AVS Forum banner
1 - 20 of 958 Posts

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #1 · (Edited)
Look what the cat dragged in




Let's have a closer look at this device. I'll update this post as we go along.

Pros
- Price (cheap compared to processors with similar feature set)
- Parametric EQs
- Sturdy casing
- Small form factor (1.5 RU height)
- Well structured on-screen menu
- Rack mountable (rack ears sold separately)

Cons
- Price (expensive compared to AVRs with similar feature set)
- No parametric EQ for back surrounds
- Only 3 parametric EQs for the subwoofer channel vs. 11 for all other channels (except back surrounds)
- Distance settings allow only for 0.3ft (10cm) steps
- No upload of EQ settings, EQ has to be set with remote (takes about 10 minutes per speaker)
- Only rudimentary manual/documentation available
- After startup volume ramps up to last used setting which can create problems with universal remote controls
- No changelog for firmware updates
- On-screen volume display hard to read on light backgrounds
- Doesn't support sample rate of 176.4kHz
- Doesn't support PLII or Neo:6 at sample rates higher than 96kHz

Bugs (Firmware version 1.52.02.47)
- Bass management filters create a 180° phase shift when a 12dB filter slope is selected (see post #686)
- LFE crossover and slope is mislabeled as "Subwoofer"
- Odd manual test tones for level calibration (see post #3 and post #50)
- Manual test tone generator is situated downstream from EQ (should be upstream so test signal passes through EQ)
- Enhanced bass option defaults to "on" although front speakers default to "small"
- Headphone output is downstream from bass management, delay, gain and EQ -> completely distorted sound in headphones
- "Direct" mode with x.0 digital input signal is downstream from bass management and delay with redirected bass muted -> all channels are high pass filtered, low frequencies are lost
- "Direct" mode with x.1 digital input signal is downstream from bass management and delay -> all processing active except EQ
- "All Stereo" with 2 channel digital or analog input signal is downstream from bass management, delay and EQ -> processed L/R is sent to surrounds
- Multichannel LPCM 7.1 output mode is mislabeled as "All Stereo"
- Reset EQ values of a speaker aren't retained after restart

Other observations
- Rubber feets smell
- HDMI switching is about 4 seconds

Tips
How to configure listening modes? (Post 130)
How does "Direct mode" work? (Post 99)
How to set speaker distance settings using REW measurements with loopback (Post 593)
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #3 ·
Measured the internal test tones with an RTA (1/48 octave, 8192 FFT length, 32 averages). Here's what they look like:




Not sure how these can be used to calibrate the device to reference level. I'd expect the signals to be either full band pink noise for use with an RTA or bandlimited for use with a SPL meter.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #4 ·
Tested bass management, delays and EQs. Signal routing looks right. Hallelujah!


The LFE crossover settings are mislabeled though. Setup > Speaker Setup > Size/Crossover > Subwoofer and Setup > Speaker Setup > Size/Crossover > Subwoofer Slope set the low pass filter and slope of the LFE channel, not the summed subwoofer channel.


I've read somewhere the device would smell and it does. It's the rubber feets. I'll remove them if the smell doesn't go away within the next few days.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #5 ·
Another obervation in Setup > Speaker Setup > Speaker Distance:

"Left Front" sets the delay for the right speaker. No other option changes the delay for the left speaker.


If tested left and right on the AN 1 input. Speaker were set to "Large".


Edit: When listening to 2 channel PCM over HDMI, delays for left and right do work.


Edit 2: When listening to correlated pink noise fed to the AN 1 inputs, the delays do work as expected. The problem is obviously with my measuring setup.


Edit 3: Nothing was wrong with my measuring setup, I just didn't understand what I was seeing
Check post #9 .
 

·
Registered
Joined
·
371 Posts
It appears UMC-200 is a half-baked product with unresolved bugs. Thanks for detailed observations. 4 seconds for HDMI switching is really lame. I was considering it despite only 4 HDMI inputs.


Does anyone make a simple pre-pro that is:


- relatively small (certainly not 8 inches tall)

- devoid of analog inputs

- devoid of video scaling

- equipped with 7 HDMI inputs with decent switching


I saw pictures of UMC-500 but should probably give up on it given how hard it has been for Emotiva to deliver properly functioning UMC-200. Too bad.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #7 ·
^

HDMI switching time of 4s unfortunately has to be considered normal, albeit unacceptable. A Onkyo 3008, 818 or Denon 2312 don't feel any faster. The problem is HDMI.


Emotiva's CEO said that the UMC-500 will be released some time after the XMC-1 which has been under development for years. So there's no point in waiting for an UMC-500.


Let's stick to the UMC-200 for now and hope they can fix the bugs. I do believe the UMC-200 can be the best affordable processor available. With the on-board PEQs the UMC-200 gives the user so much more control than any other consumer processor/AVR. I do believe one can get superior or at least similar results to any multichannel room correction system out there today. That's a bold statement. We'll see whether this is true. REW will support the UMC-200 in a future release.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #8 ·
Tried to copy a Audyssey MultEQ calibration including a -20dB DynamicEQ curve with the help of REW. The result sounds remarkably similar to the "original".


