Meanwhile I reworked my test setup to further improve THD+N performance. Loopback investigations revealed that the residual noise I observed was dominated by crosstalk from the active output @ -1dB full scale poisoning the input signal. I had a feeling that there must be some room to improve dynamic range. All in all this looked like some internal ground loop issues of the EMU-Tracker.
After some fiddling I decided that improving the sound card accordingly is beyond me. Facing an I/O crosstalk problem next approach was to isolate in- and output channels. Luckily I hace a second EMU-sound card in my stash, and so I used the EMU-TRACKER for output exclusively and the EMU-0202 solely for input. The audio path is routed balanced through a 3-pole low pass filter to reduce generator noise and an attenuator (passive). Both USB devices are driven from a single laptop PC running REW under WINE/Linux LMDE3 (meanwhile deprecated). The in/out channels were assigned in the pulseaudio sound setup. The laptop was mains powered all the time.
These are my actual results with REW:
After some fiddling I decided that improving the sound card accordingly is beyond me. Facing an I/O crosstalk problem next approach was to isolate in- and output channels. Luckily I hace a second EMU-sound card in my stash, and so I used the EMU-TRACKER for output exclusively and the EMU-0202 solely for input. The audio path is routed balanced through a 3-pole low pass filter to reduce generator noise and an attenuator (passive). Both USB devices are driven from a single laptop PC running REW under WINE/Linux LMDE3 (meanwhile deprecated). The in/out channels were assigned in the pulseaudio sound setup. The laptop was mains powered all the time.
These are my actual results with REW:
Attachments
Cool enough... 🙂
Well, i can take mine Behringer and bring it away from mains wiring (and walls) and have a good loopback measure, but as soon i will test an real amplifier, the giant 50Hz spike + friends occour; this is done with 4 different amplifier's, so it cannot be the amplifier's i guess.
I'am considering buying another soundcard, but i donno if i really need it... let's see.
Jesper.
The Behringer has a dedicated button to disable "monitoring" function. I guess the EMU-Tracker has it always on by design?
BTW, I wasn't able to take full advantage of REW... I blame it to my Win7-32bit machine. Do I understand correctly that 64-bit is necessary for that?
There is a 32bit installer as well:
https://www.roomeqwizard.com/installers/REW_windows-x32_5_20_4.exe
Alleviating 50Hz spike typically means battery powering the soundcard and PC that it connects to, so that only one item of the test setup (the DUT/amp) connects to the mains AC power. The next aspect is to ensure that the DUT/amp internally has its input gnd connected directly to its output gnd - if not then there may be other ways to alleviate that aspect.
Is this with the internal Java Audio? If so I'd try a download of the latest Behringer driver from their site and for Win 7. Then choose ASIO from R.E.W.
You could envision testing another ASIO driver like ASIO4ALL, but my experience with that on Win 10 (not 7) is that there seems to be some interference with both several ASIO drivers and other existing sound devices. I'd say make sure you use only 1 ASIO at a time (as in 'have only 1 ASIO installed') - disable the other drivers and sound devices, use a single ASIO one and ensure you reboot after installation. Installation should be launched with the unit disconnected from USB.
Doing it like this, I managed to get a rather stable operation while previously it was unbelievably glitchy.
I can't guarantee it will work because I haven't tested on Win 7, but normally it should.
I have the same problem with other soundcards and ASIO drivers and only with REW. With ARTA everything works fine.
I opened up the Behringer UMC202HD this morning and had another look at it, and while doing a search online, found this:
Comparative tests (translated)
THD is definitely better on a Focusrite Scarlett than on the Behringer. He also gets the 24kHz spike.
Additionally, the Behringer seems less good on dynamic range than the others.
Comparative tests (translated)
THD is definitely better on a Focusrite Scarlett than on the Behringer. He also gets the 24kHz spike.
Additionally, the Behringer seems less good on dynamic range than the others.
The problem with UMC202HD is that it stacks unnecessary gain stages introducing distortion and reducing DNR. This thread's project was about bypassing these stages. That leaves us with a very good audio codec and a nice usb to I2S converter at a considerably lower cost. I've been thinking about the next step on this conversion. I believe I'll have something to post in the weekend...
MagicBus, you will really like a scope for deep investigations like you are doing.
Where I am today (no mods):
And a quick test (no cal.) shows the high freq. spike on the right:
Input gain is approximately at 9 o'clock, output is at 1 o'clock or so., no preamp, no direct monitoring.
My noise floor seems worse than what others have, not sure, what do you think? What about THD here?
Where I am today (no mods):
And a quick test (no cal.) shows the high freq. spike on the right:
Input gain is approximately at 9 o'clock, output is at 1 o'clock or so., no preamp, no direct monitoring.
My noise floor seems worse than what others have, not sure, what do you think? What about THD here?
Moving the board on top of my desk PSU and turning the latter on, getting max peaks when the XLR region of the PCB is near the PSU:
The board makes for a pretty good mains interference detector 😀
The board makes for a pretty good mains interference detector 😀
A scope being an essential tool in many cases
As I mentioned before, for best results input gain is always at MINIMUM
this is of little help when tracing audio noise in the uV-range
No scope was needed to show that interference, just the PCB is out of the box and then I oriented it over my desk PSU.
I know I can get less distortion and noise with reducing the input gain and also reducing output gain, but then the R.E.W. cal. procedure is more difficult.
Now I see another problem with R.E.W. and ASIO. I get an error today, something like 'Device Reset Request' and 'soundcard didn't provide any input data from ASIO'. ...
I know I can get less distortion and noise with reducing the input gain and also reducing output gain, but then the R.E.W. cal. procedure is more difficult.
Now I see another problem with R.E.W. and ASIO. I get an error today, something like 'Device Reset Request' and 'soundcard didn't provide any input data from ASIO'. ...
OK, here's what I did to recover from the ASIO error above with R.E.W.:
- Disconnect and re-connect the soundcard (USB cable computer port side)
- In R.E.W., change from ASIO to Java.
- Change back from Java to ASIO
- Set the ASIO rate, Control Panel Buffer, etc...
I don't see any problem with your soundcard. As it comes from the factory, the best it can do is what you post, I think. The raised noise floor is because the enormous input gain. The same explains the sensitivity to environment conditions. Could you repeat with all gain controls -input/output- completely turned down?
Regarding this high frequency peak, it seems it's not consistent in all specimens. In mine, it was shown much lower within the audio spectrum. Having it already above 20kHz I suggest you should ignore it since this soundcard is actually meant for viewing audio only.
Regarding this high frequency peak, it seems it's not consistent in all specimens. In mine, it was shown much lower within the audio spectrum. Having it already above 20kHz I suggest you should ignore it since this soundcard is actually meant for viewing audio only.