That works only if the audio latency of the USB microphone is consistently constant from one measurement to the next. That's not the case with every system (depends on computer, operating system, drivers, whatnot). I don't recommend this; avoid it if you can.
I have done it this way with two different usb mic’s on 4 different computer systems and four OS’s (Win XP, Win 2k, Win 7, Win 10). For 6 years. It hasn’t not worked. In fact, I don’t own any regular XLR analog mics. I do have XLR mic sound interfaces. They all go back to the computer via USB. Isn’t it just having ADC in the mic vs in the box (Focusrite or whatever)?
Seems too much to be just luck?
Seems too much to be just luck?
Well, if you manually adjust the relative timing of the individual measurements of each driver until you can match the wiggles of their sum measurement, you will of course be able to guesstimate the relative timing of the drivers. However, using a dual channel measurement will diectly yield data with inherently correct (absolute!) timing, with no extra manual guesswork. Why would you not prefer this way?
to AllenB's point. I recently tried an off-board USB "5.1" soundcard so I could use my old school ECM8000 behringer microphone.
The mic when used with hard wired PCI soundcard in an old lenovo desktop running Windows XP would get absolutely consistent impulse responses between measurements. From memory all 2.8msec.
When I used the USB soundcard - the timing of the impulse was wildly off (by as much as 50msec) - making measured phase crossover design absolutely useless.
Now - this might be a symptom of the crappy USB sound card, windows, the lenovo machine etc.... I dunno.
Back on topic....
IF you can / want to measure - REW and a USB umik1 will do the job
IF you want to simulate - VituixCAD graph tracing (caveats if you are interested)
The mic when used with hard wired PCI soundcard in an old lenovo desktop running Windows XP would get absolutely consistent impulse responses between measurements. From memory all 2.8msec.
When I used the USB soundcard - the timing of the impulse was wildly off (by as much as 50msec) - making measured phase crossover design absolutely useless.
Now - this might be a symptom of the crappy USB sound card, windows, the lenovo machine etc.... I dunno.
Back on topic....
IF you can / want to measure - REW and a USB umik1 will do the job
IF you want to simulate - VituixCAD graph tracing (caveats if you are interested)
That's exactly why one needs to add the reference loopback channel, which allows removing the erratic latency of the audio system during deconvolution. Nothing new under the sun.
I say: full duplex sound card (on one clock) and traditional analog mike. It's not only the latency, it's also that possible differences in sample speed between input and output channel mess up your measurement.
If you use acoustic timing reference with REW and use the other channel/speaker for it, move that speaker next to the speaker-to-measure, so that they both are roughly same distance and direction from the microphone. If you measure left speaker and acoustic timing reference comes from right in a normal listening position, echoing will smear the reference signal.
This helped me getting consistent timing measurements with USB microphone, when I had nothing else at hand.
This helped me getting consistent timing measurements with USB microphone, when I had nothing else at hand.
I’ll have to get an analog mic and try out the dual channel method. I think one still needs to go into the XO simulator program to adjust the relative acoustic center of one driver to determine the acoustic offset for the speaker system though. There is no getting around that. The data obtained from a dual channel with loop back timing ref is just more precise.
Simplified text how it all works out, maybe others fill in and please ask 🙂 VituixCAD lenghty manual and measurement manual have all this in great detail.
The acoustic offsets are included within the measurement data when one uses the dual channel method and set reference plane (distance). Impulse responses of each driver (to different angles) are measured against common reference plane and the impulse responses are converted to frequency responsed using the exact same window to preserve the relative timing in the measurements.
The reference is usually the baffle plane and each driver is measured so that the mic height is at the driver center, the mic or DUT is moved.
Then you pick a reference axis, which usually is the design/listening axis of the speaker in the simulator and input each driver relative location to the reference axis (and measurement plane) to the drivers in the simulator. This could be XYZ all 0 for the tweeter and Y -120mm for a small woofer below, but X and Z of the woofer would stay 0 since it is on the same baffle and same vertical axis, directly below the tweeter. There is also few rotations available, if there was another woofer on the side of the enclosure for example. Ayway, now the simulator is able to construct the reality as well as it can based on the data you've just given, the measurements and the relative locations.
Now, if the DUT was rotated around rotation axis which goes along reference plane and through driver on axis, and measurements were taken say in 10 degree intervals all around both horizontal and vertical (DUT laying sideways) then you basically end up with full 3D acoustic field generated by each driver available in the simulator. Based on the data the simulator is able to calculate the distances from each driver to your listening /inspection position and display accurate combined response in that location.
Now you can inspect the output of the device to any direction and distance very easily, see power response, room interaction and all kinds of stuff in matter of seconds. very powerful.
Of course there is some error depending how many measurements one took and how carefully the reference plane and rotation axis was maintained between measurements and so on, but expect it to be minor. After a while one can cheat a bit and take less measurements knowing where there might be a bit more error and still have very accurate end results.
Check out CTA-2034 standard, available free here Standard Method of Measurement for In-Home Loudspeakers (ANSI/CTA-2034
– Consumer Technology Association(R)
Of course one can freely exploit the possibilities and capabilities of the measurements and simulators for what ever means and goals. Very educational to see what happens if driver locations were changed, how the crossover slopes affect off-axis etc 🙂
The acoustic offsets are included within the measurement data when one uses the dual channel method and set reference plane (distance). Impulse responses of each driver (to different angles) are measured against common reference plane and the impulse responses are converted to frequency responsed using the exact same window to preserve the relative timing in the measurements.
The reference is usually the baffle plane and each driver is measured so that the mic height is at the driver center, the mic or DUT is moved.
