that the ADC system is used. Higher-end users won't be happy with this design.

XLR Inputs

XLR inputs first go to a three opamp circuit that reduces the signal level by 6dB and converts it to single-ended. The resistors used in these circuits are 5% tolerance with means that common mode noise reduction from the differential amplifier function won't be great. The resistors are also thick film surface mount units that have extra noise (over Johnson noise) and distortion. As an overall note: Newer units such as the AV7705 use 1% thin film resistors in key locations, such as the differential amplifiers after the DAC's, which don't have these drawbacks.

There is no indication that XLR inputs take any different path in the AV8805 than stereo inputs.

Why would Denon/Marantz treat analog signals in the manner? My guess, YMMV, is that D/M is worried about what some users will connect to the analog inputs. Tube preamps often have high RCA output levels as do other components. The switches and volume control in these units are CMOS based. CMOS is great except it doesn't like high voltages. The voltage supply to these CMOS units is 7V DC, which is likely about all they want so see anywhere. A 6V AC sine wave input would have 8.4V peaks for example, which might causes damage. The signal input limit of the volume control is just over 4V, so care is necessary.

The 8 (7.1) external inputs are an exception, this is likely because the assumption is that the 8-channel RCA outputs feeding these inputs are from say a Blu-Ray player than will have about a 2V maximum output on each channel. In addition, 8 ADC circuits would be required instead of just 2 and that would cost money. In the user manual the 7.1 channel input is labeled for use by a Blu-ray player, DVD player, etc, not stereo.
Welcome @bigguyca and thanks for this excellent post.
It's great that Marantz publishes the schematics and, hopefully, this continues.

Why does the AV8805 has low output voltages (2.4V XLR), is this cost savings on a $4500 processor?

The two processors I am looking into are the Emotiva RMC-1L (still buggy) and the Monoprice HTP-1 (available this fall).
Currently, there are no performance specifications for either.

Since the analog path in many HT processors seems to be hobbled or non-existent, I have moved to an Benchmark LA4 for 2-channel. The XMC-1 analog volume control that compares well with the LA4 but the LA4 adds some clarity using level-matched (instantly switching). For two channel, the signal path is BRIX HTPC Roon -> Oppo UDP-205 DAC -> LA4 -> AHB2's -> Salon2's.

- Rich
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
Welcome @bigguyca and thanks for this excellent post.
It's great that Marantz publishes the schematics and, hopefully, this continues.

Why does the AV8805 has low output voltages (2.4V XLR), is this cost savings on a $4500 processor?

- Rich
It's clear from other reviews the AV8805 puts out 5V and possibly more, and it's fairly low distortion (0.002%) until higher frequencies about 1/8 of the sampling rate where aliases enter the picture because of the slow reconstruction filter.

Since I observed an even more draconian 2V XLR limitation (distortion rising rapidly above 2V) with Integra Research RDC7, also a $4500 processor but from 2001, I'm thinking home theater processors tend to keep their output voltages low so as not to clip power amps. With many different amplifier channels, it's more likely some of them might be driven well past clipping, and you're watching the movie (most users) and not focussing so much on the music, you might let it pass. And also I suspect there may be deliberate "soft clipping" in the line outputs which may cause distortion to rise starting at half power. It's hand holding for HT types. And it appears to be nicely covered up for marketing purposes with a "rated output" specification around 2 volts.

(Edit: Also the processors have lots of EQ and other functions which could cause momentary clipping. So the soft clipping and/or voltage limitation helps to keep power amps from blowing up.)
 
Last edited:

RichB

Senior Member
Joined
May 24, 2019
Messages
447
Likes
454
Location
Massachusetts
It's clear from other reviews the AV8805 puts out 5V and possibly more, and it's fairly low distortion (0.002%) until higher frequencies about 1/8 of the sampling rate where aliases enter the picture because of the slow reconstruction filter.
Can you provide links to these measurements from other sites?

