I wrote to Dr Jordan as I wanted to procure a microphone with a MK301E capsule.
I’ll be looking at HF and LF units too.
@A4eaudio will be covering the linear distortion (frequency response) side of things in weeks to come…
I’ll be looking at HF and LF units too.
@A4eaudio will be covering the linear distortion (frequency response) side of things in weeks to come…
Thank you @tktran303 for all your effort.
Purifi could use the "S mic" in the future.
Obviously the D3 harmonic of common measurement microphones is very high, therefore current-drive (or resistor-drive as I implemented and measured it) never became the normal mode of operation.
Best regards
Bernd
Purifi could use the "S mic" in the future.
Obviously the D3 harmonic of common measurement microphones is very high, therefore current-drive (or resistor-drive as I implemented and measured it) never became the normal mode of operation.
Best regards
Bernd
Last edited:
What are you talking about? All the measurements here show that microphones mainly have H2 and all higher harmonics are magnitudes lower.
@Hörnli
I’m glad you encouraged me to push forward with testing via dynamic microphones. Their high SPL handling capability means that their own self distortion are closer to being able to match the classically powered B&K and G.R.A.S. capsules with external polarisation.
As we can see, there’s variation in both H2 and H3. How much depends on how close the mic is to its limit, but also peculiar differences in the mic. Is it the mic capsule or the electronics?
I remember we were commiserating about how “DIY is dead”. 🤣 I wasn’t ready to accept that yet, so I dutifully followed your lead to look at STEPS. As I had to RTFM to set it up...I realised that Ivo had already down a mic comparison decades ago.
Well knock me down with a feather!
"Three microphones were compared: an inexpensive model (MM-1 T-Bone, costing around € 35), a mid-range model (Audix TM1, around € 300), and a class 1 reference microphone (NTI M2210, cost around € 1100)"
I was suspicious that €€€€ could be the only factor, so went to look at the mic specs a bit closer and…
MM-1 has max SPL of 110dB (1% THD)
TM1 has max SPL of 130 (1% THD)
NTI M2210 has max SPL of 144dB (3% THD)
( ?134dB @1% THD)
@IamJF showed us his ECM8000, MicW M215 and Earthworks M50 and B&K 40BD
—> large difference in maximum SPL, and corresponding differences in THD at specific observed SPL.
Now, with a dynamic omnidirectional microphones do not have flat perfectly flat frequency response, so there’s no free lunch (yet). A user has to create an electronic compensation file from a known reference. My own compensation file was created using the Earthworks M23 as a baseline comparator. And Earthworks mics calibrated using an ACO Pacific microphone AFAIK. So we are comparing a mic to a compared mic, so any error might add up, or average out; it’s hard to know which.
Sonarworks appears to be comparing their mics to a “ANSI reference mic”, and it appears that miniDSP are also using a microphone substitution technique.
But suppose we are off by even +/- 1dB), that seems close enough for distortion measurements with modern software.
So it’s not calibration by an electrostatic actuator setup, but it’s certainly not “plug and play” . If one is not experienced in measuring speakers or microphones, I wonder how much error would creep in from using the generic curve from the data-sheet.
As @bwaslo explained succinctly, a mic’s (or recording chain’s) non-flat frequency response doesn’t affect the ability to measure the frequency response- as long as it is known and be taken into account. On the other hand, it may affect dynamic range and noise if the compensation is excessive.
How much is too much?
Well, I’d think anything around 3dB compensation is reasonable. 6dB at the maximum. More than means that boosting the noise by double creates more imprecision. As you can see to the right of the graph, the uncertainty goes up as the frequency response of the Sennheiser MD42 falls off a cliff in the top octave.
Other microphones options include Sennheiser MD21-U, a classic dynamic microphone that has been around for 60 years. I’m not sure how the older models do, like speakers and voice coils probably shift with increasing age.
There’s also the Audio Technica 4022 omnidirectional condenser mic with self noise 13dB(A) and of 160dB @3 % THD. ?110dB with 0.01%. This is as close as I can find without going for a classical setup.
