That still depends how you want to look at distortion.
This can be on an absolute way or relative way.
Absolute = having real accurate and absolute figure to the measurements
Relative = numbers don't resemble actual values but give enough indication when problems occur and what to use for crossover filtering.
On a relative way the cheapest microphones are capable enough for this.
In combination with many 3rd party measurements these days some people can definitely get away with a cheaper microphone as long as it's calibrated well.
If you really want to compare different drivers at especially higher volumes, this won't cut the mustard.
To measure down below 0.01% (-80dB below fundamental) i have to blast the dynamic omnidirectional with above 110dB SPL because of the noisefloor of the mic preamp in the interface. That is easily acomplished by putting it (5cm) near the DUT. This also solves the problem with noise and reflections from the room.
There is no way at all , be it absolute or relative, to measure distortion
if the microphone (or the noise btw.) dominates the nonlinear harmonics !
Last edited:
@Hörnli
this approach is sure to work, but which dynamic Omni directional microphone is flat from 20 to 20 K?
Or do you ignore absolute frequency response accuracy?
Or do you have two microphones for different tasks?
Edit: I read some of the earlier comments in the thread, and I missed that you use different microphones for different tasks.
I thought this would be necessary and probably are more affordable solution than a 1/2” electret condenser microphone designed for laboratory/scientific purposes, a niche market and thus costs an order of magnitude more than “prosumer” device costing multi hundred dollars.
I too, like @mainframe99 would like to find one single microphone solution that can do accurate SPL measurements for crossover design, but also distortion down -80dB (for unpublished/unattainable or prototype drivers)
this approach is sure to work, but which dynamic Omni directional microphone is flat from 20 to 20 K?
Or do you ignore absolute frequency response accuracy?
Or do you have two microphones for different tasks?
Edit: I read some of the earlier comments in the thread, and I missed that you use different microphones for different tasks.
I thought this would be necessary and probably are more affordable solution than a 1/2” electret condenser microphone designed for laboratory/scientific purposes, a niche market and thus costs an order of magnitude more than “prosumer” device costing multi hundred dollars.
I too, like @mainframe99 would like to find one single microphone solution that can do accurate SPL measurements for crossover design, but also distortion down -80dB (for unpublished/unattainable or prototype drivers)
Last edited:
That is pretty obvious, but what is your point?
A simple cheap Dayton mic can easily follow such distortion numbers that are more- or less in line with 3rd party measurements that use much higher quality microphones at regular sound pressure levels.
I have used both types (cheap en proper quality ones) and for the purpose of designing crossovers for the vast majority of speakers a cheap one is adequate especially when this is being used in additional with 3rd party data.
Knowing in absolute numbers how low the lowest distortion is, is not really relevant for making a crossover filter for an average speaker.
It's maybe interesting for a couple of super nerds like us to be used for further research for very specific reasons.
The frequency response of my dynamic omni directional microphone is just about linear enough to make useful distortion measurements. Apart from that, measuring a relatively big cone 5cm away from the dustcap is not linear at all anyway.
Of course one may buy an externaly polarized microphone base unit and two capsules, one for low noise and one for high SPL. Something like that costs rock solid four figures.
Yes, there is no way around it. As of today both together cost 300€ list price. As most diyAudio members already own an elektret measurement microphone, i promote choosing (lending from a friend) a cheap dynamic microfone - to enter the world of no/reduced distortion loudspeakers.
Of course one may buy an externaly polarized microphone base unit and two capsules, one for low noise and one for high SPL. Something like that costs rock solid four figures.
That is my point ...and i am nerdy about loudspeaker distortion. According to my observation, this is the last parameter that can only be cracked in DIY.
That's all great, but let general nerd advice be there for nerds.
It's pointless to advise very specific needs for a general public.
Deep diving into the world of measurements is an interesting part of DIY audio. Some people have an interest in this, and that is great. Measuring acoustic harmonic distortion down to -80 dB is really challenging, and I am impressed by anyone who can do it.
