Wesayso ,
How important do you think it is that we have the audio chain removed how much err is in the audio chain, and lastly , what about in a 4 way with amps of different brand and topology, how would that be accomplished?
Thanks🙂 ,
Andrew
The analog electronics should have no appreciable phase shift in the audio range unless there is a filter (crossover/equalizer) or some marginal electronics that doesn't have flat response across the audio band. You should look at a driver as the composite of its electrical and acoustic response which would come from the microphone.
For a system the best way to test is the composite system exciting it from the electronic input to the acoustic output. You will need an acoustic test setup like ARTA or REW and you will need to be able to back out the delay through the electronics (any DSP) and the air path. Then you will have the real phase/time response. The other option is the Hilbert transform of the acoustic response which assumes that the speaker system is "minimum phase" as in now extra delays or phase shifts. With DSP this assumption may not hold.
While its nice to have a microphone with effectively perfect phase response well beyond the audio band you will find the drivers will be the bigger issue. Cancellation from diffraction across the front of the speaker will make a mess of your phase and frequency response.
In testing a Pioneer ribbon with response to 70 KHz I was shown what I was told to expect- moving the microphone 1/8" will change everything. The sound source is large enough to cause cancellations all by itself. I could get useful results to about 35 KHz. The phase response was all over the map.
Fortunately hearing doesn't work that way. Above about 4 or 5 KHz you are hearing envelopes of HF, not discrete waveforms. Can you really tell the difference between 12 KHz and 13 KHz bursts? Low IM and good tone burst performance may be much more important than phase coherent waveforms.
For a system the best way to test is the composite system exciting it from the electronic input to the acoustic output. You will need an acoustic test setup like ARTA or REW and you will need to be able to back out the delay through the electronics (any DSP) and the air path. Then you will have the real phase/time response. The other option is the Hilbert transform of the acoustic response which assumes that the speaker system is "minimum phase" as in now extra delays or phase shifts. With DSP this assumption may not hold.
While its nice to have a microphone with effectively perfect phase response well beyond the audio band you will find the drivers will be the bigger issue. Cancellation from diffraction across the front of the speaker will make a mess of your phase and frequency response.
In testing a Pioneer ribbon with response to 70 KHz I was shown what I was told to expect- moving the microphone 1/8" will change everything. The sound source is large enough to cause cancellations all by itself. I could get useful results to about 35 KHz. The phase response was all over the map.
Fortunately hearing doesn't work that way. Above about 4 or 5 KHz you are hearing envelopes of HF, not discrete waveforms. Can you really tell the difference between 12 KHz and 13 KHz bursts? Low IM and good tone burst performance may be much more important than phase coherent waveforms.
With a program like REW you could loop the output to your mic or line in and run a sweep. Once you put in amplifiers you'd need to attenuate the sound before looping back into the input. I didn't dare to put the amp in between but checked the output of my DAC and processing chain with loop back.
The amplifier will be in the chain during speaker tests, so it gets it's turn separately. If you trust your microphone enough to accept the results. 🙂
It's quite revealing to test the DAC or any processing chain you might have. Not all of them test as perfect as we might assume. I was able to optimise the buffer settings on my Asio driver to get a better result. You might notice some DAC's are optimised for a specific sample rate (which may not be the sample rate it's used with) which can be a cause of high frequency ripple in other sample rates.
@ 1audio,
I am actually more interested in the phase shift on the lower frequency spectrum.
About the procedure you described, are you talking about using the microphone as a speaker and then measuring it's output? But how would you measure it? Wouldn't you need to trust the measurement microphone used? Which is kind of a catch 22 situation, again.. 😕
I am actually more interested in the phase shift on the lower frequency spectrum.
About the procedure you described, are you talking about using the microphone as a speaker and then measuring it's output? But how would you measure it? Wouldn't you need to trust the measurement microphone used? Which is kind of a catch 22 situation, again.. 😕
Actually the 1" B&K make good test sources in the lab for testing microphones but not for this.
I was probably unclear. first the microphone will be free pf phase shift down to around 100 Hz. below that they typically have a rear port that causes a rolloff with an associated phase shift. if you can plug the port(mostly at least) then the LF response goes even lower without phase shift. Its all pressure then. What will get you is the room's interaction at low frequencies. Take the speaker outside and mike close to get the most accurate speaker response.
I think Arta compares the input to the amp to the signal from the mike which should encompass all the pieces you would need. In your case the input to the DSP crossover stuff.
I was probably unclear. first the microphone will be free pf phase shift down to around 100 Hz. below that they typically have a rear port that causes a rolloff with an associated phase shift. if you can plug the port(mostly at least) then the LF response goes even lower without phase shift. Its all pressure then. What will get you is the room's interaction at low frequencies. Take the speaker outside and mike close to get the most accurate speaker response.
I think Arta compares the input to the amp to the signal from the mike which should encompass all the pieces you would need. In your case the input to the DSP crossover stuff.
Demian,
If you plug the rear port I suspect changes in atmospheric pressure will destroy the capsule. If you try it do let me know!
I would be tempted to just use two drivers with reciprocity.
ES
If you plug the rear port I suspect changes in atmospheric pressure will destroy the capsule. If you try it do let me know!
I would be tempted to just use two drivers with reciprocity.
ES
Lots of ways to skin the cat. Plugging the rear was suggested by a mike vendor to me. You can't ship by air afterwards. Really all you need is to restrict the port. The 100 Hz or 50 Hz they pick is usually because these are for voice applications and the LF just causes problems. Same with MEMS mikes. The MEMs mikes have better consistency (+/- 1 dB sensitivity is common) but big rear ports and convoluted acoustic paths. Again plug the rear ports for use in feedback noise cancelling. . .
