Doing these measurements in 4m distance ... I'm acutally pretty envious about their measurement room
I did a near field measurement on the tweeter in my latest build, and thought I would post it. It's a typhany 4ohm tweeter, and I compared it with and without a 8,2ohm resistor in series. I liked this combo in an earlier build, and maybe the low distortion has something to do with it. As before, I hve no SPL calibration, only 'by ear' to what would be a high listening level in my case.
Ah yeah I didnt understandit, compared the images to each other the ITU named image tone is 100Hz, 200Hz, 300Hz... so yeah harmonics covered.
Btw something like "the grass" above the bandwidth would happen in active multiway speaker with voltage amp and without series impedance Can you do similar image that has response with series resistor compared to loopback? even better with series coil if you happen to have one.
See how "naked" the distortion is past "pass band". If distortion is audible I suspect its this one, its not masked in a multiway speaker muchas it extend out of band and probably boosted and beamed by cone resonance. Tweeter would have putput tgere and mask some, but its at different physical location and not beaming. On active speakers that is.
the ITU named image tone is 100Hz, 200Hz, 300Hz... so yeah harmonics covered. Btw something like "the grass" above the bandwidth would happen in active multiway speaker with voltage amp and without series impedance
Can you do similar image that has response with series resistor compared to loopback? even better with series coil if you happen to have one. See how "naked" the distortion is past "pass band". If distortion is audible I suspect its this one, its not masked in a multiway speaker muchas it extend out of band and probably boosted and beamed by cone resonance. Tweeter would have putput tgere and mask some, but its at different physical location and not beaming. On active speakers that is.
Last edited:
Yes, I agree I have coils, and did some test with that before, but the coil did almost nothing, just as the resistor. The first one in post 44 is a comparison with and without the resistor, not much difference actually.. It is the resistor measurement seen in the background of the loopback too. It seems the 18sound mid is not very responsive to increased 'lookback resistance'. I don't think the distortion of it is catastrophic, but I have very few references. It would be nice to see some multitone measurements of some of the acknowledged low distortion mids/woofers as a reference. It is a bit tricky though, since many mics have high distortion themselves, so it's hard to compare measurements taken with different mics and at different levels.
I have coils, and did some test with that before, but the coil did almost nothing, just as the resistor. The first one in post 44 is a comparison with and without the resistor, not much difference actually.. It is the resistor measurement seen in the background of the loopback too. It seems the 18sound mid is not very responsive to increased 'lookback resistance'. I don't think the distortion of it is catastrophic, but I have very few references. It would be nice to see some multitone measurements of some of the acknowledged low distortion mids/woofers as a reference. It is a bit tricky though, since many mics have high distortion themselves, so it's hard to compare measurements taken with different mics and at different levels.
Jeah perhaps these kind of big pro drivers don't benefit as much as some smaller "hifi" type without shorting rings and what not.
I need to take IMD measurements on my system, got coils and what not, but haven't had a chance yet. There is quick THD measurement on my 8" pro mid, which I'm not sure is correctly done, but shows difference with and without coil. The coil is quite big though, 2.2mH and starts to rise impedance right after main resonance impedance peak where the shift happens from mechanical impedance to electrical, around 300-400Hz.
I don't know about levels on the measurements, about good listening level. There is clearly drop in 3rd harmonic assuming the measurements are legit. Its so low already that I cannot hear any difference, haven't done proper AB test either. The measurements are with high and low pass filters on, DSP is adjusted roughly to compensate the coil low pass effect. I think there is enough measurements on web that distortion drops with series impedance but if it matters any is not clear Perhaps if loud enough level so that distortion starts to creep to audibility, so wee bit more "headroom".
I need to take IMD measurements on my system, got coils and what not, but haven't had a chance yet. There is quick THD measurement on my 8" pro mid, which I'm not sure is correctly done, but shows difference with and without coil. The coil is quite big though, 2.2mH and starts to rise impedance right after main resonance impedance peak where the shift happens from mechanical impedance to electrical, around 300-400Hz.
I don't know about levels on the measurements, about good listening level. There is clearly drop in 3rd harmonic assuming the measurements are legit. Its so low already that I cannot hear any difference, haven't done proper AB test either. The measurements are with high and low pass filters on, DSP is adjusted roughly to compensate the coil low pass effect. I think there is enough measurements on web that distortion drops with series impedance but if it matters any is not clear
Perhaps if loud enough level so that distortion starts to creep to audibility, so wee bit more "headroom".Attachments
Last edited:
Two-tone distortion tests are useful for evaluating Le(x) and Le(i). See: http://www.zaphaudio.com/nonlinear.html
For the lower tone, pick a frequency that causes significant excursion. For the upper tone, pick a frequency that is above the point where the impedance starts becoming inductive. For a 6.5" midwoofer you might choose something like 100Hz and 2kHz.
