It doesn't need to be omni. Isn't it contradictory that we always apply windowing, but insist on using omni mics? We always want to measure direct sound, not reflected. Omni is required for room acoustics, but not for loudspeaker measurements. When focusing on distortion measurements, it's not necessary to have flat frequency response either, but it would help. It's got to be full range though. Right tool for the job gives best results.
Most measurement mics are condenser mics, but it's stupid how much high performance mics cost. UMIK-2 has good specs, but it's an USB mic. Similarly specced XLR mics cost >500 € which just isn't right. Heck it's even difficult to find a mic that is on par with UMIK-2 specs let alone price.
Most measurement mics are condenser mics, but it's stupid how much high performance mics cost. UMIK-2 has good specs, but it's an USB mic. Similarly specced XLR mics cost >500 € which just isn't right. Heck it's even difficult to find a mic that is on par with UMIK-2 specs let alone price.
IMO 100€ el-mics with calibration files are totally ok for speaker measurements. Distortion appears as speaker spl goes up and is always different with different speakers. Room noise is much more a problem. And now we are talking about us amateurs!
I guess there are studies about this.
I guess there are studies about this.
Last edited:
But frequency response will be distance dependent.
If you only do relative distorsion measurements, that's probably perfectly ok.
I'm not sure what you mean by that, but using windowing relieves us DIYers from building anechoid chambers, rooms without echo. We want to measure only the direct sound. If FR changes by distance, it's most likely because of room reflections and if measuring multiple drivers at once, summation/lobing. Lobing should show up also in highly directional microphones.
After some listening since the 'high drive impedance mod' I'm actually convinced the sound has improved. Higher mids seem more open and relaxed, less 'congested' if that is a valid description? I will keep the coil and resistor in place for now, maybe a modified amp to adjust output impedance could be in the future projects list..
Since I'm using one of those expensive measurement mics (B&K), it seems to be enough to do distortion measurements at least to abt -60dB, so I deem that to be enough for now. I don't think frequency response is important to do back to back comparisons of distortion levels. As JC found, the SM58 should work just fine.
Since I'm using one of those expensive measurement mics (B&K), it seems to be enough to do distortion measurements at least to abt -60dB, so I deem that to be enough for now. I don't think frequency response is important to do back to back comparisons of distortion levels. As JC found, the SM58 should work just fine.
Have a look at this graph of a supercardioid mic with responses depending on distance:
You can equalize it, but only for a specific distance.
Source:
https://mynewmicrophone.com/complete-guide-to-microphone-frequency-response-with-mic-examples/
You don't understand the difference between a pressure mic and velocity microphone. Look it up for details.
A velocity microphone (everything but omni) changes it's frequency response when comming in the nearfield of a speaker. The closer, the more bass. And not a little - that's 20-30dB. Nearfield depends on the frequency - it's a mess all over.
It's not stupid - it's a complete different technology! Getting a real Class 1 capsule + electronics for 500,- is a STEAL! These are WAY more stable, way less THD, higer max SPL etc. And fit in a stadard 1/2" calibrator.
A professional mic + electronics + preamp is >2000,-
And what happens with your measured THD when the mic has a 10dB boost? And don't reproduce >10kHz?
Using such a microphone and draw conclusions from the measurements ... is a mistake.
(until you have A LOT of knowledge about behaviour of microphones and that specific microphone. But then you would not use it first hand
)
It should be ok as long as you just measure thd in percent and don't overload the mic.
Edit: @IamJF I stand corrected. See next post!
But the again I really don't know much about mics. Not as much as you probably know, that is.
The mic graph was really just an example for disrance-dependency - not recommended for measuring!
Last edited:
You measure the THD components 10dB to high! With a boost at 3kHz you measure wrong H2 at 1,5kHz, wrong H3 at 1kHz, wrong H4 at ...
I started my career at AKG, developing electronics for condenser microphones. Learned a bit about it at university and doing acoustic measurements for >20 years. And NEVER I had a use case where a SM57 would bring ANYTHING helpful
I started my career at AKG, developing electronics for condenser microphones. Learned a bit about it at university and doing acoustic measurements for >20 years. And NEVER I had a use case where a SM57 would bring ANYTHING helpful
Okey, I was not aware of different microphone technologies and how they would respond in different situations. Looks like pressure mics are inherently omnidirectional, but that doesn't mean that the pattern as a mathematical model is to blame. In fact, you can have directional recording of sound using more then one omni mics if you place them on an array and apply some math while reading the input. That yields results as if the diaphragm of the mic was the size limited by the array. Array microphones exist (like miniDSP UMA-16) and the math exists (like in space telescopes)(but it's not that hard, the principle is rather simple), but I'm haven't heard anyone combining them yet in context of speaker measurements. In theory you could record and visualize the wave front with the array mic, which would give DIYers some new in low cost tools.
