I should have said nothing wrong with the MiniDSP, Behringer or Dayton mic for the purpose. I think they have similar internals but that doesn't matter compared to consistent test setup and results.
I had a look at Speaker Bob's link and they had this Calibrated mic:
https://www.gear4music.com/Recordin...y5lwc*_ga*MjAyNzE4MTI2Ni4xNjc0MTEyMTg0*_up*MQ..
Would this be a good compromise, or would I be better getting the ECM8000 for timing and a UMIK1 for frequency as recommended by Bob?
From what some people say, not all calibration files are equal....!
https://www.gear4music.com/Recordin...y5lwc*_ga*MjAyNzE4MTI2Ni4xNjc0MTEyMTg0*_up*MQ..
Would this be a good compromise, or would I be better getting the ECM8000 for timing and a UMIK1 for frequency as recommended by Bob?
From what some people say, not all calibration files are equal....!
Something to consider: most of these small capsule mikes are very linear, due to their size. Even the ECM8000 is. The main consideration is: is the mike free field corrected or diffuse field corrected. The latter correction (found on the Beyerdynamic and the Behringer) leads to a rising response on-axis (we speaker designers need on axis response). Mind you, they all measure flat up to 4-5kHz, so for crossover design all these mikes mostly do fine.
The diffuse field correction can easily be corrected (sort of) by tilting the mike and measure on about 45 degrees. You keep deviations above about 10kHz, but these are small and certainly negligible for starters and occasional users.
The diffuse field correction can easily be corrected (sort of) by tilting the mike and measure on about 45 degrees. You keep deviations above about 10kHz, but these are small and certainly negligible for starters and occasional users.
My Umik has a 0 degree and 90 degree calibration file, so I'd assume most capsule mics are not truly omnidirectional. (think my ECM did as well)
I'm in the beginning phase of learning to take speaker measurements. Are you saying that in a 2-way speaker for instance, the HF and LF drivers should be measured separately?
Yes that is correct. After you install the drivers in the cabinets, make acoustical and impedance measurements of the drivers. At this point there is no crossover. For best results, you will make horizontal polar measurements from 0 to 180 degrees for each driver. The gated far field measurements are merged with near field measurements to get a composite response that is used in simulation.
No, i was replying to Qts, who wants to measure left and right speaker at the same time. You need to read the post above and not to take my reply out of context.
Just like hifijim said, you measure each driver separately, besides on axis, i measure of axis to give me an idea about beaming. Then you select crossover point, steepness, and so on. There are programs and web pages which will give you optimum. Then you buy the parts, assemble, measure separately each driveragain, and together. I measure both polarities to reveal proper phase, just in case. Then you adjust if needed, use l-pads or fixed resistors to match sensitivities.
Goal is flat on axis and well behaved of axis.
But you never develop crossover by measuring left and right speaker at the same time. Idiots do. There is even one guy measuring commercial speakers in his room, both left and right speakers at the same time with one mic. And he is wondering why he has all the dips and peaks on high frequencies. In contrast to manufacturers data or reviews in stereophile. He even corrects this with minidsp and publishes his correction file. Not realizing double mistake, besides the comb filter, its only applicable for his room.
Cheers!
Okay, what time? I am not interested in to use 22 mins looking that guy.
And by the way adason. You can call me idiot if you like but check first what is dual channel measurement. It is not left and right together, at least when we are talking about measurement equipment for speaker development.
And by the way adason. You can call me idiot if you like but check first what is dual channel measurement. It is not left and right together, at least when we are talking about measurement equipment for speaker development.
Last edited:
I think the standard is actually 90°. Ideally it should be specified in the datasheet, but that's too much to ask often...
Attachments
I'm setting up to do this sort of work as well. On the mic I chose Dayton USB mainly because a calibrated SPL can be measured directly with any weighting I want to use including none. So say I find I must have a none USB mic. I have a Behringer which I assume I could calibrate if needed. Actually I already had one but would still go down the same route having seen comments about purchasing a decent sound meter and maybe a calibrater for that.
