USB Audio interface for measurement

Hi All,

I've been inactive in the DIY speaker world (and on this forum!) for a few years but time and opportunity has finally come to finish some speakers that have been half finished in bits in storage for years for lack of time, working space and money... :)

Years ago (since 2003) I used a Desktop Windows XP PC with a Sound Blaster Audigy 2 ZS as my measurement card - one of the few prosumer cards in the early 2000's that supported true 24 bit 192Khz recording and playback without re-sampling and had very good analogue performance, and I used it extensively for speaker / crossover measurement with various software including most recently ARTA.

Unfortunately in the last year that PC has gone belly up, and as the card was an original 32 bit PCI card that most PC's today won't support I can't really put it in anything else, even if I had another desktop PC to put it in...

Besides, it was very inconvenient using a large, noisy desktop PC for speaker measurements!

Nowadays I use a Windows 7 Laptop but it only has a single shared mic/speaker socket, and even if it had both I'm sure the quality would be dreadful compared to my old card anyway and inadequate for precise speaker/crossover measurement.

So I'm in the market for a USB sound card, and my requirements are:

1) Preferably under £100, in the UK.

2) Full duplex, naturally.

3) True 192Khz 24 bit without any re-sampling or other post processing, and the flat frequency response and low distortion/noise floor that you would expect of a good quality 192/24 card.

4) Must work well with ARTA.

5) Inputs that support both XLR with phantom power for the measurement microphone, and line level input for direct connection to crossovers, making impedance measurements etc.

6) Inputs must be fully independent - eg independent selection of XLR or Line, independent gain control etc.

This is important to be able to make proper use of dual channel mode in ARTA - typically you would connect the measurement microphone to Left in XLR mode with phantom power, and connect the right input in line level mode (possibly with a suitable L-Pad) to the input terminals of the speaker to cancel out any response errors in the amplifier, random delay variations in the sound card that might affect phase measurement etc.

I did a quick 10 minute browse tonight and came across one promising candidate at about half the maximum I'm prepared to spend, the Behringer U-PHORIA UMC202HD USB Audio Interface:


Behringer U-PHORIA UMC202HD USB Audio Interface | DV247


It seems to meet all my requirements above, and seems almost too good to be true for the price. Is it ? (Or have things just moved on a long way in the 10 years since I last looked at USB sound cards!)

The only thing it seems to be missing that would be good is a "high impedance" mode for the inputs, which would be useful for more accurate measurement of crossover outputs without loading them down, and also be necessary for impedance measurements.

I'm assuming it will have an input impedance of about 600 ohms or so in line level mode ?

Another interface that does have a high impedance mode for one of its inputs and thus would be better for impedance measurements and direct crossover measurements is the Steinberg UR-22 MkII:

Steinberg UR-22 MkII | DV247

It's also pretty similar in other regards, however it's about twice the price.

Am I going to run into problems with impedance and component measurements without a high impedance mode ?

I don't have access to an RLC bridge anymore so I would quite like to be able to measure crossover coils and caps (mainly coils, as my multimeter can do caps) using ARTA as well using an RL / RC test configuration.

Any thoughts and an opinions of these two boxes, or recommendations of other ones to check out in my price range ?

On a related note - in the house I currently live in I won't be making most of the measurements in the living room (not enough reflection free window time) and whilst my laptop and a USB interface are far more portable than my old system for measurements I still have the issue of the amplifier.

I now only have one amplifier and it is well and truly fixed in place in the main AV cabinet, getting it out to take to another room to do speaker measurements would be a major PITA.

So it occurs to me that a small, portable amplifier that I can use with the USB audio interface purely for the purpose of driving a single speaker to take a frequency response measurement would be extremely handy, if not essential.

It doesn't have to be fancy - it doesn't really need anything more than a volume control, doesn't need to be stereo, doesn't need tone controls, lights etc. What it does need is very flat frequency response, low output impedance and an output of about 10 watts. Something that is not much bigger than the USB audio interface would be great, as together with the laptop and USB audio interface it would make it possible for me to measure just about anywhere including the holy grail - outdoors!

