Emotiva XMC-2 Review (AV Processor)

[msmucr]

The main difference is that I was only driving two channels and they are driving all 8. Someone commented that the bug in bass management may be the problem here and this would sort of point to that direction.

Amir,

thanks for the review.

Just what came to my mind about tilted noise spectrum you've measured.
To me this looks like rather typical issue with accumulated roundoff artifacts in recursive (IIR) filters.
Say for example if someone implements common direct form 2nd order HPF with single precision floats all the way through (input, coefficients, intermediate values for calculation, output product) in a straightforward way, then its FFT plot will be similar to what you've got (sans other distortion from DAC and analog circuitry of course).
In those cases the issue comes down recursive filter nature, where feedback imply use of previous values for calculation. With coefficients for low corner frequency and long decaying time domain response towards low frequencies of signal you are quickly in a situation, where algorithm has to deal with comparatively very small and very large float numbers. So despite of huge inherent range of floating point, there is relatively significant accumulated error, which shows on the plot as this tilted spectrum rising towards lows.
See my example attached pic with single prec HPF to 80 Hz and how it can look like on FFT.

There are general ways to push those artifacts well bellow any sane dynamic range of audio signals.. Most of high quality filters uses 64 bit double precision floats for critical parts of its algorithm, when used platform allows that. Also some filters are possible to redesign with different structure, which improves level of artifacts from calculations there.
Fixed point filter implementation and calculation is of course different in nature without "sliding" range of floats, but also in such case one has to deal with its fixed dynamic range for recursive calculations in terms of possible overflows and truncation, se even there it is advantageous to have more bits for numerical representation and calculation of critical parts.. say like in case of old Motorola/Freescale 56k with its 48bit double precision fixed point registers and associated 56 bit accumulators, those were commonly used for pro audio for high quality effects and crossovers. So ways are definitely there, if platforms allow and a designer wants to do that

I haven't seen this AV preamp, but isn't possible, there was some enabled bass management for measured channels...? Like some AV receivers has default 80 Hz HPF for front speakers, which are set to "small" at its setup. It should be apparent from its frequency response measurement.
If this was enabled, it could explain difference to the other measurement from Emotiva forum without that LF junk.

Michal

 

[Spocko]

Considering the margin of error between Amir's and Lonnie's two tests how can anyone make a conclusive statement?

Is this where everyone departs from the scientific process and sides w/ their most favorite test/reviewer? I hope people are capable of separating any emotional attachment from the subject/topic in attempts to be objective.

You must also consider quality control and manufacturing defects - it is not surprising if @amirm received a unit that was defective from the start but this is not Amir's fault as he tests whatever members send him. If Emotiva wants accuracy, then they simply need to send it directly to Amir and he will work with them - this is his policy when manufacturers send products for review. For example when Trinnov heard that I forwarded my Alt16 to Amir for testing, they immediately responded and sent Amir their latest Alt16 as it was just upgraded after I bought mine; this gave Trinnov a chance to review Amir's results beforehand and troubleshoot any unexpected measurements.

 

[Loron]

The Emotiva Lounge thread is going as usual along two general lines:

(1) It sounds great so who cares about measurements; and,
(2) Amir doesn't know what he's doing, SINAD is a bad metric (maybe we should call it THD+N, that seems more widely accepted, same measurement), etc.

The Lounge has always bled Emo blue with a "shoot the messenger" mentality if any flaws are highlighted.

As best I can tell after a quick skim, the RMC-1 and XMC-2 were tested the same, though FW revs are different, but there may be something flat-out broken in the unit Amir tested. I keep wanting to trade in my XMC-1, but almost two years later the XMC-2 still seems to have a lot of bugs.

Until then I'll just suffer along with my exorbitantly-priced third-party-engineered crappy old SDP-75 and accept the abuse from ASR naysayers (fora denizens are the same everywhere). At least it works.

I do wish Emotiva would return to the days of providing a very nice set of test results with their products. They used to be one of the good guys in that, but it seems to have fallen by the wayside, even for the amplifiers.

For now, I'd like to find the thread on the FTC amp specs update, wonder how that's going? I submitted my comments, as did a number of the rest of us techie types, curious.

I obviously am not reading it the same way. A lot are asking the right questions and a lot respect ASR work. It is obvious, except for a couple of individuals, that generally most care about measurements. The thread is more about why the XMC-2 isn’t measuring as close at it should to the RMC-1.which is what Emotiva advertised.
The denial part, the ASR credibility part and the listening test part are old and the thread has moved on.

