A good measurement omnidirectional mic has flatter frequency response through a much wider range of angles than directional mics. This is important because in many instances the sound being measured doesn't originate from a single point source, but rather multiple sources, including reflections. An example: assessment of environmental noise in a work environment like an office, factory, etc or capturing wildlife sounds like bird songs or animal calls.
If you're only using the mic on-axis, as with most loudspeaker measurements, a non-omni (typically, cardioid) mic that has flat FR on axis and low enough self-noise would probably work fine.
An Aside: Although 1/2" and smaller diameter capsules are most widely used in test mics, a 1" capsule usually has lower self-noise due to the bigger surface area. The same reason that a larger diameter woofer usually has lower distortion than a smaller one.
For a long time my reference measurement mic was a $2000 1" ACO Pacific + 200vdc power supply with literally straight-line FR from 5Hz to 20kHz and self noise of 5~7 dBA -- quieter than the ambient of most anechoic chambers. Yet it could handle 150 dB peaks, which is amazing dynamic range. Hard to find a mic preamp or digital interface to match the low noise level. It's compatible with Bruel & Kjaer mics of the same size -- interchangeable capsules.
Listening to just about any sounds recorded through it with quality gear, in mono, through Sennheiser HD650 headphones, was usually a revelation: I heard more than I did live! Although this was partly because I could listen repeatedly. There was absolutely no sense of detail loss compared to the live sound.
Sold the mic a couple years back because it was overkill for my level of diy audio, but I still miss it sometimes.
@tktran303 -- some fascinating detailed work in your posts here, but the intent of your last few posts isn't entirely clear to me.
The first part is not true.
In fact, most cardioid mics have a very flat and even frequency response.
The biggest issue with those is the proximity effect.
For measuring loudspeakers it would actually be best to have NO additional angles except for line-of-sight.
Since all other information and data is not relevant, in fact will give us those pesky reflections that we don't want.
D
Deleted member 375592
A few more steps, and you'll acknowledge that there could not be a mic that is omni, flat, low distortions, and low noise... but you don't need one - you can have 2 (or more) each excelling in some areas. Then you cross-calibrate them.., and Bob is your uncle.
🙂
In the theory of system identification and Kalman filtering and practice of "rocket science", radar, sonar, etc, that's called the Multiple Model approach (see IMM, AFMM, etc).
🙂
In the theory of system identification and Kalman filtering and practice of "rocket science", radar, sonar, etc, that's called the Multiple Model approach (see IMM, AFMM, etc).
Actually, I think it is true that the best measurement mics have the flattest FR. The ACO mic came with a individually-tested printout of FR that looked ruler straight to ~18 kHz (ie, ~0.1 dB variation), with a minor bump/dip less than 0.4 dB in level above. This response was confirmed by 3rd party testers and myself personally in the U of BC anechoic chamber. I have not seen a flatter FR in virtually any other transducer.
Suppose we have a Cosmos APU + ADC capable of THD+N of -130 dB, or an Audio Precision APx555B, with a THD+N of -120 dB.
If we use microphone of 20KHz bandwidth and distortion of -80dB, then the limit will lie with the microphone.
The reason is because the signal, from the computer, travels through the digital analog convertor of the audio interface, to amplifier, to speaker (Device Under Test), to microphone and it's pre-amp, and back into analog to digital converter of the audio interface. It's a measurement of the complete chain.
Now suppose we want to measure a driver with a true H2 of, eg. -70dB. What equipment do we need to do that?
Let
Device 1 be an audio interface with -130dB THD+N.
Device 2 be an amplifier with -120dB THD+N
Device 3 be a microphone, with unknown THD+N
And Device 4 be the speaker (DUT)
Does one need a microphone (Device 3) to have be? -90dB? -80dB. Or will -70dB be adequate?
Let's play with some numbers:
View attachment 1355775
With a mic of -90dB THD+N, we can measure ~70dB, with an error margin of <0.1dB. With a mic of -80dB THD+N, the error margin is <0.5dB
But if the microphone has the same distortion level as the DUT, then the error is 3dB.
