Since there has been a lot of focus on driver distortion lately with Purify entering the scene, I thought I would post a question on distortion measurements.
I like running multitone measurements on amps to get a 'general overview', so I thought why not do it on speakers? I think it gives a good visual overall indication of the 'signal to distortion' vs frequency. I find two tone tests a bit hard to interpret, and it seems the tones used can influence the result if they happen to hit 'problem areas' of the driver or not. More tones, more excitement
'More is more' like Yngwie Malmsteen claimed (guitar nerd joke)!
(Edit:
)
Is there a reason for not doing it?
I ran some quick measurements where I took seven averages/measurement because of a noisy garage environment. One is a PA 8" woofer, the other is an old Mission speaker with a 8" woofer & 1" tweeter. I have no SPL reference, just my ears, and the level in this graph was what I would call normal listening level, so this is not really a scientific example, just a general question about the concept. I'm hoping Lars or other smart people would comment on using multitone for driver testing, and I would also be curious to see what a similar measurement on a purify driver looks like.
I like running multitone measurements on amps to get a 'general overview', so I thought why not do it on speakers? I think it gives a good visual overall indication of the 'signal to distortion' vs frequency. I find two tone tests a bit hard to interpret, and it seems the tones used can influence the result if they happen to hit 'problem areas' of the driver or not. More tones, more excitement
'More is more' like Yngwie Malmsteen claimed (guitar nerd joke)! (Edit:
Is there a reason for not doing it?
I ran some quick measurements where I took seven averages/measurement because of a noisy garage environment. One is a PA 8" woofer, the other is an old Mission speaker with a 8" woofer & 1" tweeter. I have no SPL reference, just my ears, and the level in this graph was what I would call normal listening level, so this is not really a scientific example, just a general question about the concept. I'm hoping Lars or other smart people would comment on using multitone for driver testing, and I would also be curious to see what a similar measurement on a purify driver looks like.
Attachments
Last edited:
I'm quoting Juhanzi from here: https://www.diyaudio.com/community/threads/full-size-3-way-project.389962/post-7163483
Thought it was better to continue the measurement discussion here.
I have given it a little thought about room artefacts. Looking at the graph I posted first in this thread, the Mission speaker was close to the floor (maybe 60cm) and the 18Sound was up on a stand at maybe 160cm. So, the Mission likely had more reinforcement in the bass than the 18sound. My guess here is that will improve the signal to distortion. My logic is that the reflected overtones (distortion) are likely more damped than the lower tones. I'm guessing the same would apply if there are a lot of reflections in the higher registers too. So, my guess it that one should measure close to the driver to get more of the direct sound. When it comes to bass I guess it is what it is.. distortion is said to be less important in bass anyway, based on what we can hear (and not). However, making the drivers play bass will most likely influence IMD higher up too.
In general, I'm just curious to see some measurements around normal listening levels, No need to crank it up so it hurts your ears
Use your imagination, your 'guesstimations' are probably better than mine anyway
Hoping more people will chime in, maybe even do some measurements, and maybe we can get a general feeling for if this is a useful measurement or not. I'm hoping mr driver distortion himself @lrisbo will chime in here too
ARTA and REW can do multitone in their signal generators. Use averaging is my advice (until I'm told differently), I did seven averages/measurement.
From what I could find on the internet, it seems mics can have pretty high distortion themselves (especially at high SPL), so maybe it's a good idea to mention what microphone is used? I used an expensive B&K (together with a dedicated pre-amp) I borrowed from work, so I'm guessing it should be a good performer.
ARTA and REW can do multitone in their signal generators. Use averaging is my advice (until I'm told differently), I did seven averages/measurement.
From what I could find on the internet, it seems mics can have pretty high distortion themselves (especially at high SPL), so maybe it's a good idea to mention what microphone is used? I used an expensive B&K (together with a dedicated pre-amp) I borrowed from work, so I'm guessing it should be a good performer.
I would like a software that can play both a single tone (to set excursion), while also doing a sweep and plotting IMD. Alternatively, a qualitative version would be to use a single tone and play musical passages while listening/recording the result and judging when distortion becomes audible.
Sound like an good idea! I'm just guessing doing averages will not work unless it's a steady state measurement?