- Rich
 

RichB

Senior Member
Joined
May 24, 2019
Messages
447
Likes
454
Location
Massachusetts
Here are the specs for the Oppo UDP-205:

Frequency Response: 20Hz – 160kHz (-3dB ~ +0.05dB)
Signal-to-Noise Ratio: > 120dBr
THD+N: < 0.00018%
Output Level: (RCA) 2.1±0.2Vrms. (XLR) 4.2±0.4Vrms
Dynamic Range: > 120dB
Crosstalk: > 118dB
It's hard to understand why products costing 4 time as much, or more, cannot supply a handful of simple performance measurements.
It seems the high-end processors are above objective performance measurements. ;)

- Rich
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
Can you provide links to these measurements from other sites?

- Rich
Sure, it was listed only about 18 pages ago :).

https://hometheaterhifi.com/reviews/receiver-processor/processors/marantz-av8802-processor-review/

Very curiously, they show 10kHz measured at 2 volts only for 44.1kHz sampling rate. At 96kHz sampling rate, they show both 2V and 5V. They didn't show 10kHz at 5 volt for the 44.1kHz sampling rate, which might have revealed the kind of hugely rising distortion above 5kHz that Amir discovered at 4V in the 8805.

Perhaps there are some secrets even too secret for Secrets to reveal. For those secrets, we must come to ASR.

(The above review is for 8802, but previous discussion has determined audio circuitry identical with 8805. Their other measurements look comparable to Amir's measurements allowing for predictable differences caused by output level.)
 
Last edited:

RichB

Senior Member
Joined
May 24, 2019
Messages
447
Likes
454
Location
Massachusetts
Sure, it was listed only about 18 pages ago :).

https://hometheaterhifi.com/reviews/receiver-processor/processors/marantz-av8802-processor-review/

Very curiously, they show 10kHz measured at 2 volts only for 44.1kHz sampling rate. At 96kHz sampling rate, they show both 2V and 5V. They didn't show 10kHz at 5 volt for the 44.1kHz sampling rate, which might have revealed the kind of hugely rising distortion above 5kHz that Amir discovered at 4V in the 8805.

Perhaps there are some secrets even too secret for Secrets to reveal. For those secrets, we must come to ASR.

(The above review is for 8802, but previous discussion has determined audio circuitry identical with 8805. Their other measurements look comparable to Amir's measurements allowing for predictable differences caused by output level.)
I think each product deserves its own measurements.
Even so, the 8805 (and 8802) XLR output voltage at 2.4 Volts.

Perhaps, Marantz knows there are issues that cause 4+ Volt operation to produce increased noise and distortion and chose a number to produce a better specification.

It's not a specification that induces confidence. At this price point, there are products that produce 11+ Volts on the XLR outs.

- Rich
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
I see replying to email didn't post, so I am posting.

You don't find it at all curious that Secrets left out the 5v measurement of 10kHz at 44.1kHz, when they went to that trouble for 96khz?

There's more evidence also...notice how the distortion products shown by Secrets at 44.1kHz go on and on. That's suggestive of a digital filtering issue in the 8802 as well...slow reconstruction filter, the same basic issue (there may be others) that leads to rising distortion at 5kHz at higher outputs on the 8805.

I think we're looking at design choices here, design choices that have continued essentially unchanged from the 8802, when they did make some significant changes, though probably even the earlier models than that were same "philosophy" (slow reconstruction filter) too and probably weren't that different.

Not some kind of "mistake" they made specific to 8805, or the excuses people keep making ("oh, it's so complicated.")

It's the "bubbly" champaign sound. We can predict pretty much the "sound" of a slow digital filter. It's going to sound somewhat softer, counter intuitively, for the most part, despite (or because) the added noise. Digital edges will be free of pre-ringing, so there also may be more "pop" at times. So, softer but poppier. That's basically what DSD does also, which is another way of sacrificing noise filtering for time domain "accuracy." Marantz has been fairly big on DSD and the like, so it's not surprising this slow filter philosophy is shown here.

All of the Secrets measurements for the 8802 are consistent with Amir's measurements of the 8805, with allowance for the different levels used. By all accounts so far, the parts and circuitry is identical. Only part quality or programming could have changed, or some kind of interference which you'd hope they'd be smart enough to catch.

Now, I have an even more daring conspiracy theory...that the low SINAD results from something else...another design choice.

And that is how the pre-pro ultimately clips. I suspect there is a goal, perhaps mandated for compliance with some proprietary standard, that soft clipping be used, and that the maximum voltage not exceed 6V.