For the practical application of designing and building a high quality loudspeaker for listening to music, speech, or video content, I judge that once harmonic distortion is below some level, it becomes insignificant. That does not mean that it will always be inaudible at that level, on all music, for all listeners. I am simply saying that at some minimal level, harmonic distortion becomes less important than all the other competing priorities that must be balanced when designing a high quality system.
For me, that minimal level is H2 at -40 dB (1%) and H3, H4, H5 at -50 dB (0.3%). So for me, once distortion is below these levels, I would not trade other aspects of performance to get a lower harmonic distortion.
So in my case, I need a distortion measurement capability that goes down to -50 dB.
j.
For the practical application of designing and building a high quality loudspeaker for listening to music, speech, or video content, I judge that once harmonic distortion is below some level, it becomes insignificant. That does not mean that it will always be inaudible at that level, on all music, for all listeners. I am simply saying that at some minimal level, harmonic distortion becomes less important than all the other competing priorities that must be balanced when designing a high quality system.
For me, that minimal level is H2 at -40 dB (1%) and H3, H4, H5 at -50 dB (0.3%). So for me, once distortion is below these levels, I would not trade other aspects of performance to get a lower harmonic distortion.
So in my case, I need a distortion measurement capability that goes down to -50 dB.
j.
With a dynamic low output mike at -say- 55dB? I always wondered how Joseph Crowe got 0,0056% with a Focusrite Scarlett Solo (while that was with a condenser mike). A typical highish end differential pre for microphones (PGA2500) attached... Of course, if you get the levels up high enough, you don't need 50dB gain.
Attachments
The thing is, you’re all right. Not alright. But all correct.
We're probably all sitting on the outside looking in at an object, seeing the same thing from different angles.
When I have to increase the signal to noise by moving my microphone closer, and then have to check my gain stages / input chain, and ensure that I have an extremely quiet environment to see if my distortion measurements are accurate, I have to wonder, was that even at all necessary? Sure, I could measure the bass distortion 10dB better than before, or midrange distortion 6dB better than before, eg. H2 -60 vs -55dB, but it sounded the same to me as when I thought it measured at -55dB. Besides, in an ordinary listening environment, there's bound to be more sources of noise interference than a permanently set-up lab.
But I do it for drivers for which I don't datasheets, or for prototype drivers, then of course it's useful to measure what is going on/what happened.
And what is the current state of the art?
How does a driver that has H2 of -70dB @2.83V input, and all others harmonics under the noise floor of even a lab room (~20dB(A)) "sound like"?
An interesting observation from @hifijim - how did you come to that conclusion?
I wonder if there have been any carefully controlled experiments between says, the SBPFC drivers (no demodulation device) vs SB15/17 standard range (copper sleeve) and Satori (copper sleeves). Now the problem is even if the on-axis is tightly matched to +/- 0.1dB, the polar responses/directivity is not.
It's not clear to me whether directivity trounces distortion, or the other way round.
We're probably all sitting on the outside looking in at an object, seeing the same thing from different angles.
When I have to increase the signal to noise by moving my microphone closer, and then have to check my gain stages / input chain, and ensure that I have an extremely quiet environment to see if my distortion measurements are accurate, I have to wonder, was that even at all necessary? Sure, I could measure the bass distortion 10dB better than before, or midrange distortion 6dB better than before, eg. H2 -60 vs -55dB, but it sounded the same to me as when I thought it measured at -55dB. Besides, in an ordinary listening environment, there's bound to be more sources of noise interference than a permanently set-up lab.
But I do it for drivers for which I don't datasheets, or for prototype drivers, then of course it's useful to measure what is going on/what happened.
And what is the current state of the art?
How does a driver that has H2 of -70dB @2.83V input, and all others harmonics under the noise floor of even a lab room (~20dB(A)) "sound like"?
An interesting observation from @hifijim - how did you come to that conclusion?
