Looks like the woofer itself just peters out after 2kHz.
Are you trying to cross too high for that woofer?
Will try that thanks Terry 
Loren, it shouldn't, it should be good to 3-4Khz but that's not what's happening and I've not been able to work out why...
The current low pass filter consists of nothing more than a notch filter somewhere around the 4Khz mark. Below is the tweeter and woofer plots with the 3Khz bessel 2nd order curves super-imposed.
I know bessel is not generally used but I wanted to try it because of the superior transient response of a bessel filter, also as a 2nd order it actually (I think) rolls of more sharply than a LR.
1st pic shows the current crossovers responses and the target (red) response.
2nd pic shows the low pass response compared to the actual woofer response.
3rd pic is the current low pass circuit. the tweeter circuit for the above is simply a 4.3uF cap nothing else.
4th pic is the comparison between the notch (blue) and a 220uH coil with 100m ohm series resistance.
having done the 2nd pic shows that the 4K notch considerably affects frequencies lower than the notch frequency!!
edit: I've just realised something else as well. I originally had two notch filters in the lowpass network, but removed one, I did not re-optimize after doing that! If I optimise the current low pass network I get quite different values and a quite different fit to the curve.
Tony.
Loren, it shouldn't, it should be good to 3-4Khz but that's not what's happening and I've not been able to work out why...
The current low pass filter consists of nothing more than a notch filter somewhere around the 4Khz mark. Below is the tweeter and woofer plots with the 3Khz bessel 2nd order curves super-imposed.
I know bessel is not generally used but I wanted to try it because of the superior transient response of a bessel filter, also as a 2nd order it actually (I think) rolls of more sharply than a LR.
1st pic shows the current crossovers responses and the target (red) response.
2nd pic shows the low pass response compared to the actual woofer response.
3rd pic is the current low pass circuit. the tweeter circuit for the above is simply a 4.3uF cap nothing else.
4th pic is the comparison between the notch (blue) and a 220uH coil with 100m ohm series resistance.
having done the 2nd pic shows that the 4K notch considerably affects frequencies lower than the notch frequency!!
edit: I've just realised something else as well. I originally had two notch filters in the lowpass network, but removed one, I did not re-optimize after doing that! If I optimise the current low pass network I get quite different values and a quite different fit to the curve.
Tony.
Attachments
Last edited:
Will try that thanks Terry
Loren, it shouldn't, it should be good to 3-4Khz but that's not what's happening and I've not been able to work out why...
The current low pass filter consists of nothing more than a notch filter somewhere around the 4Khz mark. Below is the tweeter and woofer plots with the 3Khz bessel 2nd order curves super-imposed.
I know bessel is not generally used but I wanted to try it because of the superior transient response of a bessel filter, also as a 2nd order it actually (I think) rolls of more sharply than a LR.
1st pic shows the current crossovers responses and the target (red) response.
2nd pic shows the low pass response compared to the actual woofer response.
3rd pic is the current low pass circuit. the tweeter circuit for the above is simply a 4.3uF cap nothing else.
4th pic is the comparison between the notch (blue) and a 220uH coil with 100m ohm series resistance.
having done the 2nd pic shows that the 4K notch considerably affects frequencies lower than the notch frequency!!
Tony.
Is the second plot just the woofer with the crossover removed from the circuit?
If not, do you have that plot?
I just want to isolate the problem.
yes second plot, blue trace is just the woofer, no crossover components. I'm off to bed now so will view in the morning
nearfield plot is last one in post 108 the rise up to 1Khz is baffle diffraction related as far as I can tell.
Tony.
Maybe I am missing something because I joined the party late, but the manufacture's data sheet shows an almost identical SPL plot.
http://www.morelhifi.com/products/pdf/mw144.pdf
So the fallout after 1 kHz seems to be a characteristic of the driver and not a fault of the system.
That would say that the intended purpose of the woofer is to be crossed much lower than 1 kHz so that the suck out (and corresponding cone break up) is suppressed to a point where it does not impact the combined summation of the next driver.
Sorry if I am off track.
The problem here is that man. data shows a 5dB drop at 2K while the actual drop here is 8 - 10 dB. I have measured and used this driver many times and have never had this degree of a problem. The earlier incarnations on these speakers (posts 73, 74) were quite good. I think that the notch filters are stuffing things up.
Terry
You make the assumption that:
1. The measurement techniques and measurement system are correct.
2. That the environment is not a contributing factor.
3. That the cabinet is not a contributing factor.
What is happening at 2kHz does not bother me too much because I see issues with the dip at 1 kHz and see the response is no longer flat after that point. It just seems to me that we are trying to make the woofer perform in a way that it was not intended.
Have both woofers been tested to see if it isn't just a driver issue?
If all four behave the same, then it seems that is just how they work or there is a measurement or other contributing factor confounding the measurement.