There might be some difficulty in choosing the frequency for the single tone? If it was at FS (or close) current would be low, would that effect the whole test? Should be different depending on the intended use for the driver?
The listening test would probably have to be done on a single driver, since woofer excursion on a 3-way will probably not change the sound much when listening to music? On the other hand, listening to a dedicated midrange fed with a low frequency tone, might not be so relevant either?
I'm thinking the same with using multitone, it would best represent the actual use-case if the driver is tested using/simulating the intended XO. Looking at the Mission measurement in my first post (yellow trace), I'm amazed that the cheap tweeters I put in those is showing distortion around -60dB!
There might be some difficulty in choosing the frequency for the single tone? If it was at FS (or close) current would be low, would that effect the whole test? Should be different depending on the intended use for the driver?
The listening test would probably have to be done on a single driver, since woofer excursion on a 3-way will probably not change the sound much when listening to music? On the other hand, listening to a dedicated midrange fed with a low frequency tone, might not be so relevant either?
I'm thinking the same with using multitone, it would best represent the actual use-case if the driver is tested using/simulating the intended XO. Looking at the Mission measurement in my first post (yellow trace), I'm amazed that the cheap tweeters I put in those is showing distortion around -60dB!
Last edited:
German online HiFi Magazine "LowBeats" has this adopted in their standard measurements:
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
Random examples:
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
(Nice, but would be even nicer if they also publish simple FR measurements. But they won't blame their advertising customers too much, I think...)
You need a good Mic for that, e.g. I have a (calibrated) one (Monacor ECM-80) that adds much own k2 distortion so that multitone measurements are crap....
Best regards
Peter
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
Random examples:
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
https://www-lowbeats-de.translate.g..._sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp
(Nice, but would be even nicer if they also publish simple FR measurements. But they won't blame their advertising customers too much, I think...)
You need a good Mic for that, e.g. I have a (calibrated) one (Monacor ECM-80) that adds much own k2 distortion so that multitone measurements are crap....
Best regards
Peter
Great link!
Judging from the variation in the peaks amplitude, i assume there are reflections from the room etc making it more uneven. It also seems they have chosen 94dB SPL as the level where they measure? That is a bit tricky IMO, when I look at the individual peaks they are around 80dB, but the sum of them all apparently makes 94dB SPL (weighting?)
Not sure if distortion should be compared to the 80dB peaks, or the total 94dB? I find it easier to just compare it to the peaks and how far the 'grass' is below them.
They seem to cut the graph around -45dB under the peaks, so I guess everything below that is considered irrelevant?
It seems like a valid measurement method though.
On mic distortion: I saw somewhere on the web where they used two drivers playing one tone each (different frequencies), to determine the distortion from the mic by looking at the IMD products the mic was creating. Interesting method, but I don't remember the details.
Judging from the variation in the peaks amplitude, i assume there are reflections from the room etc making it more uneven. It also seems they have chosen 94dB SPL as the level where they measure? That is a bit tricky IMO, when I look at the individual peaks they are around 80dB, but the sum of them all apparently makes 94dB SPL (weighting?)
Not sure if distortion should be compared to the 80dB peaks, or the total 94dB? I find it easier to just compare it to the peaks and how far the 'grass' is below them.
They seem to cut the graph around -45dB under the peaks, so I guess everything below that is considered irrelevant?
It seems like a valid measurement method though.
On mic distortion: I saw somewhere on the web where they used two drivers playing one tone each (different frequencies), to determine the distortion from the mic by looking at the IMD products the mic was creating. Interesting method, but I don't remember the details.
Like this?
http://aplaudio.com/conc2/products/tda-im
He quoted me a price of (I think) 325 euros. If it was under a 100 US I would have purchased it.
-I've looked through the listing for low noise floor mic.s at a reasonable price and the Rode NT1-A seems to be "it" (but it needs to be "calibrated" (corrected calibration file) for a linear response on the axis you want to measure with it).
I see you have done a couple of videos on distortion measurements, also on measuring mic distortion as I mentioned before, so a lot of interesting stuff on the topic! Recommended for others viewing this thread! https://www.youtube.com/c/JosephCrowesDIYSpeakerBuilding
When you say noise floor, I guess it's mainly 'distortion floor' you're referring to?