 

[RichB]

One more comment regarding what was said in the Emotival thread about firmware update. I nearly bricked an Arcam processor for attempting to update it. I got lucky that I got it back up after it hung for many hours. So while I try to update firmware often, I always worry especially if the device is this complicated and this expensive. If a $200 streamer gets bricked, I can buy another one for the owner. But that option doesn't exist for $3,000 processor and aggravation of sending it back and for to the company to fix.

Also, firmware updates are very important when testing functionality. That is not what I do. I am sending digital bits to a box and expect analog out. If getting this right required later firmware, something fundamentally is wrong with the whole product.

While it is not the goal of ASR, observations concerning the firmware update process are valuable to the consumer.
The Emotiva updates have been good, taking < 10 minutes and with few bricks, but they do occur on many of these newer processors.

An Apple updated bricked my ATV4k and replaced it free of charge.

- Rich

 

[RichB]

Amir,

thanks for the review.

Just what came to my mind about tilted noise spectrum you've measured.
To me this looks like rather typical issue with accumulated roundoff artifacts in recursive (IIR) filters.
Say for example if someone implements common direct form 2nd order HPF with single precision floats all the way through (input, coefficients, intermediate values for calculation, output product) in a straightforward way, then its FFT plot will be similar to what you've got (sans other distortion from DAC and analog circuitry of course).
In those cases the issue comes down recursive filter nature, where feedback imply use of previous values for calculation. With coefficients for low corner frequency and long decaying time domain response towards low frequencies of signal you are quickly in a situation, where algorithm has to deal with comparatively very small and very large float numbers. So despite of huge inherent range of floating point, there is relatively significant accumulated error, which shows on the plot as this tilted spectrum rising towards lows.
See my example attached pic with single prec HPF to 80 Hz and how it can look like on FFT.

There are general ways to push those artifacts well bellow any sane dynamic range of audio signals.. Most of high quality filters uses 64 bit double precision floats for critical parts of its algorithm, when used platform allows that. Also some filters are possible to redesign with different structure, which improves level of artifacts from calculations there.
Fixed point filter implementation and calculation is of course different in nature without "sliding" range of floats, but also in such case one has to deal with its fixed dynamic range for recursive calculations in terms of possible overflows and truncation, se even there it is advantageous to have more bits for numerical representation and calculation of critical parts.. say like in case of old Motorola/Freescale 56k with its 48bit double precision fixed point registers and associated 56 bit accumulators, those were commonly used for pro audio for high quality effects and crossovers. So ways are definitely there, if platforms allow and a designer wants to do that

I haven't seen this AV preamp, but isn't possible, there was some enabled bass management for measured channels...? Like some AV receivers has default 80 Hz HPF for front speakers, which are set to "small" at its setup. It should be apparent from its frequency response measurement.
If this was enabled, it could explain difference to the other measurement from Emotiva forum without that LF junk.

Michal

Is it possible that after the factory reset the front channels were set to large so bass management was applied? It seems odd to the layperson that this would affect the results of a 1kHz test signal.

- Rich

 

[Lsc]

I obviously am not reading it the same way. A lot are asking the right questions and a lot respect ASR work. It is obvious, except for a couple of individuals, that generally most care about measurements. The thread is more about why the XMC-2 isn’t measuring as close at it should to the RMC-1.which is what Emotiva advertised.
The denial part, the ASR credibility part and the listening test part are old and the thread has moved on.

Yes I’m reading it the same way. It’s obvious that some folks who could have had a bad experience may have an agenda.

All I’m trying to get to…which was communicated by Lonnie in the Emo forums is that the XMC2 is equal to the RMC1 for stereo listening. This was what Emotiva had been saying all along but the tests may be showing otherwise. Hoping that they run tests with the RMC1 to confirm their statement.

I guess if I had an RMC1 and XMC2 and compared them side by side listening to lossless songs on Apple Music, would I be able to tell the difference?

I honestly don’t know how things can sound “significantly” better by replacing the XMC2 than what I listened to last night but as you all know, we humans and our hearing, it’s tricky. But having said all that, I’m 100% satisfied with my XMC2 both in performance (music and movies) and usability (it’s quite simple to use).

Would I recommend to a potential buyer? Well if you get a copy of the XMC2 that resembles mine then by all means I would recommend, but if you get a version like the one that Amir got, then definitely not.

 

[DonH56]

Is it possible that after the factory reset the front channels were set to large so bass management was applied. It seems odd to the layperson that this would affect the results of a 1kHz test signal.

- Rich

The bass management bug was my first thought for the LF rise as well. It adds (sums) all the bass from all the channels and that means all the noise gets added as well. It should be decorrelated among channels but that would still result in a rising noise response like Amir saw.

The phase shift is odd, but the overall SINAD is comparable to the RMC-1 IIRC.