USB latency? With more modern kit I wonder if this is old hat. The more modern standard has a lot more bandwidth than this used to have. The analogue route in - Win decided that they needed to improve that on win10 and up. How significant this is - pass. I run Linux anyway and will be switching to a low latency kernel. I mentioned this to someone else on here - they said why do I want that. Just made me smile - one reason for offering it is sound.
USB latency? With more modern kit I wonder if this is old hat. The more modern standard has a lot more bandwidth than this used to have. The analogue route in - Win decided that they needed to improve that on win10 and up. How significant this is - pass. I run Linux anyway and will be switching to a low latency kernel. I mentioned this to someone else on here - they said why do I want that. Just made me smile - one reason for offering it is sound.
You can assume that most of the cheap mikes do have a rising response, but you don’t know how much. I have an ECM here that only deviates about 2dB at 15k free field. So if you don’t know what you have, don’t go for full diffuse field correction.
Mind you, in the above curve the deviation at 10kHz is about 1dB. A lot of people will not be able to tell the difference.
have
Thank you for sharing your observations on the challenges with dual-channel measurements and USB device variability.
Do you think that a professional audio interface connected via PCI-E (like ESI MAYA44 eX)
might offer a solution if want to stick with Kimmosto´s recommended dual-channel measurement procedure for Vcad? Moving away from USB interfaces like the Steinberg UR mkII would potentially enhance the precision and consistency required for these advanced measurement techniques?
I wouldn’t know why the USB aspect would be responsible for the mentioned issue. My interfaces don’t show it. OTOH in another thread on another forum it was discussed that sometimes an Xmos chip is involved in the digital signal path and that, combined with poor firmware and wrong driver/windows settings, can lead to issues.
Hi Dave, Must say that sounds like a unique problem...I'd suspect something is wrong with your rig, if you you were having variable timing problems with an external USB soundcard.
I mean, the entire professional measurement world uses dual channel FFT with external devices. ????
Phase can (and should be, imso) measured for each driver section independently, by keeping a locked timing reference and constant mic position.
VCAD Help describes it well.
My preference is to use the HF/VHF driver as the timing reference, as it always gives the most consistent/repeatable time-of-flight (no matter whether dual or single channel)
With a ton of experience, I think the combined measurement technique can be made to work almost as well as making separate driver section measurements. But it takes a ton of experience for lower freq crossovers to do so, and is super difficult at best for subwoofer time alignments, ime.
Reason being, a full sweep HAS to tie together phase to make it representable. Which means the phase trace representation in the xover range is an average of the two sections, sections which may or may not have entirely overlaying phase traces.
The only way to see how well they truly align (overlay), is to see the phase traces separately. Tis why the "phase trace overlay method" is the most commonly touted prosound recommendation for phase and time alignment.
Actually not unique. We just discussed the same on our Dutch forum. A 2i2 2nd gen giving various results, as in no stable timing reference between out- and input. Like I said it probably boils down to the digital signal processing in the unit and as such doesn’t stem from the USB part.
Possibly a problem. I did see the impulse with no changes vary by up to 10 msec or so from memory. Although a loopback would have got relative phase to the particular impulse, it would have been useless when the impulse moved for the next DUT. In other words, I can get the flight time (measured phase) but not relative phase between drivers (with USB timing shifting the impulse). I can't recall the USB device I used nor PC. Was some time ago.
I do measure each driver separately - gated (farfield). The combined measurement is purely to work out Z (acoustic offset) for drivers when using minimum phase data in the crossover.
To prove the accuracy of the combined driver measurement method to get Z, use derived phase and not measured phase...., here is a commercial speaker I redesigned from a 2 way TMM layout to 2.5 way TMM with Tweeter replacement (brand / model entirely). Blue is final speaker measurement, red is the simulation
Good to know, thx.
Do you know of any other USB cards having such a problem? Or has it seemed confined to that unit?