I wonder if there have been any carefully controlled experiments between says, the SBPFC drivers (no demodulation device) vs SB15/17 standard range (copper sleeve) and Satori (copper sleeves). Now the problem is even if the on-axis is tightly matched to +/- 0.1dB, the polar responses/directivity is not.
It's not clear to me whether directivity trounces distortion, or the other way round.
A little study I'm working on and trying to wrap up this week...
First thing I'll mention is...how much do you trust your mic and calibration file?
Audix TM1 is manufactured to be +/- 2db from 20Hz to 20Khz (not used below)
Beyerdynamic MM1 is manufactured to be +/-1.5db from 50Hz to 16Khz (not used below)
Earhworks M23R is manufactured to be +/-0.5db from 9Hz to 23Khz
Here is SPL response using (1) a $35 mic with a $47 preamp; (2) a $70 mic with a $200 preamp, and (3) $600 mic with a $170 preamp. With what should be good calibration files and "psychoacoustic smoothing" which is about 1/6 octave which is the level of "squiggliness" we can hear.
This is a $15 tweeter in a 3D printed waveguide
SPL ~1W/1M...
Here is HD3 (at 96db/1M)...
And the noise floor in my garage...
First thing I'll mention is...how much do you trust your mic and calibration file?
Audix TM1 is manufactured to be +/- 2db from 20Hz to 20Khz (not used below)
Beyerdynamic MM1 is manufactured to be +/-1.5db from 50Hz to 16Khz (not used below)
Earhworks M23R is manufactured to be +/-0.5db from 9Hz to 23Khz
Here is SPL response using (1) a $35 mic with a $47 preamp; (2) a $70 mic with a $200 preamp, and (3) $600 mic with a $170 preamp. With what should be good calibration files and "psychoacoustic smoothing" which is about 1/6 octave which is the level of "squiggliness" we can hear.
This is a $15 tweeter in a 3D printed waveguide
SPL ~1W/1M...
Here is HD3 (at 96db/1M)...
And the noise floor in my garage...
Just to clarify, this was not to imply these microphones are not good. I would trust the calibration files that these come with unless I saw objective evidence otherwise.
I would not trust the typical calibration file that comes with $100 or less mics. But with a good calibration file a $35 mic can work fine for crossover work. My fist mic was a UMIK-1 from Cross Spectrum Labs.
My passing thought is >
Any form of Electro-Dynamic Transducer is just not going to allow distortion figures in the order of 0.01%
Thanks for that - we are on the same page re crossover design. I was basically just annoyed at the variance between the mics I have here and wanted something from someone a bit more reputable i.e. going from "good enough" to "pretty good" , while perhaps not "state of the art". I'm well aware of the dual channel and polars etc. I wanted to step up my mic. I agree only a novice would create a crossover with on axis only. I have an old full 2-way design study on ASR.
You mention tuning the final result by ear. This is sort of where my problem was, I lost trust in my mics because the USB only mic sounded better to my ear and subsequently measured the highest raw tweeter output, therefore meaning it was boosted less or reduced more in the final design. It became a pain in the *** hauling crossovers back out of the cabinets to change the tweeter by ear after the tuning with the dual channel XLR mic setup, when I knew I had the actual crossovers and polars phase dialed in. The simulations are so close to the final result now with VCAD, all I am finding myself doing is adjusting the 4k ish-20+khz range, where these cheap XLR mics were letting me down.
All this discussion on distortion, I didn't really too much care about distortion measurement accuracy. Just SPL being correct. I know the mics aren't going to be great i.e under 0.1-1%, but they are still useful for showing glaring errors in the cabinet or driver faults.
I have ordered a earthworks M23R. Thanks to all for the recommendations. Seems like the proper step up from the entry level Umik (although handy asf due to USB), the Dayton and the sound-works. While still cheaper than some of the drivers I have here.
Some folks believe otherwise.My passing thought is >
Any form of Electro-Dynamic Transducer is just not going to allow distortion figures in the order of 0.01%