Hi Terry and Loren, The original datasheet that I had seen was a different one, this one: http://www.moreleurope.com/data/mediablocks/MW144.pdf this is a very different animal.
I had seen the datasheet that you posted Loren, I think I may have even made comparisons to it in another thread, The thing that I find strange with that one is that the chasm (smaller as Terry notes) is at 1Khz not 2Kz as I am seeing...
1. There is a possibility that there is a problem with the measurements, and Terry has also suggested this, I don't think so but it is certainly possible.
2. The environment again is possibly a contributing factor, but I've tried quite a few different places for measurements and get the same results.
3. I think that it almost certainly is, and perhaps an unfortunate number of things are adding together to make the problem worse. baffle width, distance to top and bottom edges, internal dimentions could all be contributing (negatively) to the problem.
I've measured (at some point in the past) all four woofers, there are some differences but they all have large dips at 2Khz. It might be worth me doing so again as my measurement technique is now MUCH better than when I first started
Terry, the crossover in post 73/74 had the same notch filter as the current one, it just had a second one to get rid of the 1Khz hump as well, though it wasn't very effective, probably due to the high an ESR of the 100uF electrolytic I was using.
I've been listening (only TV) to the speakers with the tweeter pad resistor shorted for the last week and I'm happy with the sound. I've just put on Pink Floyd Momentary lapse of reason to hear some music.
The 2K dip is there, I don't think I can drop the crossover frequency enough to completely eliminate it, the current crossover is masking it, but I'm not sure that is the best approach.
Rabbitz suggested in am other thread for me to try putting some "wings" on the sides of the speaker and measure to see if the problem is reduced, I haven't done that but should, I think I've invested too much time in the cabinets and didn't want to find that they were the problem :rolleyes"
One final point. I really should get on with my active crossover/preamp project! when listening to music, I have been using the HTPC, because I can't use my marantz DV18 (due to no preamp/volume control) some of the problem may be with the sound card/drivers of the PC!
Tony.
I had seen the datasheet that you posted Loren, I think I may have even made comparisons to it in another thread, The thing that I find strange with that one is that the chasm (smaller as Terry notes) is at 1Khz not 2Kz as I am seeing...
1. There is a possibility that there is a problem with the measurements, and Terry has also suggested this, I don't think so but it is certainly possible.
2. The environment again is possibly a contributing factor, but I've tried quite a few different places for measurements and get the same results.
3. I think that it almost certainly is, and perhaps an unfortunate number of things are adding together to make the problem worse. baffle width, distance to top and bottom edges, internal dimentions could all be contributing (negatively) to the problem.
I've measured (at some point in the past) all four woofers, there are some differences but they all have large dips at 2Khz. It might be worth me doing so again as my measurement technique is now MUCH better than when I first started
Terry, the crossover in post 73/74 had the same notch filter as the current one, it just had a second one to get rid of the 1Khz hump as well, though it wasn't very effective, probably due to the high an ESR of the 100uF electrolytic I was using.
I've been listening (only TV) to the speakers with the tweeter pad resistor shorted for the last week and I'm happy with the sound. I've just put on Pink Floyd Momentary lapse of reason to hear some music.
The 2K dip is there, I don't think I can drop the crossover frequency enough to completely eliminate it, the current crossover is masking it, but I'm not sure that is the best approach.
Rabbitz suggested in am other thread for me to try putting some "wings" on the sides of the speaker and measure to see if the problem is reduced, I haven't done that but should, I think I've invested too much time in the cabinets and didn't want to find that they were the problem :rolleyes"
One final point. I really should get on with my active crossover/preamp project! when listening to music, I have been using the HTPC, because I can't use my marantz DV18 (due to no preamp/volume control) some of the problem may be with the sound card/drivers of the PC!
Tony. 
Tony, I was reading your comments on the other thread and thought I'd take a look. It seems you are dabbling in the cone breakup region here and the response you measure won't give the true picture owing to off axis and energy storage issues.
I'd like to make a suggestion. The woofers have a predictable small bump around 1k3 followed by a fall at 2k. Start your rolloff to drop the 1k3 bump and to exploit the 2k drop as part of the crossover. Get the driver down by 6dB before you hit 2k. Then make sure that from 3k onward is down as far as possible. This will operate your woofers where they should sound smooth and content.
I'm not sure how this will place your tweeters but this will need to be dealt with. Using the woofers to fill the hole will only bring issues. The woofers are already starting to become directional where I've sugested you cross, if you let them run higher it will additionally go all unpredictable in the breakup region.
I'd like to make a suggestion. The woofers have a predictable small bump around 1k3 followed by a fall at 2k. Start your rolloff to drop the 1k3 bump and to exploit the 2k drop as part of the crossover. Get the driver down by 6dB before you hit 2k. Then make sure that from 3k onward is down as far as possible. This will operate your woofers where they should sound smooth and content.