Is anybody using a small portable amplifier for measurement tasks and have a recommendation ?
 
Last edited:

PRR

Member
Joined 2003
Paid Member
...seems almost too good to be true for the price. Is it ? (Or have things just moved on a long way in the 10 years since I last looked at USB sound cards!)

The only thing it seems to be missing that would be good is a "high impedance" mode for the inputs.... I'm assuming it will have an input impedance of about 600 ohms or so in line level mode ?

EVERY musician is recording the next Hit Record, in their bedroom, with gear like this. It is cut-throat pricing. Quantity moved is large enough that they generally use good chipsets.

I have not seen a True 600 Ohm Input in decades (except vintage and clone-vintage gear).

Those bedroom stars will not be happy playing guitar into 600r, 10K, even 100K. Beeringer knows that.

Did you find the Owner Manual? (They don't make it easy.) Input is 3K Mike, 1Meg hi-level.

https://imagescdn.juno.co.uk/manual/583637-01U.pdf --page 21 of PDF
 
Yes they do make the manuals hard to download - I had already found the one you linked to soon after posting but annoyingly it is only a "quick start guide" which says go to the website to get the "full" manual - well there isn't one! Not that I can find anyway. :confused:

Yeah I dunno where I got 600 ohms from. It's been quite a few years since I last tried to measure any speakers - as you point out XLR microphones inputs are usually 3k impedance as that's what the Microphones are designed to drive into especially condensor mics with pre-amps like mine.

However I think you're mistaken about "High level" being 1M - if you read the spec page carefully there are three input modes - Mic, Line, and Instrument, with a switch to switch between Line and Instrument.

I'm assuming the balanced XLR input is always Mic level regardless of the switch - so -4dBu and 3k impedance.

The TRS input is switchable between Instrument and Line. Instrument says a sensitivity of -3dBu and 1M impedance, however Line has a sensitivity of +20dBu and NO input impedance stated, at least not in the quick start guide... :(

So the instrument mode probably switches in an additional preamp that is bypassed in line mode, so we can't assume the impedance is the same. Maybe it is, but I suspect its more likely to be about 10k, which is what the line in on my old sound card was and I think is typical for line level.

10k is fine though - on my old setup when I was measuring the high level output of an amplifier or the output of a passive crossover I used a small high quality audio transformer as a step down transformer - I don't recall the exact ratios but I think I was using a 3k winding as the input side and 300 ohms on the output side to the sound card input.

I originally started using this instead of an L-Pad due to earth loop problems between PC, amplifier and speaker, the transformer gives full galvanic isolation on the input side and even allows neat things like non-ground referenced measurements, such as measuring the high side driver in a series crossover which is otherwise infeasible. :)

It also worked out that the ratio was perfect to step the voltages you'll see from a typical amplifier down to a level that matched line level inputs well without any additional attenuation, and it also meant that the 10k input impedance provided by the card was effectively increased to about 100k.

So I'd just use that same transformer here when measuring crossovers in line input mode.

For small signal measurements not including the amplifier like measuring impedance of a driver I suspect I can just use the more sensitive instrument mode with the 1M input impedance directly - which should be fine for what I'll be measuring.

I'm actually liking the look of the next model up, the umc204hd:

Behringer U-PHORIA UMC204HD USB Audio Interface | DV247

It's the same as the 202HD but also has Midi input/output which would be handy, as I also lost my only midi interface when my desktop PC died, I did sometimes use that for example saving backups from my DEQ2496 which uses a Midi interface for data transfer.

It also has 6 channel output instead of 2 - again it would be nice to have a FULL manual because it's not clear if the additional outputs are usable only for things like monitoring or whether it has 6 genuine discrete digital channels from the PC like my old Audigy 2 ZS did.

If it is 6 discrete PCM channels then it opens up the possibility of using software DSP based crossovers for an up to 3 way system in the future.

