Oscilloscope noise measurements on Behringer UMC 202 HD sound card

[torquil]

Hi!

I am using a Behringer UMC 202 HD USB sound card to produce a 500 Hz tone into some 10 Ohm headphones. The problem is that the sound is too noisy. I'm pretty sure that the sound card must a hardware problem, and I'm wondering if anyone here has an idea of where to look first inside the unit, and what is likely wrong. I also have access to a different USB sound card, a Terratec Producer Phase 26. In order to analyse the noise spectrum, I generate a 500 Hz tone using Audacity, and analyse the signal on the headphone outputs of the sound cards using my oscilloscope. The 10 Ohm headphones is connected to ensure that the headphone output has the correct load. For both cases, I obtain waveforms and FFTs using my Siglent SDS1104X-E. The Behringer outputs a high frequency noise far above the audible range. Please see the screencaptures below.

Note that since the purpose is to produce a 500 Hz tone at a given intensity, I have matched the signals coming from the two different sound cards by adjusting the level to reach 0 dBm for the 500 Hz peak in the FFT. I am pretty sure that I have not made "gain staging errors" (inadvertently overdriving intermediate gain stages in the signal path) when setting up the signal paths in these to cases.

The problem is that the output sound from the headphones is too noisy, and I'm afraid that this massive amount of noise production at super-audible frequencies is loading down the performance of the output stages in the sound card. Anyone seen stuff like this before and have some advice on how to continue? Open the unit up and start scoping around the output stages? As seen in the bottom images, the line out from the Behringer looks much better than the headphones out. I guess that is a good thing and indicates that perhaps something bad is happening around the output stages for the headphone output? I have not yet compared the right and left headphone output, but if the noise is only present on one side, that would narrow it down.

Behringer FFT:

Behringer waveform:

Terratec FFT:

Terratec waveform:


On the line out of the Behringer the signal looks better:

[themadhippy]

10 ohms load seems a bit low,the standard seems to be settling around 32 ohm these days,but higher isn't unusual.

[tszaboo]

You typically cannot debug Audio gear with an oscilloscope. Well, not everything. The scope has 8 bits of vertical resolution, and the audio gear has 16-24 bit, with some 18-is bit noise free. That's a job for a dynamic signal analyzer or dedicated audio test gear.

[torquil]

10 ohms load seems a bit low,the standard seems to be settling around 32 ohm these days,but higher isn't unusual.

I know, but I don't think that is the problem here. The noise problem on the Behringer sound interface occurs also with higher impedance loads.

You typically cannot debug Audio gear with an oscilloscope. Well, not everything. The scope has 8 bits of vertical resolution, and the audio gear has 16-24 bit, with some 18-is bit noise free. That's a job for a dynamic signal analyzer or dedicated audio test gear.

But surely it would be possible to do some useful oscilloscope probing in this case, since the noise outputted by the sound interface is far above the noise floor of the scope, and I'm seeing dramatic differences when comparing different USB sound interfaces. I have an old Rhoede & Schwartz audio analyser here. I could give it a whirl, but I doubt it will add much to what I already have measured. I'll try it, but I'm not even sure it is in working condition.

I'll report back after I have opened it up and probed around a bit looking for the source of the problem.

[torquil]

Just for completeness since the use of the oscilloscope was questioned above, here are the noise floors for both sound cards measured on a Rohde & Schwarz UPL on the maximum frequency span of 120 kHz. So this is without producing the 500 Hz tone. Qualitatively, the difference between the noise floors of the two audio interfaces is the same as when using the oscilloscope FFT.





There might be some differences caused by me forgetting to attach the transducer in this case, but it is easy to see that the tendency for high frequency noise from the Behringer interface is the same.

[shakalnokturn]

Just a thought:
SMPS noise? How are the groundings and interlink of various sound "cards" and computer itself?

Edit:
And you have confirmed the 500Hz actually sounds distorted with your ears without the scope hooked-up?