Here's the resulting curves, 1/6 smoothing (Sub + L = green, Sub + R = red):

 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #9 ·
There's a bug in how the UMC-200 sets delays (speaker distance setting). I've used the AN 1 input and measured the the in-room response of the right speaker.


1) Impulse response for distance R=0, distance L=0:




2) Impulse response for distance R=0, distance L=0.9ft:




3) Impulse response for distance R=0.9ft, distance L=0.9ft:





If distance for L is increased, R is delayed. Although changing the distance setting for L should vary only its own delay, shifting all other speakers is OK, as long as delays are applied to all other channels accordingly. This is not the case in the UMC-200 (I've checked by changing delays for C and surrounds).

Furthermore setting delays for L and R to the same value brings them back to the same overall delay as setting them both to 0.
 

·
Registered
Joined
·
1,861 Posts
Dang it Markus, now you're making me paranoid!



I can do some group delay measurements of just the UMC-200 at line level without involving a microphone, like I did for the bass management. I have a Tascam US-144 MKII, which allows having an analog loopback timing reference while providing a signal to the device under test via analog or S/PDIF. Or I can use HDMI inputs, but that's a little more problematic because of the REW glitches with HDMI and ASIO4ALL.


I can't get to it right away, but I can do it sometime this coming week. I have to remove it from my rack and do the measurements on my test bench. I propose to measure each of the left- and right-channel group delays under the following conditions:


Left distance = 0, Right distance = 0

Left distance = 0, Right distance = 32.8 ft

Left distance = 32.8 ft, Right distance = 0

Left distance = 32.8 ft, Right distance = 32.8 ft


So I'll provide 8 graphs: two for each of the above conditions (right- and left-channel delays). The expected data are delay ~= 0 for distance = 32.8 ft and delay ~= 29.8 msec for distance = 0 in all cases.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #11 ·
^

Yes, please do those measurements.


Simply use the right channel for loopback and check Preferences > Analysis > Use Loopback as Timing Reference. Now the peak of the in-room impulse response will show exactly how much a signal is delayed by the UMC-200 and speaker distance.

Alternatively you could also directly measure the preamp outputs.


You'll find that setting the delay for the left speaker will change the delay for the right speaker and vice versa. Furthermore the delay for L=0/R=0 will be the same as R=32.8/L=32.8.
 

·
Registered
Joined
·
1,861 Posts
^

Back when I did the measurements of the bass management, the loopback timing reference was a little tricky in some cases. With S/PDIF it was easy though. I could apply a signal to either right or left individually using the Tascam S/PDIF output, and set the loopback timing reference as channel 2 analog out of the Tascam back to its channel 2 analog input. In this way, the timing reference was invariant to changes in distance settings of the DUT. But to apply an LFE signal to the DUT I needed to use HDMI. I first tried to use the channel 2 analog output and input of the Tascam as timing reference, but the Tascam analog out was out of sync with the HDMI out of the laptop, resulting in erroneous delay measurements.


The fix was to make the loopback timing reference go through the DUT itself. In this way, the two channels of laptop HDMI out were in sync with each other, as were the two analog ins of the Tascam. However, there is a potential gotcha with this technique. The DUT channel used as timing reference must have its distance setting held constant for all delay measurements of the channel(s) under test to be consistent. Suppose you're measuring the delays of channels A and B, and using channel X as timing reference. If you need to change the distance settings of both channels A and B, then X must be neither A nor B, because the delay of the timing reference channel X must be invariant for all measurements of A and B.


For this test I can use S/PDIF as the stimulus of the DUT, and use channel 2 analog out of the Tascam to its channel 2 analog in as the timing reference.
 

·
Registered
Joined
·
12,396 Posts

Quote:
Originally Posted by markus767  /t/1491348/official-emotiva-umc-200-thread#post_23776976


There's a bug in how the UMC-200 sets delays (speaker distance setting). I've used the AN 1 input and measured the the in-room response of the right speaker.


1) Impulse response for distance R=0, distance L=0:




2) Impulse response for distance R=0, distance L=0.9ft:




3) Impulse response for distance R=0.9ft, distance L=0.9ft:





If distance for L is increased, R is delayed. Although changing the distance setting for L should vary only its own delay, shifting all other speakers is OK, as long as delays are applied to all other channels accordingly. This is not the case in the UMC-200 (I've checked by changing delays for C and surrounds).
The behavior shown in the three plots is how distance adjustments work in all surround processors. If the L speaker is 1' further away than the R speaker (the L sound arrives 1 ms later), shifting the delay in the other speakers is not only OK, it is the only option available.


If the corrective delay is being added to some channels but not others, that would be a flaw.
Quote:
Furthermore setting delays for L and R to the same value brings them back to the same overall delay as setting them both to 0.
Exactly as it should.
 