Then you pick a reference axis, which usually is the design/listening axis of the speaker in the simulator and input each driver relative location to the reference axis (and measurement plane) to the drivers in the simulator. This could be XYZ all 0 for the tweeter and Y -120mm for a small woofer below, but X and Z of the woofer would stay 0 since it is on the same baffle and same vertical axis, directly below the tweeter. There is also few rotations available, if there was another woofer on the side of the enclosure for example. Ayway, now the simulator is able to construct the reality as well as it can based on the data you've just given, the measurements and the relative locations.
Now, if the DUT was rotated around rotation axis which goes along reference plane and through driver on axis, and measurements were taken say in 10 degree intervals all around both horizontal and vertical (DUT laying sideways) then you basically end up with full 3D acoustic field generated by each driver available in the simulator. Based on the data the simulator is able to calculate the distances from each driver to your listening /inspection position and display accurate combined response in that location.
Now you can inspect the output of the device to any direction and distance very easily, see power response, room interaction and all kinds of stuff in matter of seconds. very powerful.
Of course there is some error depending how many measurements one took and how carefully the reference plane and rotation axis was maintained between measurements and so on, but expect it to be minor. After a while one can cheat a bit and take less measurements knowing where there might be a bit more error and still have very accurate end results.
Check out CTA-2034 standard, available free here Standard Method of Measurement for In-Home Loudspeakers (ANSI/CTA-2034
– Consumer Technology Association(R)
Of course one can freely exploit the possibilities and capabilities of the measurements and simulators for what ever means and goals. Very educational to see what happens if driver locations were changed, how the crossover slopes affect off-axis etc 🙂
Last edited:
tmuikku described the process well.
Having worked with both a USB mic and a 2-channel audio interface + XLR mic, I see the advantages of both types. It appears that with the timing reference feature of REW, it may be possible to collect the kind of data that VituixCad needs to do a proper simulation... but it will be a slow process.
With a 2-channel audio interface and an XLR mic, the process to collect impulse responses in 10 degree (or 15 degree) increments is rapid. In ARTA, we can set the gate window and time-of-flight offset for all the many measurements all at once, instead of individually for each curve. I believe REW has the same feature. This is what makes it so fast.
j.
Having worked with both a USB mic and a 2-channel audio interface + XLR mic, I see the advantages of both types. It appears that with the timing reference feature of REW, it may be possible to collect the kind of data that VituixCad needs to do a proper simulation... but it will be a slow process.
With a 2-channel audio interface and an XLR mic, the process to collect impulse responses in 10 degree (or 15 degree) increments is rapid. In ARTA, we can set the gate window and time-of-flight offset for all the many measurements all at once, instead of individually for each curve. I believe REW has the same feature. This is what makes it so fast.
j.
Question is, in what way does REW determine timing reference. ARTA does the same in single channel mode at (from my mind) 20% of the max level of the impulse. Problem is the impulse response of a tweeter is quite different from that of a woofer (rise times differ) so the 20% level shifts in time with the bandwidth of the DUT. Theoretically one could compensate for such. But then again, the problem of non-synchronous sampling remains. To me there is not much to say for USB mics when designing loudspeaker systems.
What is the advantage of a "USB microphone" over a good old analog mic on a 2-channel sound card?
I Agree, assuming a person has a two channel rig, and has worked out all the setup, calibration, and bugs. For a person with no (zero) microphone or recording experience, such as I was two years ago, a USB mic is a good way to start. There are so many fewer opportunities for error. My USB mic was a turn-key system that enabled me to get started right away. There is no ambiguity to it's measurements, 96 dB SPL is for certain 96 dB SPL... no uncertainty about preamp gain or mV/Pa.
As an analogy, if someone with no wood working experience asked me for advice on sanding a maple table top to get it smooth, and the only tools available were (1) a hand sanding block and a stack of 80 grit paper, or (2) a belt sander... Well I would recommend using the sanding block. It will take a lot longer, but the chance of ruining the table is limited.
j.
Last edited:
Ok, I see where you are coming from. However, hooking up an analog microphone and a USB soundcard is hardly more difficult than hooking up a "USB microphone". The analog mic with a soundcard will allow accurate timing, which is very challenging (if not impossible) with a "USB microphone".
Forgive my ignorance, what is the difference between a 2-channel USB audio interface, and a microphone with a built-in USB interface? More to the point, what exactly is the difference that allows proper time-of-flight/phase with the former but not with the latter?
Dave Bullet mentioned that a USB sound card still had timing issues... Do we need something onboard or PCIe-based?
EDIT: Sorry - seems like it's the loopback that is important. So a USB interface would be fine as long as the the IO are analog.
Last edited:
From the REW help file:
"If the input and output are on the same device and so share a common clock longer sweeps will provide higher signal-to-noise ratio (S/N) in the measurements. Long sweeps may be problematic when the input and output are on different devices, such as when using a USB mic, as their sample clock rates will differ. Over a long sweep a significant difference in clock rates could cause severe distortions in the shape of the impulse response and affect the phase response."
REW has added several features (e.g. acoustic timing reference) that compensate for the timing problems that can occur with USB mics. You do need to read the help file for other measurement and level calibrations peculiar to a USB mic. I don't know how well the USB mic features work as I've always used an analog microphone.
"If the input and output are on the same device and so share a common clock longer sweeps will provide higher signal-to-noise ratio (S/N) in the measurements. Long sweeps may be problematic when the input and output are on different devices, such as when using a USB mic, as their sample clock rates will differ. Over a long sweep a significant difference in clock rates could cause severe distortions in the shape of the impulse response and affect the phase response."
REW has added several features (e.g. acoustic timing reference) that compensate for the timing problems that can occur with USB mics. You do need to read the help file for other measurement and level calibrations peculiar to a USB mic. I don't know how well the USB mic features work as I've always used an analog microphone.