Reference:
https://sengpielaudio.com/calculator-thd.htm
Suppose I don't have the Cosmos APU or Audio Precision. I have a regular soundcard + amplifier that does this:
Here it is:
Here is the PTT8.0XNAA,
As we can see the SPL is around 89dB to 91dB between 100Hz to 2KHz, with H2 trending around 25dB and 10dB in that range.
This makes H2 -64dB to -80dB relative to the fundamental
Here is the measurement by @HiFiCompass, Measured at 31.5cm- and H2 is -65dB and -75dB down.
Below 200Hz the blue H2 line looks a bit ragged. Let's take a look at the nearfield measurement with the microphone just 2cm from the cone. This rejects a lot of room reflections and gets a clearer picture of what's happening in the bass:
Between 100Hz and 2KHz, H2 (blue line) is between -63 and -80dB down.
Finally, let's take a look, as measured by the Klippel Near Field Scanner.
Between 100Hz and 2KHz, H2 seems to be about -52 to -68dB relative to the fundamental
The Klippel NFS measurement is some 10dB worse
(the plot thickens)
If we use microphone of 20KHz bandwidth and distortion of -80dB, then the limit will lie with the microphone.
The reason is because the signal, from the computer, travels through the digital analog convertor of the audio interface, to amplifier, to speaker (Device Under Test), to microphone and it's pre-amp, and back into analog to digital converter of the audio interface. It's a measurement of the complete chain.
Now suppose we want to measure a driver with a true H2 of, eg. -70dB. What equipment do we need to do that?
Let
Device 1 be an audio interface with -130dB THD+N.
Device 2 be an amplifier with -120dB THD+N
Device 3 be a microphone, with unknown THD+N
And Device 4 be the speaker (DUT)
Does one need a microphone (Device 3) to have be? -90dB? -80dB. Or will -70dB be adequate?
Let's play with some numbers:
With a mic of -90dB THD+N, we can measure ~70dB, with an error margin of <0.1dB. With a mic of -80dB THD+N, the error margin is <0.5dB
But if the microphone has the same distortion level as the DUT, then the error is 3dB.
Reference:
https://sengpielaudio.com/calculator-thd.htm
Suppose I don't have the Cosmos APU or Audio Precision. I have a regular soundcard + amplifier that does this:
Here it is:
Solid line H2, dotted line H3, dashed line- noise
What do I need?
-69.546 is within 0.5dB of -70dB
What do I need?
-69.546 is within 0.5dB of -70dB
Here is the PTT8.0XNAA,
As we can see the SPL is around 89dB to 91dB between 100Hz to 2KHz, with H2 trending around 25dB and 10dB in that range.
This makes H2 -64dB to -80dB relative to the fundamental
Reference:
Purifi PTT8.0X04-NA datasheet
Purifi PTT8.0X04-NA datasheet
Here is the measurement by @HiFiCompass, Measured at 31.5cm- and H2 is -65dB and -75dB down.
Below 200Hz the blue H2 line looks a bit ragged. Let's take a look at the nearfield measurement with the microphone just 2cm from the cone. This rejects a lot of room reflections and gets a clearer picture of what's happening in the bass:
Between 100Hz and 2KHz, H2 (blue line) is between -63 and -80dB down.
Finally, let's take a look, as measured by the Klippel Near Field Scanner.
Between 100Hz and 2KHz, H2 seems to be about -52 to -68dB relative to the fundamental
The Klippel NFS measurement is some 10dB worse
(the plot thickens)
The microphone provided by Klippel for the Near Field Scanner is the Microtech Gefell MK255 (+ MV 210 preamplifier). This is a pre-polarized 1/2" capsule of the 50mV/Pa class, capable of 3Hz to 20KHz, +/- 2dB, and is attaches to a preamplifier with a IEPE (aka CCLD, constant current power) power supply.
Klippel specify it's Max. SPL before clipping as 135dB.
This is the 1% clipping point, as evidenced by the manufacturer's datasheet: MM215 i.e. MK 255 capsule on MV 210 preamplifier
If the 135dB is the 1% (-40dB) distortion point of the microphone, then it follows that
The -80dB distortion point of the microphone is ~40dB lower. i.e. 135-40 = 95dB.