Although I'm designing a passive 2 way system at the moment, I do like to make use of active EQ/DSP during the development phase to enable rapid prototyping of crossover changes digitally that will be implemented later passively, as it greatly speeds up the process and reduces the number of components I get through!

So for an extra £18 and still well within my budget I think it's a no brainer to get the MIDI and 4 extra output channels for greater versatility, and the box is only a little bit bigger and still USB powered.

I won't be able to buy anything until the end of the month so in the meantime I'll keep looking around and check out any suggestions from anyone but for the moment it looks like the U-PHORIA UMC204HD will meet my needs perfectly for a reasonable price. :)

It also occurs to me that when it's not being used for speaker testing it can be used in my A/V cabinet as a high quality DAC - at the moment most of my sources are HDMI via the TV with the analogue out from the TV going to the amplifier - OK for watching TV and playing Xbox games but not ideal for music.

The only sources I currently have connected directly to the amplifier are an Echo Dot and an old Apple Airport Express with the analogue output. The analogue quality of the Echo dot is unknown but probably not great! It sounds OK on the midi system speakers I'm currently forced to listen to but I doubt it will cut it once I have decent speakers again.

The airport express is a bit better - it at least streams from the client device in 1.4Mbps uncompressed PCM, (albeit only 44.1Khz 16 bit) but I think the DAC in the Airport Express is only average quality. (It does also have SPDIF output - but I don't have anything to connect that to)

Something like this Behringer unit would be a step up from any of my other sources as a playback DAC, plugged into a small Mediacenter PC of some sort..
 
Last edited:
Just an update - I ended up buying the Behringer U-PHORIA UMC204HD, and so far I like it. :)

Just to clarify, it has 4 playback channels not 6 as I assumed earlier.

On the rear it has 4 unbalanced line level RCA outputs, which is what I'm using for my amplifier. (Although I'm only using the "front" pair since it is only stereo)

These outputs are playback only and do not have monitor loopback from the inputs, so these are the best ones to use for full duplex speaker measurements. They also have no level adjustment so are full volume all the time.

Also on the rear are two 1/4" TRS jack plug outputs - despite looking like the socket is a "stereo" TRS that could have been a balanced output these are still only unbalanced signals. These have a gain adjustment on the front panel and also a mix control that lets you mix only playback, only input loopback or a combination of the two.

On the front is a headphone jack with the same mix facility (it shares the same mix control) but with a separate volume control of its own. I find the headphone output very useful, it has a LOT of gain available and sounds excellent.

Left and right inputs on the front support XLR microphones with/without phantom power, line level or instrument level.

Here's where things started to get a little interesting. When I did a loopback test from output to input with ARTA it showed relatively high levels of distortion - 2nd, 3rd, 4th, and so on... :confused:

I was initially worried that I might have bought a turkey...:rolleyes:

If I turned the playback level down about 10dB all but the 2nd harmonic distortion disappeared but 2nd harmonic was still only about 60dB down or about 0.05% - not terrible, but not audiophile/measurement grade either...

It took me a while to work out what the problems were.

Problem 1 - it turns out that ALL the inputs are balanced inputs, not just the XLR input. So even in line input mode using a 1/4" TRS jack the input is balanced. (The "manual" makes no mention whatsoever of which inputs/outputs are balanced or unbalanced - I had to determine this experimentally...)

By using a 1/4" TRS mono to RCA adaptor I was shorting one side of the balanced input to ground - this is not harmful and the inputs are designed to work this way, however the result is some 2nd harmonic distortion due to the imbalance.

I then tested with a floating source (my iPhone 6, which has surprisingly low distortion, barely measurable above the noise floor) connected between positive and negative balanced inputs using a stereo 1/4" jack and not connected to ground - no visible 2nd harmonic distortion at all. Touch a link between negative and ground and the 2nd harmonic distortion is back again, about 0.05%.

So, because it has balanced inputs, if you want to do accurate measurement of very low distortion you cannot use the line inputs in unbalanced "mode" with a ground referenced unbalanced signal.

This doesn't actually bother me because I have two main input sources that I will be using - XLR microphone (already balanced) and an audio isolation transformer for doing line level/crossover measurements.