Hopefully @bwaslo can pull me out of this microphone deep dive by offering a wide bandwidth microphone out to 40KHz with good SPL handling capacity and low noise in the next iteration of Omnimic!
I’m glad you encouraged me to push forward with testing via dynamic microphones. Their high SPL handling capability means that their own self distortion are closer to being able to match the classically powered B&K and G.R.A.S. capsules with external polarisation.
As we can see, there’s variation in both H2 and H3. How much depends on how close the mic is to its limit, but also peculiar differences in the mic. Is it the mic capsule or the electronics?
I remember we were commiserating about how “DIY is dead”. 🤣 I wasn’t ready to accept that yet, so I dutifully followed your lead to look at STEPS. As I had to RTFM to set it up...I realised that Ivo had already down a mic comparison decades ago.
Well knock me down with a feather!
"Three microphones were compared: an inexpensive model (MM-1 T-Bone, costing around € 35), a mid-range model (Audix TM1, around € 300), and a class 1 reference microphone (NTI M2210, cost around € 1100)"
I was suspicious that €€€€ could be the only factor, so went to look at the mic specs a bit closer and…
MM-1 has max SPL of 110dB (1% THD)
TM1 has max SPL of 130 (1% THD)
NTI M2210 has max SPL of 144dB (3% THD)
( ?134dB @1% THD)
@IamJF showed us his ECM8000, MicW M215 and Earthworks M50 and B&K 40BD
—> large difference in maximum SPL, and corresponding differences in THD at specific observed SPL.
Now, with a dynamic omnidirectional microphones do not have flat perfectly flat frequency response, so there’s no free lunch (yet). A user has to create an electronic compensation file from a known reference. My own compensation file was created using the Earthworks M23 as a baseline comparator. And Earthworks mics calibrated using an ACO Pacific microphone AFAIK. So we are comparing a mic to a compared mic, so any error might add up, or average out; it’s hard to know which.
Sonarworks appears to be comparing their mics to a “ANSI reference mic”, and it appears that miniDSP are also using a microphone substitution technique.
But suppose we are off by even +/- 1dB), that seems close enough for distortion measurements with modern software.
So it’s not calibration by an electrostatic actuator setup, but it’s certainly not “plug and play” . If one is not experienced in measuring speakers or microphones, I wonder how much error would creep in from using the generic curve from the data-sheet.
As @bwaslo explained succinctly, a mic’s (or recording chain’s) non-flat frequency response doesn’t affect the ability to measure the frequency response- as long as it is known and be taken into account. On the other hand, it may affect dynamic range and noise if the compensation is excessive.
How much is too much?
Well, I’d think anything around 3dB compensation is reasonable. 6dB at the maximum. More than means that boosting the noise by double creates more imprecision. As you can see to the right of the graph, the uncertainty goes up as the frequency response of the Sennheiser MD42 falls off a cliff in the top octave.
Other microphones options include Sennheiser MD21-U, a classic dynamic microphone that has been around for 60 years. I’m not sure how the older models do, like speakers and voice coils probably shift with increasing age.
There’s also the Audio Technica 4022 omnidirectional condenser mic with self noise 13dB(A) and of 160dB @3 % THD. ?110dB with 0.01%. This is as close as I can find without going for a classical setup.
Hopefully @bwaslo can pull me out of this microphone deep dive by offering a wide bandwidth microphone out to 40KHz with good SPL handling capacity and low noise in the next iteration of Omnimic!
I find these dips with the "S mic" kinda strange, there isn't anything to believe that something would behave that way.
Btw, it's important to disclose and tell what brand and name that mic is, otherwise it makes the results not really valid.
From the start post I understand it's the Sennheiser MD42?
It's a bit confusing to open yet another mic topic again, since I just posted some results in the other one..............
I am also confused why this is in the multi-way loudspeaker category and not in the equipment & tools or acoustics section?
Btw, it's important to disclose and tell what brand and name that mic is, otherwise it makes the results not really valid.