The -70dB distortion point of the microphone is ~30dB lower i.e. 135-30 = 105dB (refer to post 180)
In other words, the observation limit for the Microtech Gefell MK255 + MV 210 is 95dB to 105dB, if one wants to characterize the H2 of a driver is between -70dB to -60dB (error margine <~0.5dB)
So why isn't the NFS doing this?
TBC...
Klippel specify it's Max. SPL before clipping as 135dB.
This is the 1% clipping point, as evidenced by the manufacturer's datasheet: MM215 i.e. MK 255 capsule on MV 210 preamplifier
If the 135dB is the 1% (-40dB) distortion point of the microphone, then it follows that
The -80dB distortion point of the microphone is ~40dB lower. i.e. 135-40 = 95dB.
The -70dB distortion point of the microphone is ~30dB lower i.e. 135-30 = 105dB (refer to post 180)
In other words, the observation limit for the Microtech Gefell MK255 + MV 210 is 95dB to 105dB, if one wants to characterize the H2 of a driver is between -70dB to -60dB (error margine <~0.5dB)
So why isn't the NFS doing this?
TBC...
Last edited:
Stop asking the right and obvious questions 😀 😀 😀
(no actually don't stop 🙂 👍 )
Well, technically we can calibrate the entire chain out of the equation, even in sense of distortion.
In theory this is even possible with microphones, in practice not that easy unfortunately.
An image from post #187 is missing.
Here it is again.
It shows that a microphone (Device 3) with self distortion of 0.01% (-80dB), when measuring a speaker with harmonic distortion -70dB (Device 4) is close be being transparent- the resulting is ~ -69.586
Let's move forward with actual data, for which I have the 8 ohm version of the PTT8.0X driver. Here is an image taken from page 1 PTT8.0X08-NAB-01 datasheet:
author's own measurements taken with B&K mic 31.6cm from cabinet:
Compare with Klippel Near Field Scanner measurement for the 4 ohm driver, measured at 96 dB:
Coming up:
Other mics at 31.6cm
Near field testing for 60Hz to 600Hz
Here it is again.
It shows that a microphone (Device 3) with self distortion of 0.01% (-80dB), when measuring a speaker with harmonic distortion -70dB (Device 4) is close be being transparent- the resulting is ~ -69.586
Let's move forward with actual data, for which I have the 8 ohm version of the PTT8.0X driver. Here is an image taken from page 1 PTT8.0X08-NAB-01 datasheet:
Here are the harmonics transposed to dB relative to fundamental, made possible with VituixCAD -> Tools -> SPL Trace, and imported and displayed in REW.
NB. Midrange harmonics- from 400Hz to 2KHz is below -70dB for H2, and from 100Hz to 1.4KHz, ~-75dB for H3
author's own measurements taken with B&K mic 31.6cm from cabinet:
Compare with Klippel Near Field Scanner measurement for the 4 ohm driver, measured at 96 dB:
Coming up:
Other mics at 31.6cm
Near field testing for 60Hz to 600Hz
Last edited:
How high can you cross this midwoofer with a cone diameter of ~17cm (surround crest to crest). And what about the harmonics above 2KHz?
Recall Purifi's datasheet, measured @40cm with 2.83V drive, in their hemi-anechoic chamber:
SPL response shown for 1m:
Here I’m using REW's built in EQ function. A few PEQs to flatten the on-axis response, for baffle step compensation, and inserted a a couple of low pass filters to enable an acoustic LR4 crossover at 2KHz:
Then we re-measure with all this filter in place. No need for external DSP box, Equaliser APO etc. Measured directly with Filters in place.
Here's the fundamental, and harmonics:
Do we really need the low distortion filter using passive parts ?
Recall Purifi's datasheet, measured @40cm with 2.83V drive, in their hemi-anechoic chamber:
SPL response shown for 1m:
Here I’m using REW's built in EQ function. A few PEQs to flatten the on-axis response, for baffle step compensation, and inserted a a couple of low pass filters to enable an acoustic LR4 crossover at 2KHz:
Then we re-measure with all this filter in place. No need for external DSP box, Equaliser APO etc. Measured directly with Filters in place.
Here's the fundamental, and harmonics:
Do we really need the low distortion filter using passive parts ?