That limitation, and the limitation for soft clipping, creates high distortion 5-6V, and lesser and lesser at lower levels. But still the 0.002% we see at 4V at 1kHz, which is considered "acceptible" even if far removed from limitations of the dac.

OK, this may not fully explain the sinad, that may be simply the poor choice of attenuator following the dac, etc, but it might explain why there does seem to be some ceiling (which nobody has yet revealed) to the output level, and this may complicate other matters. I've never been near an 8805 so I can't tell you what the maximum output level is, and how it clips, but my prediction is softly at 6V.
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
I see replying to email didn't post, so I am posting.

Y
Now, I have an even more daring conspiracy theory...that the low SINAD results from something else...another design choice.

And that is how the pre-pro ultimately clips. I suspect there is a goal, perhaps mandated for compliance with some proprietary standard, that soft clipping be used, and that the maximum voltage not exceed 6V.
And yet another theory, that no professional tells us the clipping limit, and how it clips, because they have all signed nondisclosure agreement with unspoken company.

So it would be up to us amateurs to find out. I'll accept AV8805 donation...
 
Last edited:

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,168
Likes
1,643
Location
Sin City, NV
I have a couple of the earlier 8801A's (purchased at closeout for $1200/ea) and I've always wondered where that 'value' would have been at MSRP when new - not sure it's even there at the discount really. However, the convenience factor trumps everything else in my two applications: multiple HDMI sources, with active monitors in surround configurations. I guess the comfort I can take from these measurements is that, there's no reason to even consider an upgrade to the newer model. Although I expected better than the cheaper version somewhere... I can't say I'm surprised there isn't - at least all the copper looks pretty when viewed through the vents. :cool:
 

RichB

Senior Member
Joined
May 24, 2019
Messages
447
Likes
454
Location
Massachusetts
The measurements here are for the DAC/DA performance, so it represents the best it will be.
Are there measurements with the DSP's engaged?
I suspect they will be in normal use, even if that is not desired ;)

- Rich
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,168
Likes
1,643
Location
Sin City, NV
The measurements here are for the DAC/DA performance, so it represents the best it will be.
Are there measurements with the DSP's engaged?
I suspect they will be in normal use, even if that is not desired ;)

- Rich
As I didn't see a statement at the beginning of "All tests were conducted in Pure-Direct mode" - I think it's a reasonable assumption that those are with DSP engaged... since I think it's only defeatable that way (if it's even actually doing what it says then - which is rather doubtful IMO).
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
8,152
Likes
10,108
And yet another theory, that no professional tells us the clipping limit, and how it clips, because they have all signed nondisclosure agreement with unspoken company.

So it would be up to us amateurs to find out. I'll accept AV8805 donation...
Here is an paragraph about distortion at various levels in the AVR 7701 (Marantz lists the same identical specs for all their pre-pros going back more than 10 years and measurements back that up.) So I can't with 100% certainty say the 8805 is like this, but it likely is.


With 0 dbFS input and volume set to 0 db the output is a touch over 3 volts where these measurements were taken. You can turn the volume up to +18 db. With max level digital input the distortion begins to rise, but not very much. It is still .005% at 4.5 volts. Distortion has risen to 1 or 2% by +12 db and 6 volts out. It puts out 16 volts at max volume, and above +12 db the increase in voltage is less than it should be. Also it appears to be an analog amplification. If you feed it some lower level digital signals and turn the volume up distortions stays very low until you pass 4.5 volts output even if volume controls are +18 db. So useful for unusually low sound levels. Will not be harmful if all you are doing is turning it up to hear at normal levels.