I've talked about it over in the ARTA thread but for measuring the output of an amplifier or crossover I use a small audio transformer with a 1.3k input impedance (connected to the circuit under test) and a 30 ohm output impedance connected to the measurement device.

This provides a 40:1 step down ratio so that I don't need an L-Pad and gives full galvanic isolation which means no hum loops or noise pickup.

So I simply connect the output of this transformer as a balanced signal to the balanced inputs and connect ground to the can of the transformer. (Which also has a faraday shield between primary and secondary) Balanced input problem solved.

Problem 2 was that I found that the line level output of this interface is actually higher than what the line level inputs can accept - by at least 6 dB. It has very "hot" outputs at 0dB digital...

So if you do a straight loopback test from output to input with a direct connection not only do you have the unbalanced to balanced issue introducing some 2nd harmonic distortion, you are overdriving the input as well unless you turn the output volume down about 10dB...

To prove both these points I used my transformer between output and input providing some step down in voltage and also providing unbalanced to balanced conversion, set the output level to maximum and adjusted the input gain control to about 2dB below clipping - and the result was a nice flat noise floor a bit over 100dB down with no visible distortion at all above the noise floor in spectrum analyser mode. :D

I also measured the frequency response in 44.1/48/96 and 192Kbit modes - it does vary very slightly between modes (as it seems to on all cards) but in all sample rates it is flat within 0.5dB from below 10Hz up to close to the nyquist limit for the sample rate.

(I might post some frequency response and distortion measurements from ARTA if I get time if anyone is interested)

So my conclusion is that this £70 interface is more than adequate quality for doing speaker and crossover measurements to a high degree of accuracy provided that you are aware of both the output being higher level than the inputs (important during loop back tests or impedance testing etc - turn the output level down a bit or use an L-Pad) and that the inputs are balanced and that you will get a small amount of 2nd harmonic distortion if you don't provide a balanced input signal such as a transformer isolated signal. So it is not ideal if unbalanced line level inputs are your only choice and you are trying to measure distortion.

Allow for these two points when configuring your test setup and the quality is excellent.

I've also set it up with Volumio running on a Cubox-i as a small music media center when I'm not using it for measurement purposes and used it for a few hours over the weekend as a DAC for music playback and can say that I'm impressed there too.

It has plenty of output to drive the amplifier and sounds terrific. Self powered from the Cubox-i I have to turn the amplifier volume up to 12 o'clock on silence (which would normally be VERY loud) with me a couple of feet from the speakers before I hear any noise, which is primarily residual white noise from the Cubox power adaptor - no hum.

I also initially tried it on a Raspberry Pi 2 running Volumio however there is significantly more electrical noise on the 5V from the Pi which could be heard, and the Pi also suffers from audio dropouts/glitches due to the USB driver/chipset being a bit inadequate. (Something I'm still looking into)

So I wouldn't recommend this (or any USB audio) interface to be used with a Pi if you care about quality - if you have a Pi you're better off with an I2S sound card, but for devices other than a Pi this interface is certainly an option for an affordable but high quality USB DAC which can also double as a versatile testing and measurement interface for the same price.
 
Last edited:
It took me a while to work out what the problems were.

Problem 1 - it turns out that ALL the inputs are balanced inputs, not just the XLR input. So even in line input mode using a 1/4" TRS jack the input is balanced. (The "manual" makes no mention whatsoever of which inputs/outputs are balanced or unbalanced - I had to determine this experimentally...)

By using a 1/4" TRS mono to RCA adaptor I was shorting one side of the balanced input to ground - this is not harmful and the inputs are designed to work this way, however the result is some 2nd harmonic distortion due to the imbalance.
I've just had a thought regarding the issues with trying to feed an unbalanced, ground referenced line level signal into this unit, or using direct output (unbalanced) line looped back through to the input or nearly directly (simple resistor impedance measurement jig for example) and remembered that this unit also has Left and Right effects "insert" sockets at the back...