Last edited:
The PTT8.0X has the lowest iron distortion anong all our woofers so you gain very little with the passive notch trick. you can just see the subharmonic HD3 peaks just above and below 2 kHz in the data sheet graph and they are pobobaky in the noise floor of your plot 😎
In general that depends on what kind of distortion criteria you use and what you THINK is still audible or not.
Some people seem to have an insanely low acoustic noise floor in their room and seem to also found a way to cope with the psychological trouble such low noise floor will create.
Lars,
It appears that I (and Wolfgang, Vance) need to find a new harmonic measurement methodology that allows for higher than 80dB range.
It appears that I (and Wolfgang, Vance) need to find a new harmonic measurement methodology that allows for higher than 80dB range.
Last edited:
I've read through this huge thread twice but have problems digesting it.
My background is measuring speakers (and mikes) using 1/2" B&K 200V mikes with 1902 & 2010 gear but I can no longer get the steam to power these today. My experience is that B&K and GRAS stuff was a better than ACO in da previous Millenium.
tktrans303's measurements with his 1/2" B&K bring back fond memories of the times when these things were important to me. His curves show what I consider to be 'true' distortion measurements as well as illustrate the times when noise is the limiting factor.
But has anyone answered his original question? What is the best cheap measurement mike for distortion?
BTW, a given noise level gives you a minimum theoretical possible time to do a measurement (both response & distortion) and achieve a given accuracy. There's several B&K papers on this subject
Angelo Farina's method is the only one that does this properly. All other methods, TDS, MLS, stepped sine bla bla will either take more time or be less accurate in that time. If you can't do zillion point FFTs, you can average several of his log sweeps for similar efficiency. The limit for averaging is set by the reverb time of your 'room' but is usually not a problem with the measurements we DIYs (and professionals) do.
I came up with his method in the 90s when I was looking at how to measure a speaker's response & distortion in the shortest possible time. But then, the computing power and more particularly good CODECs were too $$$ for the many units I wanted for factory test. Angelo was amazed when I showed my Jurassic code to him this Century.
My background is measuring speakers (and mikes) using 1/2" B&K 200V mikes with 1902 & 2010 gear but I can no longer get the steam to power these today. My experience is that B&K and GRAS stuff was a better than ACO in da previous Millenium.
tktrans303's measurements with his 1/2" B&K bring back fond memories of the times when these things were important to me. His curves show what I consider to be 'true' distortion measurements as well as illustrate the times when noise is the limiting factor.
But has anyone answered his original question? What is the best cheap measurement mike for distortion?
BTW, a given noise level gives you a minimum theoretical possible time to do a measurement (both response & distortion) and achieve a given accuracy. There's several B&K papers on this subject
Angelo Farina's method is the only one that does this properly. All other methods, TDS, MLS, stepped sine bla bla will either take more time or be less accurate in that time. If you can't do zillion point FFTs, you can average several of his log sweeps for similar efficiency. The limit for averaging is set by the reverb time of your 'room' but is usually not a problem with the measurements we DIYs (and professionals) do.
I came up with his method in the 90s when I was looking at how to measure a speaker's response & distortion in the shortest possible time. But then, the computing power and more particularly good CODECs were too $$$ for the many units I wanted for factory test. Angelo was amazed when I showed my Jurassic code to him this Century.
I’m still working on it!
I hope to compare mics, but also methods and processes, such that one doesn’t have purchase a laboratory grade/accident prone 4191+2669
So far one leading contender is the Line Ausio Omni1, which is a suggested microphone in REW, by @JohnPM and the FSAF method by @mikets42
where you can listen to the distortion aka recording residual (via ? headphones)
https://www.roomeqwizard.com/betahelp/help_en-GB/html/fsafmeasurement.html#top
More to come…
Last edited:
You just described how 99.8% of all threads go and why they are hard to follow 😄🤣🤣
Post in diyAudio are fixed after 30 minutes, which I don't like so much, in case I present erroneous data.
On the other hand, original post CAN be updated and modified, as time goes on.
But the search continues...
On the other hand, original post CAN be updated and modified, as time goes on.
But the search continues...
Wanna tell us what these are?
There are immutable Laws of Nature that relate mike directivity (however it is achieved) to proximity. You may know them as the Wave Equation.
There are directional mikes that are sorta cardioid and have MORE proximity than 'simple' cardioids but none with less.