From measurement I did on the 7701 here:
https://www.audiosciencereview.com/forum/index.php?threads/marantz-avr-7701-dac-measurements.3485/
 

Sal1950

Major Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
6,922
Likes
5,091
Location
Central Fl
As I didn't see a statement at the beginning of "All tests were conducted in Pure-Direct mode" - I think it's a reasonable assumption that those are with DSP engaged... since I think it's only defeatable that way (if it's even actually doing what it says then - which is rather doubtful IMO).
If the DSP you refer to is Audyssey I'm pretty sure it was not in the loop. And yes you can disable Audyssey while still using/measuring the DAC.
Amir would have to answer exactly how it was configured while testing.
I have a couple of the earlier 8801A's (purchased at closeout for $1200/ea) and I've always wondered where that 'value' would have been at MSRP when new - not sure it's even there at the discount really.
A lot of the "value" we pay for is access to the latest-current surround software like Dolby HT and DTS-M. Once it's supplanted by something newer the value of the old tech is nill.
Now aren't you just dieing for IMAX Enhanced by DTS. o_O
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
As I didn't see a statement at the beginning of "All tests were conducted in Pure-Direct mode" - I think it's a reasonable assumption that those are with DSP engaged... since I think it's only defeatable that way (if it's even actually doing what it says then - which is rather doubtful IMO).
Previously presented evidence shows matters getting no better and possibly worse in "pure direct" whatever. Thus ensued much debate. It's possible that under some conditions it's possible to completely avoid DSP, but "pure direct" may not necessarily engage them. It was either Amir or perhaps someone else who did this test on another 880x unit.

I have another theory. Actually, when it goes through the AD converter, it's cleaner, when SPDIF is the source. That would explain why in previously measurements matters got worse. Such measurements did NOT have DSP functions enabled, so pure direct should have been pure direct. The alternative analog bypass is noisier than the chip in my theory. However, Amir or someone else failed to do the definitive test of bypassing digitization--using the multichannel inputs as Kal suggested.

Even with a complete round trip through digital and back, S/N better than 115dB could easily have been achieved, absent other goals. Counterintuitively, AD converters are often quieter and better than DA converters.
 
Joined
Jul 9, 2019
Messages
11
Likes
1
Location
San Antonio, TX, USA
Here is an paragraph about distortion at various levels in the AVR 7701 (Marantz lists the same identical specs for all their pre-pros going back more than 10 years and measurements back that up.) So I can't with 100% certainty say the 8805 is like this, but it likely is.


With 0 dbFS input and volume set to 0 db the output is a touch over 3 volts where these measurements were taken. You can turn the volume up to +18 db. With max level digital input the distortion begins to rise, but not very much. It is still .005% at 4.5 volts. Distortion has risen to 1 or 2% by +12 db and 6 volts out. It puts out 16 volts at max volume, and above +12 db the increase in voltage is less than it should be. Also it appears to be an analog amplification. If you feed it some lower level digital signals and turn the volume up distortions stays very low until you pass 4.5 volts output even if volume controls are +18 db. So useful for unusually low sound levels. Will not be harmful if all you are doing is turning it up to hear at normal levels.

From measurement I did on the 7701 here:
https://www.audiosciencereview.com/forum/index.php?threads/marantz-avr-7701-dac-measurements.3485/
Thanks! That sure sounds like soft clipping to me. If not, once you get to 1% distortion it really takes off. If they weren't aiming for slow clipping, better than 0.005% at 4.5V might have been possible. And the takeoff of distortion above 5kHz might not have been so bad. So this is all consistent with my theory of design choices, not simply crappy implementation.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
25,104
Likes
52,150
Location
Seattle Area
There's more evidence also...notice how the distortion products shown by Secrets at 44.1kHz go on and on. That's suggestive of a digital filtering issue in the 8802 as well...slow reconstruction filter, the same basic issue (there may be others) that leads to rising distortion at 5kHz at higher outputs on the 8805.
FYI someone contacted Marantz and gave them a link to my measurements and asked about the filter roll off. They contacted Japan engineers and answer came back that they on purpose have selected a slow roll off filter because they like the sound of it! So our measurements are accurate in that regard. They should provide filter selections if they are going to do this and select something that the theory says is incorrect.
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,168
Likes
1,643
Location
Sin City, NV
Unlikely considering there's a correlating Denon offering for almost any Marantz (and more of course)... unless there was a significant cost savings - at the component level - I would guess any differences are merely 'hobbling' of features that are still present. In many cases they're probably more 'hard-set' at the factory than on the Marantz version. Though this would merely be conjecture on my part based on owning several models from both brands over the years. I can't see adding internal amps equaling less noise (and a lower MSRP)... but anything is possible.
 
Top Bottom