Maybe one 'flat' higher distortion mic could be used for FR measurements, and another for distortion.. some wiggles in mic FR should not matter much when testing distortion.
Rode NT1-A? Some clones exist too I think? Price of the original seems to be around 200Eur, so maybe not too bad considering what his hobby can cost.. Have you tested that one yourself too?
When you say noise floor, I guess it's mainly 'distortion floor' you're referring to?
Maybe one 'flat' higher distortion mic could be used for FR measurements, and another for distortion.. some wiggles in mic FR should not matter much when testing distortion.
Rode NT1-A? Some clones exist too I think? Price of the original seems to be around 200Eur, so maybe not too bad considering what his hobby can cost.. Have you tested that one yourself too?
I'm not Joseph Crowe; he actually posts here from time-to-time. I quite like his videos.
"noise floor" listed as *"self noise" on spec. sheets. For the NT1-A self noise is only 4.5db. My UMIK-1 is about 30db (though it also includes the ADC). My Line Audio OM1 is about 18db.
*https://www.dpamicrophones.com/mic-university/the-basics-about-noise-in-mics
I've not done any distortion testing of mic.s myself (..yet). (..I've had to many other projects dumped on me of late, a multitude of which are ahead of the line of "things to do".)
Note: in addition to spl-reading with the UMIK-1 I also purchased it from Cross Spectrum to effectively "calibrate" other microphones (the OM1, and quite likely the NT1-A at some point) to achieve a closer result to flat (and vocal microphones like it often have some loss in pressure at lower freq.s along with irregular responses at higher frequencies).
"noise floor" listed as *"self noise" on spec. sheets. For the NT1-A self noise is only 4.5db. My UMIK-1 is about 30db (though it also includes the ADC). My Line Audio OM1 is about 18db.
*https://www.dpamicrophones.com/mic-university/the-basics-about-noise-in-mics
I've not done any distortion testing of mic.s myself (..yet). (..I've had to many other projects dumped on me of late, a multitude of which are ahead of the line of "things to do".)
Note: in addition to spl-reading with the UMIK-1 I also purchased it from Cross Spectrum to effectively "calibrate" other microphones (the OM1, and quite likely the NT1-A at some point) to achieve a closer result to flat (and vocal microphones like it often have some loss in pressure at lower freq.s along with irregular responses at higher frequencies).
Sorry for the confusion, people (like me) don't always have usernames corresponding to their real name
As I understand it, self noise might be low, but max SPL before distortion hits might also be low, so there is not more dynamic range even if noise is lower. I guess it depends on what levels should be measured, in what range the mic is 'at it's best'? Also, a mic with low noise and high dynamic range might still not have low distortion? Is the Rode actually a low distortion mic, or just low noise?
JC seems to like the SM58 for it's low distortion. There seems to be many versions of that mic Sm58, Sm58S, Sm58LC.. some are claimed to have 'brigtened mids & bass' etc https://www.thomann.de/se/search_dir.html?sw=sm58&smcs=8af55d_3372
Seems they have some built-in filters to reduce wind noises etc. Not sure if those are electrical or 'mechanical screens' thogh.
As I understand it, self noise might be low, but max SPL before distortion hits might also be low, so there is not more dynamic range even if noise is lower. I guess it depends on what levels should be measured, in what range the mic is 'at it's best'? Also, a mic with low noise and high dynamic range might still not have low distortion? Is the Rode actually a low distortion mic, or just low noise?
JC seems to like the SM58 for it's low distortion. There seems to be many versions of that mic Sm58, Sm58S, Sm58LC.. some are claimed to have 'brigtened mids & bass' etc https://www.thomann.de/se/search_dir.html?sw=sm58&smcs=8af55d_3372
Seems they have some built-in filters to reduce wind noises etc. Not sure if those are electrical or 'mechanical screens' thogh.
Here is a company that sells some software for measuring mic distortion. I wonder if there is some free software for this..?
https://www.listeninc.com/products/...ermodulation-distortion-measurement/#tab-id-3
https://www.listeninc.com/products/...ermodulation-distortion-measurement/#tab-id-3