Budget DAC Review: behringer UMC204HD

[amirm]

I'm not understanding why the digits are still so far from full-scale (see waveform) when distortion sets in.

Could you (@amirm) perform a similar test with your 204 and see what you get?

I need more data first. You are playing that clip you created and then recording it with 204hd and seeing just -66 db or so level? Is this an analog loop through or are you playing it through speakers? And have you maxed out the levels using front-panel controls?

 

[RayDunzl]

Commenting on the last part first, I too saw that high peak at 60 Hz which caused me to not want to use it for measurements.

It isn't there when the preamp is "off"... Nothing else changed.

UMC202HD noise floor:

Not perfect but a heck of a lot cleaner than my 25 foot unbalanced analog cable to the PC digitizer.

 

[DonH56]

Ground loop causing 60 Hz? Or just poor ground... Or it's from the preamp/source?

Rising spurs below 0 dBFS could be either the DAC itself or any of the circuits after that (like the image filter and/or output buffers inside the box), to include noise causing clipping someplace in the chain. Datasheets routinely spec everything at -1 dBFS.

 

[RayDunzl]

Ground loop causing 60 Hz? Or just poor ground... Or it's from the preamp/source?

The UMC202 has USB power (from across the room) and long mic cables (coiled up) to preamp. Since it doesn't show 60Hz with the preamp off** I don't blame it.

I'm assuming it is from:

The preamp - sitting on top of the isolation transformer. It "looks" good there.
Cables - rats nest of AWAW* wiring configuration

The 60Hz (and harmonics) are inaudible (maybe unmeasurable in the air) even at maximum preamp output, so, I'll rate it inconsequential for now. The ceiling fan hum is audible when I listen for it. And there is low-speed underclocked PC fan noise. And the refrigerator. and I-75 about 3 miles away. And helicopter rotor wash. And HeatPump compressors... And golf carts out back on the 4th tee. Lots of innocuous low level stuff floating around. Right now I'm being audibly subjected to General Hospital.

The only audible noise in the system is a slight hiss from the JBL LSR 308 if I put my ear up to it. Even it doesn't show up on measurements at the listening position, or maybe it barely does, I forget.


*AnyWireAnyWhere

** The preamp has no power switch, but has two output zones, seems able to "power off" one or either or both zones, but is still powered on, overall, eternally, contributing billions of tons of CO2 annually

 

[RayDunzl]

I'm not understanding why the digits are still so far from full-scale (see waveform) when distortion sets in.

Ok, Operator Error:

I thought the USB out would reflect the channel input level. And, of course, it does.

But there is also an "Output Level" knob which affects the digital USB output level, which didn't enter my thought stream, thinking it was more for control of analog out (headphones and line out).

So, the Output Level was a bit low, not permitting full-scale digital output on USB, before the onset of clipping occurred at the analog input and/or ADC function.

 

[RayDunzl]

The only audible noise in the system is a slight hiss from the JBL LSR 308

Here's a measurement...

https://audiosciencereview.com/forum/index.php?threads/jbl-lsr-308-in-the-house.1066/#post-27832

 

[Blumlein 88]

Here's a measurement...

https://audiosciencereview.com/forum/index.php?threads/jbl-lsr-308-in-the-house.1066/#post-27832

Looks like the hiss is only from the tweeter. And what about 6 db on top of what is otherwise room noise? And this right next to the speaker.

 

[RayDunzl]

Looks like the hiss is only from the tweeter. And what about 6 db on top of what is otherwise room noise? And this right next to the speaker.

I figure the lower "hiss" is the woofer... cross at 1.7k

One amp chip, two outputs, digitally crossed, making the woofer and tweeter hiss within their ranges.

I only notice it when listening for it right at the speaker. I'm deaf so don't go by that (which is why I measured it a little).

 

[ships]

Hi all. I've been on the hunt for a decent budget DAC to pair with a Magni 3 amp to drive HD6xx. I've been using the very affordable Signstek mini USB DAC but I'll be giving that to someone else soon so I'll be needing a replacement. Originally I figured I'd pick up a Modi 2 and call it a day, but amirm's measurements and impressions have me a bit weary on buying any more Schiit. People have noted that the output from this Behringer DAC is low. My question, is that from both the 1/4 inch main out, and the RCA playback outputs? If it's low, anyone know if the Magni 3 has enough power to overcome the low output to sufficiently drive an HD6xx? Thanks, this forum has been a wealth of information!

 

[amirm]

Welcome aboard Ships. Unfortunately I don't remember the balanced levels. I will measure and report back.

But the real good news is that I also have the Schiit Magni 3 that I have yet to review. I can plug it in quickly and let you know how loud it can get.

 

[ships]

Hi Amir, that would be fantastic. Thanks again.

I must admit that I'm pretty new to all things audio/engineering. I'm in the biological sciences, but my physics is pretty rusty.

Another question, is there any detriment to sound quality in running an amp like the Magni 3 at close to max?

I did a quick test to see what volume I'd have to set through Windows going through my Signstek to max out the Magni at a comfortable listening volume. It ended up being a level of only 4/100, although of course this doesn't leave any headroom for quieter recordings. In comparing the sound quality between having my Window's volume maxed at 100/100 and the Magni on hi gain at ~10 oclock vs 4/100 at 5oclock (maxed), I can't say I really heard too much of a difference other than some hiss when there is no sound playing (imperceptible with music playing). From the technical side, is there a big difference in signal between these two cases?

 

[amirm]

OK, I ran a test with Magni 3 headphone amp. Using my quiet tracks, I had to turn up the volume to 3:00 o'clock leaving 10% or so of headroom. This was across three different headphones including my inefficient Sennheiser HD650. While I wish there was more headroom, I say you are safe to buy the Behringer to pair it up with it.