 

[DonH56]

I obviously am not reading it the same way. A lot are asking the right questions and a lot respect ASR work. It is obvious, except for a couple of individuals, that generally most care about measurements. The thread is more about why the XMC-2 isn’t measuring as close at it should to the RMC-1.which is what Emotiva advertised.
The denial part, the ASR credibility part and the listening test part are old and the thread has moved on.

I was looking at the earlier posts and the other thread wherein a certain user basically blasted ASR and its members as a "joke" or something like that. I do not spend a lot of time on the Emo Lounge. It can be very helpful, e.g. when I was trying to get my XMC-1 working after the last video board upgrade, but IME/IMO the overall tone is against measurements, even from their technical representative in the Lounge (KL). But yah there are a lot of good guys there as well. I try to be one, but we are all human. I am usually careful to note in my comments that much of the problems lie in the inaudible range.

At this point it would be interesting to have Emotiva test the unit Amir tested and compare notes. They have done that in the past, and Amir is quite willing to work with manufacturers as he has done in the past with Emotiva and others.

 

[peng]

I am the person who started the thread and asked "Is this guy a quack".
In retrospect, I should not have written that... I took a shortcut using "quack" to just mean "is this guy [Amir} credible". I ran it past a few friends and they all said I should not have worded it this way and I would like to apologize to Amir and anyone else who might see this as an accusation that Amir is a quack... that was not my intention at all. I just wanted to know if this was a credible reviewer.

That would have been a better way to ask for sure (the quack vs credible). Aside from that, I don't think it is logical (okay may be to some extent is is) to start asking "is this guy credible" just because he found the XMC-2 measured worse than the XMC-1. Your points about "RMC-1 in his "green" zone, but the XMC-2 in his "red" zone." and "I don't get this as they are topologically the same... " are of course valid, but to me that would justify questioning more the potential root causes than immediately, and only questioning whether the person who did the measurements was credible.

First of all, keep in mind Amir's "green" vs "red zone" is based on the DUT's SINAD performance only. The XMC-1 is actually in the "orange" zone and SINAD is 98 dB, vs the XMC-2's 84 dB. I don't know for sure but my guess is that 98 dB vs 84 dB SINAD, all else being equal, would not result in audible difference in blind listening tests and I am sure many owners of the 2 will find their units sounding great for them.

Regarding the "topologically the same...." part, even if it is, there could still be many reasons why there one measured 14 dB worse than the other. It could be that the XMC-2 has more noise picked up somewhere due to cable routing, grounding scheme etc.., or certain parts in the signal chain being substituted from the 1 to the 2 We will not if the reasons are related to some differences in the design and/or components/parts related, without seeing the complete BOMs, that's assuming the circuitry are in fact the same, we don't know that either.

As an example, take a look of the measurements of some "topologically the same.." AVPs/AVRs tested such as the Marantz AV8805, 7705, SR7015, and the Denon AVRs, you will see their SINADs also show differences of 14 dB or even more. When Amir's bench test results first revealed such differences, especially when the cheaper Denon units measured better than even the $5,000 Marantz AVP, some readers probably would wonder the same thing, that is whether Amir did something wrong, yet tests other tests including a few Audioholics.com's as well as a couple Denon/Marantz own measurements showed very similar results. Then some of us started to wonder if Marantz's HDAM buffer stage would be the cause for the different SINADs. On the difference in FR, we don't have to wonder, because we know the difference was caused by the different choice of the DAC reconstruction filter.

Back to the "credible..." part, based on the many tests I have seen in the past 2 to 3 years, I would say yes. It doesn't mean the guy would never make any mistakes, but to me, the guys is credible enough that if I have to wonder why two similar products measured vastly different in a standard SINAD measurement, I would question the technical reasons first, not the operator of the test bench.

 

[peng]

I can't find performance specifications for Storm processors. Their amps have the absolutely minimum specs and not 20Hz to 20kHz.
High priced processor companies seem to deem performance specs unnecessary. If Emotiva and others had published measurements, then I expect ASR would not be finding these anomalies.


- Rich

Agreed, it would be nice if manufacturers would always publish measurements of their flag ship and near flag ship products. ASR should still do theirs so we have checks and balance. People often think more expensive = better sounding but tests after tests proved that's not always truth. That being said, many owners may be happier if there are no measurements published at all, then they can simply rely on subjective measurements that would work for them because of the expectation bias.

 

[peng]

Could also be part quality, tolerance, PS design, or even the layout needed to fit in a smaller chassis. I hope it’s a faulty unit, but it’s really on Emotiva to prove they didn’t muck up the XMC2s now that these measurements are out.

As best I can tell, my XMC2 sounds really good too, but it’s kind of a slap in the face that we may have been sold a much more inferior product than we were lead to believe we were spending $3k on.