Thanks!
So in my case I have a USB mic and a creative USB sound card and it goes optical into my processors than out to the amps.
So in the that instance would rew catch the difference in the measurement?
As far as mic I have umik1 and the umm6.
I also have a ecm8000 (uncalibrated) and an audio control mic from my audio control RTA from way back when.
Should I be using a xlr mic to do this?
I also use a Dirac box along with some opendrc stuff , Dirac seems to make things sound better down low, do you think I'll get a better result if I take the time to learn this and apply it. Seems to me I would benefit and I always question these things. If I make measurements with the two USB nice I have now I do get different results especially in the midrange
@1audio
My initial conundrum, still unanswered, is how are measurement microphones measured themselves, when it comes to phase over the 20Hz-20kHz range?
So far it seems that the phase is calculated through the Hilbert transform for the range where the microphone is reasonably ASSUMED to be a minimum phase device (100Hz-20kHz+), and in the 20-100 Hz range, the assumption is still made and the phase shift calculated in the same way.
Am I correct in saying this?
Could the following be a good way to actually measure (not calculate) the phase of a microphone?
You take two microphones of exactly the same kind.
You use one as a speaker (mic 1), and the other (mic 2) as a measurement microphone.
You couple them in a pressure chamber (will this change the mic response from what it would normally be under normal conditions?)
You measure the difference between the signal fed to mic 1 and the one picked up from mic 2 (after some time realigning considerations).
You analize the difference and divide the phase shift by 2.
That should be the real phase shift of the microphone.
Do you see anything wrong in my reasoning?
Do you know if any company out there goes into the trouble of doing this (or something equivalent) to provide real measured phase shift?
From there, do you think there would be any caveat in buying a microphone that has been calibrated based on the microphone above, if one can't afford buying the real thing?
My initial conundrum, still unanswered, is how are measurement microphones measured themselves, when it comes to phase over the 20Hz-20kHz range?
So far it seems that the phase is calculated through the Hilbert transform for the range where the microphone is reasonably ASSUMED to be a minimum phase device (100Hz-20kHz+), and in the 20-100 Hz range, the assumption is still made and the phase shift calculated in the same way.
Am I correct in saying this?
Could the following be a good way to actually measure (not calculate) the phase of a microphone?
You take two microphones of exactly the same kind.
You use one as a speaker (mic 1), and the other (mic 2) as a measurement microphone.
You couple them in a pressure chamber (will this change the mic response from what it would normally be under normal conditions?)
You measure the difference between the signal fed to mic 1 and the one picked up from mic 2 (after some time realigning considerations).
You analize the difference and divide the phase shift by 2.
That should be the real phase shift of the microphone.
Do you see anything wrong in my reasoning?
Do you know if any company out there goes into the trouble of doing this (or something equivalent) to provide real measured phase shift?
From there, do you think there would be any caveat in buying a microphone that has been calibrated based on the microphone above, if one can't afford buying the real thing?
Just curious how in the world could you use a condenser as a speaker ?
I'm following all these posts and reading along with everybody ,
That's the part that I don't understand sax512 are you talking about using diaphragm mics?
I'm following all these posts and reading along with everybody ,
That's the part that I don't understand sax512 are you talking about using diaphragm mics?
electret or DC biased "condenser mics" are also "ESL speakers", AKA "reciprocal transducers"
RF modulator types aren't reciprocal
RF modulator types aren't reciprocal
I reread post 63 and it didn't really specify I just wanted to make sure were all on the same page here ,
If I look at a diagram of a condenser mic , it looks like it has some sort of diode attributes to it .
If it behaves like a Dioed , in anyway wouldn't that make it impossible ?
Sorry I'm not keen on electronic components or schemes.
An externally hosted image should be here but it was not working when we last tested it.
free photo hosting
That triangle is the capsule's preamplifier.
But your observation is still valid, as it would need to be confirmed if the preamplifier allows the signal to go from its output to its input in the same "optimized" way as the other way around (impedance matching, etc..).
I guess it depends on the design of the preamp.
But your observation is still valid, as it would need to be confirmed if the preamplifier allows the signal to go from its output to its input in the same "optimized" way as the other way around (impedance matching, etc..).
I guess it depends on the design of the preamp.
No reciprocity applies only to the transducer itself. Electrets have a built in bias, purely capacitive condensers need external bias just as an electrostatic speaker, electrostatic force is attractive so unbiased the plates would be attracted to each other on both 1/2 cycles.
Last edited:
I only tested my digital audio chain with loop back. This included REW -> JRiver (as my player with it's DSP engine for FIR and EQ) -> SP/Dif -> DAC output to line in on soundcard.
This is the result in APL_TDA, I also have the REW files, but this one is more visual:
I have a behringer calibrated mic, calibrated by iSEMcon;
I also use their mic pré-amp as phantom power source. The calibration contains phase and FR tables.iSEMcon said:
I accept the error of not knowing how far the B&K reference might deviate in phase response for the low end.
My FIR corrected arrays (stereo left + right) as measured at the listening position in APL_TDA:
My phase is following the minimum phase of the band pass behaviour of my speakers, I have tried linear phase correction too. This minimum phase correction is what I ended up with and sounds most natural to me.
This isn't a product of digital correction, it's absorption panels plus choice of speakers and their room integration combined or topped off with DSP.
But that's one of the main points of this thread!
I appreciate that you accept this uncertainty, still I can't believe that that's the best one can hope for.
- Status
- Not open for further replies.