These are also stereo 1/4" TRS jack sockets, line level, and they are a loop through connection designed to connect 3rd party effects processors (fuzz pedals, harmonisers etc) that is placed after the balanced pre-amps, but before the main ADC.

These insert connections must be unbalanced because you only have three connections available - ground, in, and and out on a TRS plug.

So I should be able to "abuse" this insert connector as a direct unbalanced line level input which bypasses the balanced pre-amps. :D It will also bypass the input level control on the front panel but for certain kinds of measurements like impedance measurement using analog loopback it should be ideal.

I'll report back on whether this works well and gives 2nd harmonic distortion free unbalanced line level input - I can't see why it wouldn't.
 
Last edited:
Finally got a chance to try this and yes, the insert input can be used for unbalanced signals instead of the balanced inputs on the front. :)

Some comparisons. Here is a -6dB signal looped back from the unbalanced line output to the balanced line level input:

643776d1509743215-usb-audio-interface-measurement-balanced-input-pad-png


You can see high levels of 2nd harmonic distortion as well as some 3rd and 4th. Here is the pad enabled and gain turned up on the unbalanced input:

643777d1509743215-usb-audio-interface-measurement-balanced-input-pad-png


Reduced distortion especially 3rd but a 20dB increase in noise floor due to turning up the gain.

And finally here is the same -6dB unbalanced signal fed into the unbalanced insert loop at the rear which also bypasses all the preamps:

643779d1509743348-usb-audio-interface-measurement-unbalanced-insert-input-png


Very low noise floor and almost no second harmonic distortion. Also the insert input is able to take the full 0dB output of the line outputs (not shown in these measurements) without additional distortion while the balanced inputs cannot.

I should point out that there are no problems with distortion when the balanced inputs are provided with a balanced signal such as a transformer coupled signal, only when connected to an unbalanced and/or high level signal.

I'll be doing some impedance measurements tomorrow so I'll be using the unbalanced insert inputs directly looping back to the line level output and the sense resistor so I will see how that goes, but it looks as if it will work well.
 

Attachments

  • balanced input pad off.png
    balanced input pad off.png
    74.6 KB · Views: 4,156
  • balanced input pad on.png
    balanced input pad on.png
    73.7 KB · Views: 4,095
  • unbalanced insert input.png
    unbalanced insert input.png
    68.3 KB · Views: 4,063
Last edited:
Did some impedance measurements using LIMP today - worked perfectly. :)

What I ended up doing was using the adjustable level headphone output on the front as the signal source - at 12 oclock on the control that gives about the right output level to drive a speaker sufficiently but not overload the inputs.

I used both left and right insert inputs on the rear to connect to either side of the 100 ohm reference resistor. (ring connection for input on the plug)

So my test rig was just three 1/4" Stereo jack plugs, some cable and a 100 ohm resistor.

So I've now done every type of measurement using this interface that I would need to with speakers successfully.

For the reference input on dual channel measurements or direct connection to a crossover for crossover electrical measurement I am using an audio transformer to provide isolation (eliminates earth loops) and unbalanced to balanced conversion so that I can use the front balanced inputs, however it looks like using the unbalanced rear insert inputs with an appropriate L-Pad would also give good results.

Thumbs up for a £70 interface, when used properly, and using ASIO driver mode it works perfectly with ARTA at 192Khz 24bit and with LIMP at 96Khz 24bit. (LIMP does not do 192Khz)
 
Last edited:
Hi Simon:

Next time you do this measurement, instead of using 1,000Hz try using a slightly different frequency, one that is centered on an "FFT bin". This will eliminate the wide skirt at the base of the fundamental so you can see all the distortion harmonics in much better detail.

You can also simply select a window function in ARTA other than "Uniform" (which is the same as no window), I generally recommend Kaiser5 or Kaiser7, but this will not be quite as accurate as using the Uniform "window" and selecting a bin centered frequency per the following instructions.

Try it both ways and see what you think. :)

Example

Assume you have your hardware interface set to a 48kHz sample rate. Also assume you have ARTA's FFT size set to 32K (which is actually 32768 (32 x 1024)).

