A couple days ago I measured a set of speakers (in-room). After a few minutes I repeated the measurement with the exact same settings (maybe apart from signal level and mic sensitivity), without changing the placement of mic and speakers. To my surprise both measurements were quite different. Today I repeated the process, and again, two measurements a couple minutes apart show very different results.
Measurement setup
Amp: Hypex NCore based (AudioPhonics MPA-S250NC)
Audio Interface: Behringer UMC22
Mic: Dayton EMM-6
Software: REW 5.31.3
Driver: ASIO4All 2.16
Today I loaned an Audio Interface from a friend of mine, a Focusrite Scarlett 8i6. I was hoping to eliminate this link in the chain. Below are a few measurements.
As can be seen, there's much less variance between these measurements. But still over 4dB, which seems way too much to me.
Has anyone else had issues with inconsistency in measurements?
What whould be the most likely cause?
Measurement setup
Amp: Hypex NCore based (AudioPhonics MPA-S250NC)
Audio Interface: Behringer UMC22
Mic: Dayton EMM-6
Software: REW 5.31.3
Driver: ASIO4All 2.16
Today I loaned an Audio Interface from a friend of mine, a Focusrite Scarlett 8i6. I was hoping to eliminate this link in the chain. Below are a few measurements.
As can be seen, there's much less variance between these measurements. But still over 4dB, which seems way too much to me.
Has anyone else had issues with inconsistency in measurements?
What whould be the most likely cause?
Last edited:
Try looking at the ungated measurement. Any differences in gate length or set-up can skew things. I would expect about a 1-2 db variance. The shorter the gate the less accurate as you go lower, If you look at your measurement you can see the error's getting worse with frequency going lower. Can you go outside? Or see what your repeatability is ungated and use the gated to mark out the room modes.
Rob 🙂
Rob 🙂
Last edited:
Psychoacoustic smoothing. Measurement was done in-room, with mic in the listening position, and speakers in their normal position. Between all these measurements the placement of mic and speakers was the same.
The intention was to measure the response at the listening position, and have REW generate a PEQ profile that I wanted to set in my WiiM Pro.
The intention was to measure the response at the listening position, and have REW generate a PEQ profile that I wanted to set in my WiiM Pro.
Hmm, I only have one amp, it's a Hypex NCore (NC252MP based) one. I doubt it's to blame, but I'll try to find another amp to verify anyway. I did change out the XLR cable between the mic and audio interface, but that didn't solve it.
My amp can handle 2 ohm loads, but my speakers don't dip below 3.8 ohm. The RCA cable could be the culprit, but I used the same cable for both the Behringer and the Focusrite audio interfaces, but the Behringer is much more inconsistent. So I'm hesitant to blame the RCA cable.
Good idea. I did do one soundcard calibration in REW for the Behringer UMC22 (using a loopback cable). Tomorrow I'll do a few more, with a couple minutes in between. Trying another mic might prove difficult. I don't know anyone that owns a measurement mic.
I have a Dayton EMM6 mic that is defective, and it was defective from day one. It has a dramatic roll off in the low end. It rolled off at about 3 dB per octave from 500 Hz down to 100 Hz, then about 4 dB per octave from 100 Hz down to 20 Hz. The calibration file did not correct this response problem.
Some wise jedi on this forum advised me that there was probably some leakage around the capsule, and this caused a cancellation of low frequencies. If your mic has a leakage which is intermittent, it could cause the non-repeatable error you are experiencing.
Some wise jedi on this forum advised me that there was probably some leakage around the capsule, and this caused a cancellation of low frequencies. If your mic has a leakage which is intermittent, it could cause the non-repeatable error you are experiencing.
What is that 1/3 octave or 1/6? If you are measuring in a couple of meters away with no averaging you are going to get variations. You have an RTA function? Just use that to take a look, the lower the number of samples the greater the variation especially down low. Or if that is an RTA like measurement change the sample size from say instantaneous to 10/100/1000 and watch the variation get smaller between measurements.
Rob 🙂