You are aware that these algorithms fall appart when wavelength com in the dimension of the distance between the mics? You hear that with most mic arrays (ok I'm picky as sound engineer but when you know how a comb filter sounds you hear it).
I got the newest Sony noise chancellation headphone with my phone and to be honest - I expected better from one of the tech leaders. When switching to "full through mode" you hear the phasing/comb filtered microphones at high frequencies.
I got the newest Sony noise chancellation headphone with my phone and to be honest - I expected better from one of the tech leaders. When switching to "full through mode" you hear the phasing/comb filtered microphones at high frequencies.
I have been listening to these for a while now, and thought I lost something in the mid-bass, voices, snappy bass, drums etc sounded 'dull'. I was prepared to go back on the RL I connected in series with the mid. However, since the sims are lost, I decided to do some measurements first. I had the woofer and mid on the same delay setting (no time alignment), and that is not how it was before, so I decided to do some measurements with the wrong polarity on the mid, and find the delay setting for deepest suck-out around XO (230Hz atm) with the mic placed half way between the drivers. Turns out 0.5-0.6ms was a good value, so I settled on 0.5ms. It amazes me how much that changed the sound. The FR was basically the same before and after (even if it is a bit hard to measure in room), but mid-bass sounds 'right' again. I see a lot of posts where people claim time-alignment woofer to mid is not important, but I beg to differ.
While I was at it with measurements, I also tried some multitone measurements on the woofers, looking at IMD above XO. I thought maybe the woofers could benefit from some series inductors as well, but they seem fine as is. I used full range multitone to the Flex8, and disabled mid & tweeter, and looked at the FFT. No IMD grass was visible above the 60dB noise floor above XO. Actually, nothing visible at all above noise floor, but the noise floor was higher in the bass, and resolution of the FFT was also degraded, but still I don't think there is anything to gain from passive components in series with the woofer. Measurements were made in my apartment, so only normal listening levels.
I definitely feel I have made a step forward with the sound. Clearer upper mids & treble, and bass is back to what I had before. I'm a bit curious about doing some 'phase correction'. However FIR is only possible on the main channels in the Flex8, so it would need to be one correction for each channel. I played around with that a bit some years ago (on passive speakers) using EQAPO and RePhase, but never got it to sound 'right' to my ears. I tried reading a lot of threads about it then, but most of it is wiped from memory now..
While I was at it with measurements, I also tried some multitone measurements on the woofers, looking at IMD above XO. I thought maybe the woofers could benefit from some series inductors as well, but they seem fine as is. I used full range multitone to the Flex8, and disabled mid & tweeter, and looked at the FFT. No IMD grass was visible above the 60dB noise floor above XO. Actually, nothing visible at all above noise floor, but the noise floor was higher in the bass, and resolution of the FFT was also degraded, but still I don't think there is anything to gain from passive components in series with the woofer. Measurements were made in my apartment, so only normal listening levels.
I definitely feel I have made a step forward with the sound. Clearer upper mids & treble, and bass is back to what I had before. I'm a bit curious about doing some 'phase correction'. However FIR is only possible on the main channels in the Flex8, so it would need to be one correction for each channel. I played around with that a bit some years ago (on passive speakers) using EQAPO and RePhase, but never got it to sound 'right' to my ears. I tried reading a lot of threads about it then, but most of it is wiped from memory now..
Some more playing around with these speakers this weekend. First I tried a small tube-amp I built on the tweeters, and subjectively it sounds nicer. Looking at the multitone on the tweeters, I can't see a difference in the 'distortion floor', even if the distortion from the amp is significantly higher than the transistor amps I tried before, it seems to still 'hide' behind the tweeters own distortion.
After some listening I still thought there was like some voices 'resonated' is some way in the lower mids that was not there before I added the RL in series with the mids, so now I removed the resistors, and run only 1,5mH coils. This will probably increase the distortion a bit in the lower mids, but maybe it will improve the resonace issue.