1. Bin width = (SR/FFT size) = 48,000/32768 = 1.4648

2. Bin integer # = (Frequency of interest/Bin width) = 1000/1.4648 = 682.6667, rounds to 683

3. Bin centered frequency = (Bin integer # x Bin width) = 683 x 1.4648 = 1000.4584 Hz

Simply enter 1000.4584 Hz into ARTA instead of 1000 Hz and repeat your measurement. :)

Tutorial Video.
 
Hi Simon,

Regardless of the configuration (sampling frequency, resolution, + 48V power supply, computer / laptop battery ...) my UMC202HD has a quite strong peak around 23kHz. This is my second unit, because in the first UMC202HD this tone was much lower (10 kHz) and I could literally hear it. So I sent it back to the seller. Is your UMC204 silent?

Witold
 

Attachments

  • UMC202HD-192k-24b-minGain-noInputSignal.jpg
    UMC202HD-192k-24b-minGain-noInputSignal.jpg
    228.4 KB · Views: 406
Hi Simon,

Regardless of the configuration (sampling frequency, resolution, + 48V power supply, computer / laptop battery ...) my UMC202HD has a quite strong peak around 23kHz. This is my second unit, because in the first UMC202HD this tone was much lower (10 kHz) and I could literally hear it. So I sent it back to the seller. Is your UMC204 silent?

Witold
Keep in mind I have the UMC204HD so I can't speak for the UMC202HD.

However, it is probably noise from your laptop - try a different USB port on your laptop or a different laptop.

What I found with mine is that one of the USB ports in my laptop has a problem - I'm not sure exactly what is wrong with the USB port but I suspect it is a dodgy earth connection.

When I run the interface from this port if I wiggle or hold the USB plug in a certain position I see discrete noise spikes appear on the spectrum analyzer view in ARTA which must be coming from the laptop itself.

If I plug into the USB port on the other side of the computer I see no noise spikes at all, just a flat, clean noise floor and certainly not what you are seeing.

So I would be suspicious of the USB port or induced noise from the laptop. Try another laptop if possible or at least a different USB port.

Another possibility is the laptop charger - do you still see the noise spike with the laptop running from battery ? I do see a little bit of noise from the laptop power supply on mine as well - but goes away if I run the laptop from battery.

Some laptop power supplies are extremely noisy, so I do all my measurements while the laptop is running on battery.
 
Last edited:
The problem is in the 48V step-up converter. In my device this converter works with 126...127kHz (126,8kHz) and creates peak around 24kHz (23,7kHz). When I raised frequency to 143.5kHz spurious peak moved to 7kHz. 143.5-126.8=16.7 and 23.7 - 7 = 16.7, so there are no doubts.

The switch on the back panel doesn't turn off this circuit, it only disconnects 48V from front connectors.

Witold
 
@ DBMandrake. it seems you are using UMC204HD for some time. How do you rate its performance when it comes to measurements and what is the THD+N figures when loopedback. I am confused between UMC202HD and Focusrite Scarlet solo. Many folks seems to have preference for Focusrite.
 
Member
Joined 2016
Paid Member
@ DBMandrake. it seems you are using UMC204HD for some time. How do you rate its performance when it comes to measurements and what is the THD+N figures when loopedback. I am confused between UMC202HD and Focusrite Scarlet solo. Many folks seems to have preference for Focusrite.
Any news on this guys?
I am thinking of ordering a UMC202HD over Focusrite.
The UMC202HD seams to have a larger dynamic range of 20hz to 43khz (0/+0.3db) vs 20hz to 20khz (0/+0.1db)
 
Member
Joined 2016
Paid Member
I'm completely happy with my Focusrite Scarlet.
Is anyone able to post some information on the attenuation circuit that they are using for coupling the output of a amplifier to the input of the Audio interface including the audio transformer. Just trying to piece that together. Do I measure the output and input impedances and the design a Pi-pad attenuation circuit or just go off the data sheet?