[2N3055]

You typically cannot debug Audio gear with an oscilloscope. Well, not everything. The scope has 8 bits of vertical resolution, and the audio gear has 16-24 bit, with some 18-is bit noise free. That's a job for a dynamic signal analyzer or dedicated audio test gear.

As you said, you can do some things. For instance you can measure noise floor without signal. You don't need super wide dynamic range for that, just some amplification and dynamic range that is enough to for noise variation to fit vertically. OP did just that.
Also you could measure THD by first suppressing fundamental tone with notch filter and then use scope FFT to measure the rest...

You can do a lot of stuff. But problem is that by using software+audio card combination you can do much better,  and if you are doing lot of audio and want to do it right and fast investing in audio analyser is a good idea...

[GregDunn]

Just a thought:
SMPS noise? How are the groundings and interlink of various sound "cards" and computer itself?

Edit:
And you have confirmed the 500Hz actually sounds distorted with your ears without the scope hooked-up?

^^^ Important questions!  If you can't hear the noise, then it's apparent that the designer didn't consider the ultrasonics to be worth cleaning up.  Behringer is notable for producing, um, cost-effective designs, and they may have simply omitted some filtering that was deemed unnecessary.  Does the other sound card sound and/or measure better than the Behringer?

[drknob]

Many questions:
1) what is the sample rate?
2) what is the word length (i.e. bit depth)
3) what is the internal signal level of the 500Hz tone in dBFS?
4) what are the performance specifications of the two sound cards?

It's fine to look at the supersonic performance of the cards, but they weren't designed for that. Generally, noise above 20kHz is not even considered in computer sound cards and you can't hear it anyway unless it inter-modulates, which is unlikely for whiteish noise. A more realistic possibility is that out of band noise overloads subsequent circuitry and thereby adds distortion products.

But mostly, I agree with GregDunn: Behringer is not a "goto" brand for Pro Audio. And I've never heard of Terratec either.

Also: computer switching power supplies are notoriously noisy from an audio perspective. The best way to eliminate its effect on audio artifacts is to use an external sound card (or 'interface') for isolation. There are many entry level 2-channel devices available from Presonus, Focusrite, Tascam, MOTU and on and on.

You may be expecting too much performance from a base level device.....

[ultrarunner2018]

I realize that this thread is more than 1-1/2 years old now, but I was doing a Google search for the same problem and here I am.
I have the same scope, and have also observed high frequency noise on the headphone output.
i don't have my scope photos anymore, but can easily re-take them.
I used a pair of Audio Technica ATH-M50X with 38 ohm impedance to get my measurements.
I didn't get very technical about what I was looking at - it just seemed strange that there would be all that hf noise on the output. I never checked the main outputs, but will do so over the next couple of days.

After a Google search, it appears that Behringer knew that the noise was there, but decided against doing anything about it - probably to keep the price down.
I paid only $60 for my UMC202HD in 2017.

I am looking into using the main outputs into a separate headphone amp to replace the UMC's headphone output. Or, I might just ditch the Behringer and go for a new DAC/Amp for my new Senn HD600's.

It may also be interesting to compare the noise output of the UMC202HD with the ATH-M50X and the HD600's. The 600's are 300 ohms, so maybe higher impedance will change the hf noise?

[David Hess]

Are we sure the ground for the headphone output is actually ground?  Headphones are isolated loads so there is no requirement that the headphone ground be "ground".

[DL2XY]

Seen this before...
The UMCx0x HD interfaces do not have any reconstruction filters at the outputs.
In audio applications this is not a real issue since the inputs are responsible for proper bandwidth limitation.

Do simply put in a RC lowpass at about 30-40 khz and everything is fine.

 

[sordid_resistor]

I have a Beh UA202, use it a lot, it's very good. However the biggest problem for me was noise, the answer is simple in my case, it was caused by the computer. I got one of those USB opto isolators so there is no direct connection between the UA202 and computer and the noise disappeard. I made a separate 5V supply.  However not all of those USB isolators are made equal, noise can sometime be present but I now know where to look.

This may not be your issue and you may have even eliminated this.