I'm not sure how this will place your tweeters but this will need to be dealt with. Using the woofers to fill the hole will only bring issues. The woofers are already starting to become directional where I've sugested you cross, if you let them run higher it will additionally go all unpredictable in the breakup region.
Hi Allen, Thanks for the suggestion! I hadn't seen your comment until just this morning, Hopefully what I've just implemented is along those lines I've only listened mono because I've only done one crossover so far. It sounds very smooth and even when I cranked it up I didn't hear any nastiness.
Picture of the final crossover is below. It is at 2.8Khz and is 4th order bessel acoustic slope. The off axis measurements I did before (with the original crossover) showed very consistent behaviour without any nastiness, so I'm doubtful it is breakup (4khz and up is though I think).
Also attached is the simmed response. Once I get a real measurement I guess I will know if it was truely successful
Tony..
I've only listened mono because I've only done one crossover so far. It sounds very smooth and even when I cranked it up I didn't hear any nastiness. Picture of the final crossover is below. It is at 2.8Khz and is 4th order bessel acoustic slope. The off axis measurements I did before (with the original crossover) showed very consistent behaviour without any nastiness, so I'm doubtful it is breakup (4khz and up is though I think).
Also attached is the simmed response. Once I get a real measurement I guess I will know if it was truely successful
Tony..
Attachments
A few more images
1st pic shows the individual responses of the simmed crossover and how they fit to the target 4th order bessel at 2.8Khz. I'm particularly pleased with the phase tracking through the crossover region, and spent quite a bit of time fiddling in the simulation to get it so.
2nd pic is sim for the old crossover which was nominally 2nd order bessel at 3Khz.
third pic shows my mounting solution for the crossover
fourth pic shows partially completed crossover board. Yes I made the coils myself....
fith pic shows the completed board.
and sixth pic shows it mounted on the back of the speaker.
Note I had great faith in the sim, in that I built the board without first doing a mockup. At this stage I'm not disapointed, but won't know the full story till I've done the second one.
Tony.
1st pic shows the individual responses of the simmed crossover and how they fit to the target 4th order bessel at 2.8Khz. I'm particularly pleased with the phase tracking through the crossover region, and spent quite a bit of time fiddling in the simulation to get it so.
2nd pic is sim for the old crossover which was nominally 2nd order bessel at 3Khz.
third pic shows my mounting solution for the crossover
fourth pic shows partially completed crossover board. Yes I made the coils myself....
fith pic shows the completed board.
and sixth pic shows it mounted on the back of the speaker.
Note I had great faith in the sim, in that I built the board without first doing a mockup. At this stage I'm not disapointed, but won't know the full story till I've done the second one.
Tony.
Attachments
I think you'll be OK. As you know I simulated the needed electrical response too, after finding what I liked with an active crossover. My targets were also 4th order (acoustical) - originaly Bessel but ended up Linkwitz-Riley. They are very close.
With a good impedance measurement of the actual speakers and the filter software - you used S.W., I used PCD - it's pretty easy to find the component values needed. Mine turned out right on target, yours should too, as you did even better filter sims than I did.
With a good impedance measurement of the actual speakers and the filter software - you used S.W., I used PCD - it's pretty easy to find the component values needed. Mine turned out right on target, yours should too, as you did even better filter sims than I did.
Hopefully what I've just implemented is along those lines
I think so. You've found a result that looks good for a number of reasons. And congratulations on fixing your dip.
Thanks Pano, and Allen, it has certainly been a journey up to this point!
I've attached in room response of one speaker at the listening position. This one measured with REW in RTA mode with pink noise. The big spike at 38 Hz is interesting!! Not too bad considering the room has no treatments of any kind. It's been 1/3 octave smoothed. Definitely quite a bit of room gain, as the output below 100Hz is not that much, the box resonance is around 90Hz.
Tony.
I've attached in room response of one speaker at the listening position. This one measured with REW in RTA mode with pink noise. The big spike at 38 Hz is interesting!! Not too bad considering the room has no treatments of any kind. It's been 1/3 octave smoothed. Definitely quite a bit of room gain, as the output below 100Hz is not that much, the box resonance is around 90Hz.
Tony.
Attachments
Here is a sim vs actual for the impedance of the speaker. No gated acoustic measurements as yet, might be a while before I can set up for that.
Note I didn't reboot, and fiddle with the sound card settings to get the real measurement (black) hence it is a bit ragged (and that is 1/8th octave smoothed) It should be smoother than that. The sim is based on original 1/8th octave smoothed measurements of the woofers and tweeter impedances separately.
I think the correlation is not bad, dips and peaks are a bit off in the sim.
Tony.
Note I didn't reboot, and fiddle with the sound card settings to get the real measurement (black) hence it is a bit ragged (and that is 1/8th octave smoothed) It should be smoother than that. The sim is based on original 1/8th octave smoothed measurements of the woofers and tweeter impedances separately.