If you choose two frequencies where the drivers impedance is resistive, e.g. 400Hz and 500Hz for a 6.5" midwoofer, the test result might not reveal anything that wasn't apparent in a single-tone measurement at either frequency.
In the former case the driver may have exceptionally low harmonic distortion when single-tone measurements are performed at 100Hz or 2kHz but bad IMD when a two-tone measurement is performed.
For the lower tone, pick a frequency that causes significant excursion. For the upper tone, pick a frequency that is above the point where the impedance starts becoming inductive. For a 6.5" midwoofer you might choose something like 100Hz and 2kHz.
If you choose two frequencies where the drivers impedance is resistive, e.g. 400Hz and 500Hz for a 6.5" midwoofer, the test result might not reveal anything that wasn't apparent in a single-tone measurement at either frequency.
In the former case the driver may have exceptionally low harmonic distortion when single-tone measurements are performed at 100Hz or 2kHz but bad IMD when a two-tone measurement is performed.
I think the 18sound should actually have some shorting ring, but I thought the higher impedance should have more benefit with 'basic' drivers, compared to really advanced motors with a lot of measures taken to reduce distortion with voltage drive.
TMM: I kind of like doing the multitone to get a quick visual of the distortion levels. However, I think I should at least use some HP and not feed a mid with LF since that's not how it will be used (XO at 300Hz or so). I'm guessing that playing around with sine sweeps (single or dual tone) would be a good method when trying to find problem areas (distortion spikes) and possibly correcting them, seeing the structure of harmonics etc, but gets kind of time consuming, and I'm lazy
I'm more of a 'overall picture' type, rather than diving in and spending months investigating details.
I find nothing is more valuable than doing a harmonic distortion sweep, to view 2nd to 5th harmonic for all frequencies. Lots of high order (5th) harmonic distortion at upper frequency ranges of a given driver (e.g. 1-3kHz for a midwoofer) can be an indicator of high Le(x) or Le(i). Then you can do an IMD/multi-tone test to confirm.
Here is an example of my setup measuring the raw response of a 10" woofer (no crossover or tweeter measurement). In ARTA, Fs(Hz) is set to 8000 which is required for resolution in the bass frequencies. FFT is 65536, any higher and the measurement takes too much time to collect IMO. Hanning window and exponential averaging. You can also use linear averaging. 12 band per octave really fleshes out and stresses the driver similar to what music does.
From the results below you can see the Scanspeak 26W/8565-00 has good distortion performance at a 95dB test signal SPL at 1m distance, but notice my SM58 mic is actually only 20cm from the driver. This maximizes the signal to noise ratio of the test setup. For lower SPL test signals you'll need to make sure there are no environmental noises such as computer fans or refrigerator noise. I have to use my laptop for these measurements to make sure the room is dead quiet. Side note...the Scanspeak 26W is -62dB down at 300Hz...That's 0.08% distortion at a very high test SPL...this is up there with the best I've heard and measured. Based on the data I would want to low pass this woofer at 800Hz or lower. I've shown harmonic distortion as well for comparison.
From the results below you can see the Scanspeak 26W/8565-00 has good distortion performance at a 95dB test signal SPL at 1m distance, but notice my SM58 mic is actually only 20cm from the driver. This maximizes the signal to noise ratio of the test setup. For lower SPL test signals you'll need to make sure there are no environmental noises such as computer fans or refrigerator noise. I have to use my laptop for these measurements to make sure the room is dead quiet. Side note...the Scanspeak 26W is -62dB down at 300Hz...That's 0.08% distortion at a very high test SPL...this is up there with the best I've heard and measured. Based on the data I would want to low pass this woofer at 800Hz or lower. I've shown harmonic distortion as well for comparison.
Last edited:
Welcome to the thread
I have not focused much on bass distortion (at least yet) as it seems to be assumed the ear is not so sensitive to it. Ok, harmonics will pretty soon be up in the more sensitive midrange..
That is the SM58 you're using, right? Seems to give some pretty good IMD results, so even if FR is not great, it should be good for IMD measurements I assume? Not too expensive either.