·
Registered
Joined
·
12,396 Posts

Quote:
Originally Posted by markus767  /t/1491348/official-emotiva-umc-200-thread#post_23754143


Measured the internal test tones with an RTA (1/48 octave, 8192 FFT length, 32 averages). Here's what they look like:




Not sure how these can be used to calibrate the device to reference level. I'd expect the signals to be either full band pink noise for use with an RTA or bandlimited for use with a SPL meter.
I'm assuming the red curve is the L channel. That is not materially different from the noise used by other surround processors, such as THX noise (500 Hz - 2 kHz) or Dolby noise (center weighted at 500 Hz, 9 dB/oct slopes either side). The main thing is that the noise be bandlimited to avoid errors due to different bass rolloffs and room modes, and inconsistent HF speaker/room responses.


Wideband noise is not ideal for subwoofer calibrations, as it changes level depending on the crossover frequency and the rolloff point in the room. But IMHO sub cals are best left to either much better tools or the ear.


One problem with the UMC-200's internal noise is that it does not pass through the EQ, so it is possible that some gain offset will result, depending on the nature of the EQ. Usually one EQs the system, the tweaks the gain trims. If you use an external test disc, that will pass through the EQ fine, but then you cannot adjust the gains because that is only possible while the internal noise is turned on. So you have to write down the level offsets then map them in as a separate step. Not the most elegant process, and one easily fixed in the firmware if so inclined.


Is there a second firmware update coming that takes care of all the issues left over from the first one?
 

·
Registered
Joined
·
1,861 Posts
It seems to me that in any sane design, the delay of each channel should depend only on the distance setting of that channel, and not on the distance setting of any other channels. That delay should be as follows:


added_delay = (max_distance - current_distance_setting) / c


where c is the speed of sound expressed using the same distance units used in the pre/pro.


For the UMC-200, the max distance is 32.8 feet, so when the actual distance setting is 32.8 feet, you get an added delay of zero. When the actual distance setting is 0, you get a delay of something like 29 msec. This is consistent with the delay measurements of the UMC-200 I've done so far.


As to how some room correction algorithms set this distance, that's a separate issue.
 

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #16 ·

Quote:
Originally Posted by Roger Dressler  /t/1491348/official-emotiva-umc-200-thread#post_23783982


The behavior shown in the three plots is how distance adjustments work in all surround processors. If the L speaker is 1' further away than the R speaker (the L sound arrives 1 ms later), shifting the delay in the other speakers is not only OK, it is the only option available.

In other words, the speakerb that is farthest away is defined as delay = 0 and all other speakers are delayed relative to that reference point.


Lets assume the following speaker distances:

L = 1

R = 2

SL = 3


Then SL will become the reference point, i.e. 0 and the other speakers are delayed by:

L = 3-1 = 2

R = 3-2 = 1
 
  • Like
Reactions: andyc56

·
Registered
Joined
·
12,396 Posts

Quote:
Originally Posted by markus767  /t/1491348/official-emotiva-umc-200-thread#post_23784120


In other words, the speaker that is farthest away is defined as delay = 0 and all other speakers are delayed relative to that reference point.
Exactly! This is taken care of automatically by the processor, in contrast to the "good ole days" where users had to dial in the delays manually. That was more confusing.
 

·
Registered
Joined
·
12,396 Posts

Quote:
Originally Posted by andyc56  /t/1491348/official-emotiva-umc-200-thread#post_23784081


It seems to me that in any sane design, the delay of each channel should depend only on the distance setting of that channel, and not on the distance setting of any other channels. That delay should be as follows:


added_delay = (max_distance - current_distance_setting) / c
Using this method, if all of the speakers in the system were 5' away, the decoder would impose 28 ms of delay in every channel (33-5=28). That would a) eat up memory and b) cause an A/V sync offset. The goal of the processor is to minimize both.


In actual products, the added delay for my example will be zero. No added latency, no memory used.
 
  • Like
Reactions: andyc56

·
Registered
Joined
·
15,012 Posts
Discussion Starter · #19 ·
Just checked delay settings again and they do work exactly as Roger explained. In my former measurements I probably didn't increase other speaker channels enough to see a change. Sorry for the confusion this might have caused.
 

·
Registered
Joined
·
1,861 Posts

Quote:
Originally Posted by Roger Dressler  /t/1491348/official-emotiva-umc-200-thread#post_23785808


Using this method, if all of the speakers in the system were 5' away, the decoder would impose 28 ms of delay in every channel (33-5=28). That would a) eat up memory and b) cause an A/V sync offset. The goal of the processor is to minimize both.


In actual products, the added delay for my example will be zero. No added latency, no memory used.

Ohhhhh, I understand now.
That's a really subtle issue I hadn't even considered.
Quote:
Originally Posted by markus767  /t/1491348/official-emotiva-umc-200-thread#post_23784120


In other words, the speaker that is farthest away is defined as delay = 0 and all other speakers are delayed relative to that reference point.

Great way of putting it. I'm glad this issue came up, as I learned some really useful stuff from it. Thanks to Roger and Markus for this edification.
 
1 - 20 of 958 Posts
Top