I think the correlation is not bad, dips and peaks are a bit off in the sim.
Tony.
Attachments
Nice correlation. By the way, did you include the DCRs on the simmed inductors? Only reason I ask is because I found out that when I changed inductor values in SW it scaled the DCRs. After a few changes this produced some excessive values which I sometimes failed to notice at first, and found a nuisance so I stopped doing it this way and used discrete resistors. Could also mention ESR on the Cs.
Hi Allen I did put in the DCR of the coils, though I should double check it. I didn't put in ESR of the caps, I should probably do that. It will be low but still will probably have an influence. The ability to do DCR on the coils was one of the reasons I preferred SW to PCD, which you can only specify DCR on coils in the main crossover and not on ones used for notches.
Tony.
Tony.
I finished the second crossover last night. I'm not going to post any subjective comments for a while because I need to get used to the new sound I always find that after a change it sounds "wrong" but that after listening for a while I get used to it, brain recalibration if you will. So after listening for quite a while with too prominent midrange they sound a bit mid-bassy now. That may actually be because I took the Baffle Step comp too far, but I guess in time that will become apparent.
One thing that is immediately obvious is that the vertical polar has improved. Much less of a difference between sitting and standing than before. With the original crossover even 5cm up or down made a fairly big difference in the high end. It will be interesting to do some measurements.
I also really really should do something about my preamp so that I can use my DV-18 instead of the HT PC with it's onboard sound card output. I suspect I now have speakers that are good enough to show up deficiencies in the source.
Tony.
I always find that after a change it sounds "wrong" but that after listening for a while I get used to it, brain recalibration if you will. So after listening for quite a while with too prominent midrange they sound a bit mid-bassy now. That may actually be because I took the Baffle Step comp too far, but I guess in time that will become apparent. One thing that is immediately obvious is that the vertical polar has improved. Much less of a difference between sitting and standing than before. With the original crossover even 5cm up or down made a fairly big difference in the high end. It will be interesting to do some measurements.
I also really really should do something about my preamp so that I can use my DV-18 instead of the HT PC with it's onboard sound card output. I suspect I now have speakers that are good enough to show up deficiencies in the source.
Tony.
OK time for an update
My parents came to visit a week ago and I sat them down and played Ella and Louis (the whole album) and then, "Private Investigations" off Love over Gold.
They had both heard the previous incarnation, to which they had said they sound nice, but that was about it. This time my mum said. "you've nailed it" They are so clear, and all the instruments are well defined and not mushed together" She also said that the balance was very good (she had previously commented that they were a bit shy in the lower frequencies, which was backed up by the objective measurements.
I've also been very happy, and yesterday I built a new speaker testing stand http://www.diyaudio.com/forums/equi...w-cost-speaker-testing-stand.html#post2776334 and did the obligatory objective measurements.
Below the first image is the holm impulse measurement for comparison between the left and right speakers. The measurement was taken at approx 1.3M as can be seen from the impulse response I didn't have the second speaker in exactly the same location as the first.
The second image shows the on axis and every 15 degrees off axis through 90deg for one speaker.
The third shows the simulated result that I showed before (blue) and the actual measurements Black and Green) I only did quick measurements this time, and I think there is something suspect with the phase, but it's not a bad match for the original sim, especially considering the measurement has been taken using a different stand in a different location and at a different distance
Anyway at this point I'm very happy, and need to get on with the active crossover for the bass units so I can cut them in at around 300Hz.
Tony.
My parents came to visit a week ago and I sat them down and played Ella and Louis (the whole album) and then, "Private Investigations" off Love over Gold.
They had both heard the previous incarnation, to which they had said they sound nice, but that was about it. This time my mum said. "you've nailed it" They are so clear, and all the instruments are well defined and not mushed together" She also said that the balance was very good (she had previously commented that they were a bit shy in the lower frequencies, which was backed up by the objective measurements.
I've also been very happy, and yesterday I built a new speaker testing stand http://www.diyaudio.com/forums/equi...w-cost-speaker-testing-stand.html#post2776334 and did the obligatory objective measurements.
Below the first image is the holm impulse measurement for comparison between the left and right speakers. The measurement was taken at approx 1.3M as can be seen from the impulse response I didn't have the second speaker in exactly the same location as the first.
The second image shows the on axis and every 15 degrees off axis through 90deg for one speaker.
The third shows the simulated result that I showed before (blue) and the actual measurements Black and Green) I only did quick measurements this time, and I think there is something suspect with the phase, but it's not a bad match for the original sim, especially considering the measurement has been taken using a different stand in a different location and at a different distance
Anyway at this point I'm very happy, and need to get on with the active crossover for the bass units so I can cut them in at around 300Hz.
Tony.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.