From the start post I understand it's the Sennheiser MD42?
It's a bit confusing to open yet another mic topic again, since I just posted some results in the other one..............
I am also confused why this is in the multi-way loudspeaker category and not in the equipment & tools or acoustics section?
Keep in mind that those are Leq levels, meaning there is a time average.
Although often the details are very blurred at best.
This more to place certain arguments within context if people want to actually measure those kind of things.
It's important to understand the little nuances sometimes 😉
Last edited:
I did it first:
https://www.diyaudio.com/community/...ortion-measurement.402566/page-8#post-7556973 (#157)
@tktran303 did calibration. So he may drop your name next time @IamJF
Last edited:
Next we measure a 12" woofer in a 1cu ft sealed box.
With a drive level of just 1V- this should be easy work for any 12" woofer!
We measure with a (Farina) log sine sweep, 128K samples x 8. Multiple measurements average the coherent noise, with improves the measurement.
Then I move the microphone from 6"/15cm, 3"/7.5cm, 1"/2.5cm, 1cm:
Since the speaker is just driven with 1V, any increase in distortion at closer observation distance must be coming from the microphone:
So the Earthworks M23 should not be used in the nearfield to measure distortion- it contributes significant distortion of it's own.
With a drive level of just 1V- this should be easy work for any 12" woofer!
We measure with a (Farina) log sine sweep, 128K samples x 8. Multiple measurements average the coherent noise, with improves the measurement.
Then I move the microphone from 6"/15cm, 3"/7.5cm, 1"/2.5cm, 1cm:
Since the speaker is just driven with 1V, any increase in distortion at closer observation distance must be coming from the microphone:
So the Earthworks M23 should not be used in the nearfield to measure distortion- it contributes significant distortion of it's own.
Let bring back the omnidirectional dynamic microphone, and see what happens when it's placed in the nearfield to measure the bass response:
As we can see, it observes lower H2 than the small electret condenser microphone like the Earthworks (and Sonarworks Xref20/Umik-1)
Let's proceed with our harmonic measurements:
A comparison point:
Reference:
https://www.erinsaudiocorner.com/loudspeakers/kef_blade2_meta/
BONUS data:
Reference:
https://ptt.purifi-audio.com/document/share/10/7958f847-3d8c-4fc5-8a68-cb8298539ef4
Figure 6- Frequency Response @1m, 94dB- Fundamental, 2nd and 3rd harmonics:
@lrisbo
The 10" PTT is better than the 12" XXLS.👍
As we can see, it observes lower H2 than the small electret condenser microphone like the Earthworks (and Sonarworks Xref20/Umik-1)
Let's proceed with our harmonic measurements:
A comparison point:
Reference:
https://www.erinsaudiocorner.com/loudspeakers/kef_blade2_meta/
BONUS data:
Reference:
https://ptt.purifi-audio.com/document/share/10/7958f847-3d8c-4fc5-8a68-cb8298539ef4
Figure 6- Frequency Response @1m, 94dB- Fundamental, 2nd and 3rd harmonics:
@lrisbo
The 10" PTT is better than the 12" XXLS.👍
Last edited:
Did you repeat the different distance measurements with the dynamic mic? It would be a GREAT help in comparing graphs when you just repeat EXACTLY the same procedure and scaling of the graphs.
There are still a few strange effects going on. The dynamic mic records more THD at 20Hz as the Earthworks in close distance. My Earthworks M50 has about -70dB THD at 250Hz and 110dBSpl. It would not have any influence to your measurements! Not sure what causes these results - don't be to quick that it is from the microphone!
Yessir of course I did measure at different distances with the dynamic mic:
I agree that it can appear that the dynamic mic has slightly more H2 at 20Hz @1cm (-42dB) than the Earthworks M23 @1cm (-43dB). I would not take it as significant- a repeat test at quieter conditions show slightly different tests. I would say within the margin of error.