Have you drawn any conclusions with 'current drive' or just increasing the 'driver look back impedance'? Seems there are some distortion gains to be had, but it seems to depend on the drivers(motors).
EDIT: Looks like you have more settings in ARTA, is that because you have the paid version, or should I update? You see my options in post 44.
BTW impressive builds you do with the CNC!
I have not focused much on bass distortion (at least yet) as it seems to be assumed the ear is not so sensitive to it. Ok, harmonics will pretty soon be up in the more sensitive midrange..
That is the SM58 you're using, right? Seems to give some pretty good IMD results, so even if FR is not great, it should be good for IMD measurements I assume? Not too expensive either.
Have you drawn any conclusions with 'current drive' or just increasing the 'driver look back impedance'? Seems there are some distortion gains to be had, but it seems to depend on the drivers(motors).
EDIT: Looks like you have more settings in ARTA, is that because you have the paid version, or should I update? You see my options in post 44.
BTW impressive builds you do with the CNC!
Yes, I'm using the SM58 which is the same as the SM57 (musical instrument mic). I get the two mixed up because I've used both. I think the discussion on audibility needs to be separated between harmonic and intermodulation. The body of research is predominantly focused on harmonic distortion. I don't know of any studies that test the audibility of intermodulation distortion, so the verdict is still out on this aspect. I do know that it consistently correlates strongly with subjective sound quality. Over the years it's never let me down as a good metric, that must also be tied in with all the other measurement types. Harmonic distortion doesn't seem to correlate strongly to perceived sound quality and I've almost stopped measuring it, however I'm still trying to understand why sometimes a little bit of H2 and H3 can be subjectively preferable in some instances, as long as IMD is extremely low. So it's not cut and dry in my mind...that lower harmonic is not automatically better. Some of the best compression drivers I've heard have average H2 and H3 performance and excellent IMD performance. It's perplexing.Welcome to the thread
I have not focused much on bass distortion (at least yet) as it seems to be assumed the ear is not so sensitive to it. Ok, harmonics will pretty soon be up in the more sensitive midrange..
That is the SM58 you're using, right? Seems to give some pretty good IMD results, so even if FR is not great, it should be good for IMD measurements I assume? Not too expensive either.
Have you drawn any conclusions with 'current drive' or just increasing the 'driver look back impedance'? Seems there are some distortion gains to be had, but it seems to depend on the drivers(motors).
BTW impressive builds you do with the CNC!
Edit: You need to use the latest version of ARTA which has more tools on multitone testing. The cost of the software is $100 USD I believe. It's a good tool to have. The free version won't let you save, otherwise it's identical.
Last edited:
Yes, I found that sometimes the ear is not happy even if something seems to measure fine. I have not measured much on drivers, mostly amps so far, so I have no correlation between listening and measurements. On amps I have a general feeling that a multitone distortion floor below -80dB is generally 'safe', but that probably is because it's masked by the speaker
I seriously considering getting a SM58, since I'm not sure what the B&K condenser mic is capable of distortion-wise. I don't know if I'm seeing mic distortion or driver distortion. Since you are able to measure really good IMD performance with the SM58 (I think I saw measurements approaching -80dB?) it seems like a good idea to get one..
I seriously considering getting a SM58, since I'm not sure what the B&K condenser mic is capable of distortion-wise. I don't know if I'm seeing mic distortion or driver distortion. Since you are able to measure really good IMD performance with the SM58 (I think I saw measurements approaching -80dB?) it seems like a good idea to get one..
All this discussion on IMD has convinced me to order the Aco Pacific 7052PH at $1,434 CAD each, just to see if I can improve the measurement system and perhaps reveal even lower distortion figures. -78dB is about the limit with the SM58. To me it seems plausible that driver distortion is much lower than anyone suspected. This goes a long way towards debunking the long held assumptions that speakers are orders of magnitude higher distortion than upstream components. Even Klang-Ton magazine appears to have a limitation in their measurement system, where bass THD never goes below 1% for any of their speaker tests.Yes, I found that sometimes the ear is not happy even if something seems to measure fine. I have not measured much on drivers, mostly amps so far, so I have no correlation between listening and measurements. On amps I have a general feeling that a multitone distortion floor below -80dB is generally 'safe', but that probably is because it's masked by the speaker
I seriously considering getting a SM58, since I'm not sure what the B&K condenser mic is capable of distortion-wise. I don't know if I'm seeing mic distortion or driver distortion. Since you are able to measure really good IMD performance with the SM58 (I think I saw measurements approaching -80dB?) it seems like a good idea to get one..