After some listening I still thought there was like some voices 'resonated' is some way in the lower mids that was not there before I added the RL in series with the mids, so now I removed the resistors, and run only 1,5mH coils. This will probably increase the distortion a bit in the lower mids, but maybe it will improve the resonace issue.
Just to confirm the subjective impression of voices in my previous post, some voices sounded odd/resonant, and removing the 8ohm resistor in series with the mid, and only keeping the 1,5mH coil has cured it. Maybe some box or driver resonance that is improved with the added the electrical damping?
I think only using the coil is a good enough compromise. The increased lookback impedance seen by the driver in the higher frequencies, and resulting distortion reduction, has certainly improved the subjective upper midrange, so I'm no longer thinking about replacing the 18sound driver. Compensation to 'flat' was done by DSP with every iteration of the process.
I still like the tube amp on the tweeter, if does add it's 'flavor' to the sound though, best enjoyed with acoustic or otherwise less 'busy' music, but still sounds a bit nicer with metal too. I still feel like trying some other amps though, but that's just part of the hobby
I think only using the coil is a good enough compromise. The increased lookback impedance seen by the driver in the higher frequencies, and resulting distortion reduction, has certainly improved the subjective upper midrange, so I'm no longer thinking about replacing the 18sound driver. Compensation to 'flat' was done by DSP with every iteration of the process.
I still like the tube amp on the tweeter, if does add it's 'flavor' to the sound though, best enjoyed with acoustic or otherwise less 'busy' music, but still sounds a bit nicer with metal too. I still feel like trying some other amps though, but that's just part of the hobby
Sorry, no measurements saved.. all the previous documentation on this project got lost on a USB stick, and since I haven't cared much to document. From memory most of the benefits are in the higher frequencies. I have active XO at 2k, and with multitone and XO active (amp connected directly to driver), the distortion 'grass' kept going while the output signal was decreasing.
I would recommend you try it with a 1-1,5mH coil before the driver. It could have some effects in the bass as well, reacting with the impedance of the driver, so watch out for that if you run sims. I cross to the woofer above 200Hz, so not really an issue for me, so that's why used the 1,5mH coil.
If you have a sim and DSP (I can't remember if your speaker was active or passive) so you can easily compensate for the extra coil, I would recommend just trying it and listening, IMO it added some 'clarity' in the midrange. Less tiring sound too. I guess the lower you cross them, the less difference you will hear.
I would recommend you try it with a 1-1,5mH coil before the driver. It could have some effects in the bass as well, reacting with the impedance of the driver, so watch out for that if you run sims. I cross to the woofer above 200Hz, so not really an issue for me, so that's why used the 1,5mH coil.
If you have a sim and DSP (I can't remember if your speaker was active or passive) so you can easily compensate for the extra coil, I would recommend just trying it and listening, IMO it added some 'clarity' in the midrange. Less tiring sound too. I guess the lower you cross them, the less difference you will hear.
I have been listening to these for some time, and with time I started to hear something in the midrange coming from the waveguided tweeters, and I can't 'unhear' it. There is not much in measurements to explain it I think, and the tweeter/WG combo has been measured and praised by Weissmann Acoustics among others. I have also used it before in other speakers, and liked it for the detailed sound. Now it has started annoying me though, like a specific area in the midrange is 'zoomed in' with a lot of detail.
It is definitely the tweeters that irritate me, I verified it by listening to only tweeters by turning off mids and even bass in the DSP.
I had a similar issue with DXT tweeters before, but higher up in the frequency range. I was not able to explain that with measurements either.
To rule out the room, I brought home some modified (fairly good sounding) Yamaha towers I had in the garage, and there is no hint of it. They have less directivity too, and that seems to work ok in my room now after the room treatment I did.
At the moment I'm just working on some ideas in my head. I'm thinking I want to get rid of the waveguide on the tweeter. I can live with less directivity, but I'm not looking for omni
Removing the WG means it won't integrate well with the 8" mids, so I would need to add a 'filler' driver, or possibly go to another mid driver. I would also prefer to lower the LP of the 8" if I keep it, maybe 1k or even lower. I just think it sound better crossed a bit lower.
I don't want to spend a lot of money, and I would prefer to have a passive XO between tweeter an mid if I add another way to the speaker. This is to save a pair of outputs (8 outputs) on the minidsp for the possibility to add some subs in the future. I also prefer not to make a new 'mid box' in the existing upper box, it will be a lot of work because of the angles and braces in the box.
