Quick measurements of Behringer UMC22

[pkane]

I've had this little interface for many years and it's been a champ. It currently sells for about $56 and I use it for room response measurements and other tasks involving a microphone. The unit is USB-powered and provides 48v phantom power to the mic, as well as instrument input, headphone out and two line outputs.

Measured using the trusty and well-performing Element24 interface. UMC22 only supports 48kHz at 16 bits, so that's what I measured.

First, the DAC portion fed with 1kHz signal. I couldn't get 0dBFS out of it. -3.2dBFS produced 400mV at the output. THD+N of -83.5dB and an elevated noise floor below 100Hz. This didn't seem to be mains frequency related, so I'm not sure what caused it:

Multi-tone test (remember, this is still with 16-bit data). Some non-linearity in the peak level observed, these should normally all line up at the same level:

Frequency response looks pretty flat in the audible range, but phase isn't quite so nice:

I then repeated the same set of tests with the ADC portion of the interface.

1kHz with input of 265mV at the microphone input, anything greater resulted in clipping:

Multi-tone test:

Frequency response and phase for the ADC:

And here's what a 1kHz square wave looks like through the ADC:

 

[amirm]

Nice job. Promoted to home page.

 

[Jimster480]

Nice, I checked out this one when I got my UMC202HD and decided that the newer USB interface was worth paying for... Also there were no measurements of this unit before. But based on Amir's test today; this unit looks to perform somewhat close to the UMC204 and probably also the 202 then too.

 

[Tks]

Anyone know what I am supposed to see with the ADC Square Wave result?

 

[JRG]

Great job!
What software do you use to do the measurements?

 

[pkane]

Great job!
What software do you use to do the measurements?

Thanks! REW is the software. Free and completely awesome.

 

[pkane]

Anyone know what I am supposed to see with the ADC Square Wave result?

Pretty much what I posted. Some devices distort the square wave beyond the Gibbs phenomena, such as making the wave asymmetric or having larger than expected rise/fall at the edges, etc. The one I posted looks like a normal, band-limited square wave should.

 

[ofrappier]

Hi,

The 1 KHZ was generated by REW ? at 16 bits at 24 bits ? or with other hardware/software ?

I use 1 Khz file .wav at 32 bits depth, and not 16 bits .wav to have a clean 1000 Hz

 

[pkane]

Hi,

The 1 KHZ was generated by REW ? at 16 bits at 24 bits ? or with other hardware/software ?

I use 1 Khz file .wav at 32 bits depth, and not 16 bits .wav to have a clean 1000 Hz

24 bits generated by REW, 24 bits used by Element24 to convert to analog, but UMC22 is only a 16 bit device.

 

[Tatteredmidnight]

Great job! Nice to see some community posts make it to the front page. Keep up the good work.

 

[BYRTT]

.....but phase isn't quite so nice:
View attachment 38676
.....

Will guess the phase mismatch looking is either because your two interfaces is not on same physical clock, or you missed a manual allingment of impulse response for session

 

[SIY]

There's something wrong with the multitone measurements. Either a window choice or something interrupting the test signal during acquisition.

 

[pkane]

There's something wrong with the multitone measurements. Either a window choice or something interrupting the test signal during acquisition.

REW switches the window to Rectangular automatically when doing multi-tone. I can try forcing it to use something else

 

[SIY]

Aha! Yes, you don't want to use a rectangular window for that. No idea why the software would make you do this- it's certainly not the case for ARTA, Virtins, or APx software I generally use.

 

[SIY]

?? I always use the rectangular window for multi-tone.

That time-domain discontinuity is what causes the skirts.

 

[scott wurcer]

That time-domain discontinuity is what causes the skirts.

Sorry I removed the comment I don't have a time domain discontinuity the way I do multi-tones and I don't use any of the usual packages these days.

 

[SIY]

Now you're making me look like I'm drunk-posting.

 

[JohnPM]

Aha! Yes, you don't want to use a rectangular window for that. No idea why the software would make you do this- it's certainly not the case for ARTA, Virtins, or APx software I generally use.

Because the generated tones are all periodic within the FFT length.

 

[SIY]

Because the generated tones are all periodic within the FFT length.

The FFT definitely shows discontinuity.

 

[JohnPM]

Clock looks to be a long way out. REW can measure the clock error if a timing reference is used and the associated Analysis options are selected.