 

[amirm]

Another question, is there any detriment to sound quality in running an amp like the Magni 3 at close to max?

I listened to the Magni 3 with the headphones I have at max volume with music on pause and it was quiet. So assuming you don't have more efficient headphones, then there is no issue there. Otherwise, the background noise may be an issue.

I am answering this in the context of having a low-output DAC driving it. Obviously if you pump more signal into it, at high volume it could get distorted.

 

[amirm]

I did a quick test to see what volume I'd have to set through Windows going through my Signstek to max out the Magni at a comfortable listening volume. It ended up being a level of only 4/100, although of course this doesn't leave any headroom for quieter recordings. In comparing the sound quality between having my Window's volume maxed at 100/100 and the Magni on hi gain at ~10 oclock vs 4/100 at 5oclock (maxed), I can't say I really heard too much of a difference other than some hiss when there is no sound playing (imperceptible with music playing). From the technical side, is there a big difference in signal between these two cases?

Oh, I didn't read this part before posting the answer above . This tells me you do have more efficient headphone than what I have so with respect to Behringer, you should be fine.

Answering your question, you should set Windows volume to 100 and then use the analog knob on the Magni to control the volume. That way you don't raise the noise floor too much as you noticed.

 

[ships]

Answering your question, you should set Windows volume to 100 and then use the analog knob on the Magni to control the volume. That way you don't raise the noise floor too much as you noticed.

Gotcha. I'll update if I end up picking one up.

 

[thm_jpl]

Regular readers of this forum know that we have been searching for an excellent performing DAC for under $100. So far we have not found it. All DACs tested underperformed my reference, the iFi iDAC2 ($350).

Next up is a product from professional world (i.e. designed for music recording/playback), the behringer UMC204HD. I paid a meager $79.99 for it, shipping included: https://www.amazon.com/Behringer-UMC204HD-BEHRINGER-U-PHORIA/dp/B00QHURLCW

View attachment 6964

As you can see, this unit does everything but wash dishes. You have headphone amp, stereo input capture, multiple outputs, microphone phantom power, MIDI, etc., etc.

The unit itself seems quite stout and much more stable than the other budget DACs. It sits where you put it despite cables tugging on it. As with other DACs, it is self-powered through USB.

As you all know, my weapon of choice in testing its performance is to through a 24-bit, 48 KHz J-Test signal at it. Ideal output would be a single spike at 12 Khz. Here is how it did as compared to iFi iDAC3 at over 4X its cost:

View attachment 6965

Color me impressed! Yes, there is a bit of mains related noise at 60 Hz and harmonics but its level is more than 120 db below reference so of no audible consequence at all. There is also a spike at similar level at 10 Khz. Other than that, the response is very clean.

In the above graph it looks like the noise floor is lower than iFi iDAC2 but that is not entirely the case since the output of the behringer is lower. If we were to amplify its output to the same level as iFi, the two would roughly equalize.

Now let's try another test: a 7 Khz, 16-bit/44.1 Khz simple tone:

View attachment 6966

Can anyone comment if this is a valid comparison I'm making? Using Arta, generating a 12kHz sine and 7kHz sine at 0dB, output looped back to input. Input gain knobs turned to minimum, main output gain knob adjusted until 0dB FS is seen (about 75% of max). Output is unbalanced RCA, inputs are balanced XLR/TRS (but are unbalanced for this measurement). Sampling rate 96kHz (can do 192kHz).

Here is the 12kHz spectrum:

Noise floor is about -135dB, maybe slightly better than UMC.​

This is 7kHz spectrum:

Worse than UMC204HD due to harmonics.​

Product is Scarlett Solo v2, which is more expensive ($99) and has less features. No MIDI, multiple outputs, and multi-use inputs, etc.
Codec: CS4272 and USB XMOS xhra-2hpa.

 

[DonH56]

Not sure what you mean by "valid"; it is a valid measurement. What may be confusing is that the harmonics of a 12 kHz signal are at 24 kHz and up so when you cut off the spectrum at 22 kHz'ish you do not see the harmonic distortion of the 12 kHz tone. Since the sampling rate is 96 kS/s, you should be able to extend to Nyquist (48 kHz) and see the harmonics unless something else is limiting you to ~22 kHz.

HTH - Don

 

[thm_jpl]

Not sure what you mean by "valid"; it is a valid measurement. What may be confusing is that the harmonics of a 12 kHz signal are at 24 kHz and up so when you cut off the spectrum at 22 kHz'ish you do not see the harmonic distortion of the 12 kHz tone. Since the sampling rate is 96 kS/s, you should be able to extend to Nyquist (48 kHz) and see the harmonics unless something else is limiting you to ~22 kHz.

HTH - Don

Thanks, was trying to reproduce the original measurements in the thread which stopped at 22kHz.
If I extend it, the 24kHz harmonic is at -80dB, second third harmonic -82dB.

 

[DonH56]

A little higher than the 7 kHz tone as expected.

Note that normal terminology is for the fundamental (12 kHz) to be considered the first harmonic, crazy as it might seem, so the second harmonic is at 24 kHz, third harmonic at 36 kHz, and so forth.

 

[amirm]

Noise floor is about -135dB, maybe slightly better than UMC.

Hi there. Welcome to the forum! And thanks for the additional measurements!

As to your question, for the purposes of measuring noise floor, your comparison is may NOT be valid. The fourier transform sizes have to be matched for that comparison. Mine uses 32,767 points. Match that and it should be close. Looking at how fine the "grass" is in the noise floor of your measurement, I suspect it is using a lot more taps in the transform causing the noise floor to artificially be lower than my measurements.

The distortion products are not affected by this.

One other variable is the noise of my measurement system versus the input ADC on that unit. On that front, I could not get as clean of a capture as you have. That is a pretty good ADC there.