I'm thinking of using some of the following drivers:
Tweeters I have:
Current XT25 tweeters
Similar DX25 tweeters
Seas DXT (prefer not to use them)
Omnes Audio T25H (small size/depth)
Mids:
Dynaudio D54? (vintage mid dome with large round metal face plate)
Peerless TC9
Possible replacement for the 8"
SB17NBAC35 (I can buy them for a decent price, and they seem to measure well from what I've seen)
What I'm thinking at the moment is to keep the 8", use the dome mids, and possibly install the small T25H tweeters to save space on the baffle and get close spacing between the domes since XO will probably be pretty high. I want matching acoustic centers offset between the domesf or passive XO, so it's likely the mid dome would need to be recessed/mounted from the back of the baffle, with some small waveguide-like hole routed in the baffle. I did this before with some tweeters (just a 45deg bit, and rounded off the edges to the baffle), and it worked out really well after some trial and error with diameter and depth. Mounting it from the back could maybe allow the small T25 tweeter to be mounted 'on top' of the Dynaudio dome faceplate from the front of the baffle.
I'm also considering a new 'top box' with a narrower baffle and bevels, but I'm not sure I want to go that far.. A lot of work and more money involved. The baffle is removable on the upper boxes I have, so It should be fairly easy to just replace the baffle, but diffraction could become a problem with the 'less waveguided' drivers.
I can't remember what the Dynaudio dome can do, but I think it can do below 1k and up to maybe 5k. Been a long time since I measured them, but they seemed to measure ok from what I remember..
What does the audience think?
It is definitely the tweeters that irritate me, I verified it by listening to only tweeters by turning off mids and even bass in the DSP.
I had a similar issue with DXT tweeters before, but higher up in the frequency range. I was not able to explain that with measurements either.
To rule out the room, I brought home some modified (fairly good sounding) Yamaha towers I had in the garage, and there is no hint of it. They have less directivity too, and that seems to work ok in my room now after the room treatment I did.
At the moment I'm just working on some ideas in my head. I'm thinking I want to get rid of the waveguide on the tweeter. I can live with less directivity, but I'm not looking for omni
Removing the WG means it won't integrate well with the 8" mids, so I would need to add a 'filler' driver, or possibly go to another mid driver. I would also prefer to lower the LP of the 8" if I keep it, maybe 1k or even lower. I just think it sound better crossed a bit lower.
I don't want to spend a lot of money, and I would prefer to have a passive XO between tweeter an mid if I add another way to the speaker. This is to save a pair of outputs (8 outputs) on the minidsp for the possibility to add some subs in the future. I also prefer not to make a new 'mid box' in the existing upper box, it will be a lot of work because of the angles and braces in the box.
I'm thinking of using some of the following drivers:
Tweeters I have:
Current XT25 tweeters
Similar DX25 tweeters
Seas DXT (prefer not to use them)
Omnes Audio T25H (small size/depth)
Mids:
Dynaudio D54? (vintage mid dome with large round metal face plate)
Peerless TC9
Possible replacement for the 8"
SB17NBAC35 (I can buy them for a decent price, and they seem to measure well from what I've seen)
What I'm thinking at the moment is to keep the 8", use the dome mids, and possibly install the small T25H tweeters to save space on the baffle and get close spacing between the domes since XO will probably be pretty high. I want matching acoustic centers offset between the domesf or passive XO, so it's likely the mid dome would need to be recessed/mounted from the back of the baffle, with some small waveguide-like hole routed in the baffle. I did this before with some tweeters (just a 45deg bit, and rounded off the edges to the baffle), and it worked out really well after some trial and error with diameter and depth. Mounting it from the back could maybe allow the small T25 tweeter to be mounted 'on top' of the Dynaudio dome faceplate from the front of the baffle.
I'm also considering a new 'top box' with a narrower baffle and bevels, but I'm not sure I want to go that far.. A lot of work and more money involved. The baffle is removable on the upper boxes I have, so It should be fairly easy to just replace the baffle, but diffraction could become a problem with the 'less waveguided' drivers.
I can't remember what the Dynaudio dome can do, but I think it can do below 1k and up to maybe 5k. Been a long time since I measured them, but they seemed to measure ok from what I remember..
What does the audience think?