I see the major difference between the test results of the two is 98 dB vs 84 dB SINAD and the 2 has more jitter. Under "normal" listening conditions, how many people can tell the difference between 98 dB and 84 dB SINAD or even 70 dB SINAD especially if their FFT looks very similar except in magnitudes? I bet "0". So I am not surprised your XMC2 sounds really good regardless of the difference in SINAD..

 

[David_M]

@amirm ... isn't there a million ways to get the 4V rms output test signals you fed into the Emotiva?

Your input AP source could have been set at 4V rms with the Emotiva volume set at -20dB, for example, to get a 4V rms output...or
Your input AP source could have been set at 1V rms with the Emotiva volume set at -10dB, for example, to get a 4V rms output... etc.

The point I'm trying to make is that with different combinations of AP input source level and Emo's output volume control to get the 4V rms test output signal, we can get different SINAD numbers. If the AP source level was high and the volume control set to say -10dB, we could get higher SINAD than if the AP source level was low and volume Emo's knob set high, to amplify all the noise and distortion from the preceding circuitry.

BTW, I don't and have never owned Emotiva products, so I'm not a biased observer for or against their products.

 

[Spocko]

Yes I’m reading it the same way. It’s obvious that some folks who could have had a bad experience may have an agenda.

All I’m trying to get to…which was communicated by Lonnie in the Emo forums is that the XMC2 is equal to the RMC1 for stereo listening. This was what Emotiva had been saying all along but the tests may be showing otherwise. Hoping that they run tests with the RMC1 to confirm their statement.

I guess if I had an RMC1 and XMC2 and compared them side by side listening to lossless songs on Apple Music, would I be able to tell the difference?

I honestly don’t know how things can sound “significantly” better by replacing the XMC2 than what I listened to last night but as you all know, we humans and our hearing, it’s tricky. But having said all that, I’m 100% satisfied with my XMC2 both in performance (music and movies) and usability (it’s quite simple to use).

Would I recommend to a potential buyer? Well if you get a copy of the XMC2 that resembles mine then by all means I would recommend, but if you get a version like the one that Amir got, then definitely not.

Probably not. Heck, replace the XMC2 with a Denon 3600H as a 2 channel DAC/preamp and I'm wondering if people could tell the difference.

 

[AdamG]

FYI I am actively deleting off topic posts. If one of your posts got slashed, consider it an unofficial warning to stay on topic in a Review thread. We permit great amounts of conversational drift in other threads. But Official Review threads need to remain on topic so accessing relevant review information and data is not overly complicated by pages of drift and off topic posts.

Please and thank you for your understanding and support.

 

[Billy Budapest]

The measurements of the RMC-1 with firmware 1.7 and with firmware 1.9 were drastically different, for the better, as revealed by Amir’s follow up measurements. I wonder if the XMC-2’s performance can likewise be drastically improved by a firmware update?

 

[RichB]

Probably not. Heck, replace the XMC2 with a Denon 3600H as a 2 channel DAC/preamp and I'm wondering if people could tell the difference.

Since I own and both an RMC-1 and Denon 3700, I say yes.

- Rich

 

[peng]

Since I own and both an RMC-1 and Denon 3700, I would yes.

- Rich

To your trained ears, but how about untrained ones? I remember reading one of Harman's studies, that "trained vs untrained" was a significant factor, though that was a study on comparing speakers.

 

[RichB]

To you trained ears, but how about untrained ones? I remember reading one of Harman's studies, that "trained vs untrained" was a significant factor, though that was a study on comparing speakers.

The very first test I do, is to compare Stereo flat (no processing) to Direct and Pure Direct.
I have yet to hear a processor that is indistiquishable. The Yamaha A820 was absolutely awful when comparing the HDMI in to the analog in (Pure) from the Oppo player.

- Rich

 

[Lsc]

To you trained ears, but how about untrained ones? I remember reading one of Harman's studies, that "trained vs untrained" was a significant factor, though that was a study on comparing speakers.

Probably not. Heck, replace the XMC2 with a Denon 3600H as a 2 channel DAC/preamp and I'm wondering if people could tell the difference.

I could tell a difference between the XMC2 and the BAT VK33SE. Also between the XMC2 and my old Theta Casanova.

 

[sealman]

Well after a few minutes of digging I see the the RMC-1 was reviewed with 1.9 firmware and the XMC-2 with 2.3. The aforementioned bass management bug appeared with version 2.2 and continues unfixed with 2.3.

This bug is NOT SUBTLE at all and persists whether using Dirac or not. It was immediately noticeable to me after upgrading my firmware to 2.2.
Whether or not this is in any way connected to the difference between the RMC-1 and XMC-2's very different measurements I have no idea but it is a possibility I would think.