One of the things we haven't spoken about so far is how to take optimal measurements. And certainly, with a log sine sweep in REW, there's a balance of getting precise measurements that won't take too long (seconds, not minutes for a sweep for the DUT)
I am using 256K samples x 8 repetitions with a 96KHz sample rate- this takes 21.8 seconds)
My main difficulty is trying to ensure sure all the colors are consistent, the legend is actually helpful and not misleading (legend refers to an on-screen cursor- that may NOT be visible during a screenshot) and graph sizes are the same, as to not unintentionally mislead the reader. My laptop for writing not the same as the one for data acquisition. John Mulcahy called me out on this- "Why do you have such large tickboxes??!"
I don't know!? 🤷♂️
Perhaps your Earthworks M50 has a better preamp than the Earthworks M23- I certainly couldn't measure -70 H2 @ 110dB
I agree that it can appear that the dynamic mic has slightly more H2 at 20Hz @1cm (-42dB) than the Earthworks M23 @1cm (-43dB). I would not take it as significant- a repeat test at quieter conditions show slightly different tests. I would say within the margin of error.
One of the things we haven't spoken about so far is how to take optimal measurements. And certainly, with a log sine sweep in REW, there's a balance of getting precise measurements that won't take too long (seconds, not minutes for a sweep for the DUT)
I am using 256K samples x 8 repetitions with a 96KHz sample rate- this takes 21.8 seconds)
My main difficulty is trying to ensure sure all the colors are consistent, the legend is actually helpful and not misleading (legend refers to an on-screen cursor- that may NOT be visible during a screenshot) and graph sizes are the same, as to not unintentionally mislead the reader. My laptop for writing not the same as the one for data acquisition. John Mulcahy called me out on this- "Why do you have such large tickboxes??!"
I don't know!? 🤷♂️
Perhaps your Earthworks M50 has a better preamp than the Earthworks M23- I certainly couldn't measure -70 H2 @ 110dB
Last edited:
Glad you asked.
In short, we can measure at various distances without having a gain or loss in the bass response.
Omnidirectional microphones do not exhibit proximity. This is because the sound has only admittance to the front of the diaphragm. A pressure microphone is basically a diaphragm in front of a closed chamber.
Proximity effect:
Reference:
https://www.dpamicrophones.com/mic-university/source-dependent-proximity-effect-in-microphones
In short, we can measure at various distances without having a gain or loss in the bass response.
Omnidirectional microphones do not exhibit proximity. This is because the sound has only admittance to the front of the diaphragm. A pressure microphone is basically a diaphragm in front of a closed chamber.
Proximity effect:
Reference:
https://www.dpamicrophones.com/mic-university/source-dependent-proximity-effect-in-microphones
Last edited:
Well that's interesting, because I have the Line Audio Omni1 as well as the CM4 (cardioid mic).
For all practical measurements at around 1-2 meters, there is absolutely NO significant difference between the two.
Same goes for off-axis response.
Obviously that can only be done with a time window.
With enough damping material on the floor and ceiling, I can get down to about 100-150Hz.
So this comparison only counts up from there.
The response of the CM4 below 100Hz is not specified very well, so it's hard to do any comparisons below 100Hz.
For all practical measurements at around 1-2 meters, there is absolutely NO significant difference between the two.
Same goes for off-axis response.
Obviously that can only be done with a time window.
With enough damping material on the floor and ceiling, I can get down to about 100-150Hz.
So this comparison only counts up from there.
The response of the CM4 below 100Hz is not specified very well, so it's hard to do any comparisons below 100Hz.
Btw, this also assumes just a traditional single capsule cardioid.
There are other ways to make a cardioid pattern as well that don't have this proximity issue.
I will do some additional (semi) near field test to show that proximity problem for lower frequencies.
There are other ways to make a cardioid pattern as well that don't have this proximity issue.
I will do some additional (semi) near field test to show that proximity problem for lower frequencies.
The proximity effect is pretty well known by audio engineers, nothing new there. In my AV production days I often gave the ‘reporter’ a dynamic omni. Never an empty battery and no problems when he or she ate the mike.