I picked up a couple of sony floor standers for $30. They are TWW. The tweeters were blown so replaced them. Looked at the xover (all 2 components of it) and had a listen they were pretty awful. I took a measurement and the woofers were playing up to 5Khz. Took everything apart and measured woofers etc separately. They shouldn't play up that high, they get ragged after 3khz, so decided to try model a new xover.
This is my first time trying to model a xover and wanted to get opinions/help on my methods.
Here Goes:
1) my listening distance is 3m. I took measurements with ARTA at 3m at on axis, 15' off axis and 30' off axis. Took the biggest gate i could get on all measurements, extracted minimum phase and saved them for modelling.
2)For the tweet/woof measurements i decided to do the woofers separately eg do the tweet with top woof, and then do the tweet with the bottom woof and save them as reference in LSP CAD to determine offsets correctly with the above measurements.
3) Import the tweet/top woof reference into LSP Cad, load the individual measurements into the xover section simulating setup (i had a cap on tweet to measure), and adjust z-axis on woof until my measurements lined up as good as possible with reference (determining offset, i think).
4) repeat above for the tweet/bottom woof.
5) model xover.
Questions:
Does it matter if the gated measurements are slightly different?
Do the measurements have to be taken from exactly the same impulse point? eg. find where the tweet impulse starts and then make where the woofers gated measurement starts from the tweeter?
Does my method for determining the offsets sound right?
Please be kind. When i modelled the xover i managed to get the phase to track nicely through the xover region which was roughly 1750hz and they sound a heap better. I know that doesn't mean much but....
This is my first time trying to model a xover and wanted to get opinions/help on my methods.
Here Goes:
1) my listening distance is 3m. I took measurements with ARTA at 3m at on axis, 15' off axis and 30' off axis. Took the biggest gate i could get on all measurements, extracted minimum phase and saved them for modelling.
2)For the tweet/woof measurements i decided to do the woofers separately eg do the tweet with top woof, and then do the tweet with the bottom woof and save them as reference in LSP CAD to determine offsets correctly with the above measurements.
3) Import the tweet/top woof reference into LSP Cad, load the individual measurements into the xover section simulating setup (i had a cap on tweet to measure), and adjust z-axis on woof until my measurements lined up as good as possible with reference (determining offset, i think).
4) repeat above for the tweet/bottom woof.
5) model xover.
Questions:
Does it matter if the gated measurements are slightly different?
Do the measurements have to be taken from exactly the same impulse point? eg. find where the tweet impulse starts and then make where the woofers gated measurement starts from the tweeter?
Does my method for determining the offsets sound right?
Please be kind. When i modelled the xover i managed to get the phase to track nicely through the xover region which was roughly 1750hz and they sound a heap better. I know that doesn't mean much but....
Frankly, I wouldn't measure a speaker at 3m unless you have an anechoic chamber. This is why:
Assume the driver is 1m from the floor and you mic is 3m away, the first reflection from the floor has a path driver-mic of 3.6m, so difference between the direct and reflected path is only 0.6m or 1.7 milliseconds. With a gate so small you'll lose info of a sensible part of the midrange.
Ralf
Assume the driver is 1m from the floor and you mic is 3m away, the first reflection from the floor has a path driver-mic of 3.6m, so difference between the direct and reflected path is only 0.6m or 1.7 milliseconds. With a gate so small you'll lose info of a sensible part of the midrange.
Ralf
^ +1 I would not measure at more than 1 - 1.5M distance depending on what your closest object /boundary is. The measurements I ended up using for my crossover design for my MTM's were taken at 1.2M (which was slightly less than the minimum distance to the nearest reflective surface, which in my case was the ceiling).
I'm not 100% following your measurement method. Were all measurements taken with the mic in the same position? or were you moving the mic up and down?
One method you could use is to put the mic on axis with the tweeter and measure each driver separately without moving anything. This will include the offsets in the measurements (assuming that the listening axis will be on axis with the tweeter).
To work out whether your measurements are good or not, the best thing to do is to do as I said above, and then also do a measurement with your crossover (still without moving anything). Simulate the crossover using the individual driver measurements, and then compare that result to the actual crossover measurement. If it is close then you're probably on the money with your measurements
If you take the above approach, then you definitely need to lock the zero point for the subsequent measurements (I know how to do this in Holm but not in Arta) so that your time zero is consistent between the drivers, otherwise your phase data won't be accurate.
I assume you have also done impedance measurements of each driver in the cabinet for use in your sim. accurate impedance measurements are equally important
The other approach (which I have not tried) is to take the measurements of each individual driver on axis with that driver, and put the offsets into the simulation program. I decided not to take this route as I felt that I would be introducing potential inaccuracies when moving the microphone (in my case since I was doing an MTM, everything was symetric around the tweeter axis, which made the decision easy).
Tony.
I'm not 100% following your measurement method. Were all measurements taken with the mic in the same position? or were you moving the mic up and down?
One method you could use is to put the mic on axis with the tweeter and measure each driver separately without moving anything. This will include the offsets in the measurements (assuming that the listening axis will be on axis with the tweeter).
To work out whether your measurements are good or not, the best thing to do is to do as I said above, and then also do a measurement with your crossover (still without moving anything). Simulate the crossover using the individual driver measurements, and then compare that result to the actual crossover measurement. If it is close then you're probably on the money with your measurements
If you take the above approach, then you definitely need to lock the zero point for the subsequent measurements (I know how to do this in Holm but not in Arta) so that your time zero is consistent between the drivers, otherwise your phase data won't be accurate.
I assume you have also done impedance measurements of each driver in the cabinet for use in your sim. accurate impedance measurements are equally important
The other approach (which I have not tried) is to take the measurements of each individual driver on axis with that driver, and put the offsets into the simulation program. I decided not to take this route as I felt that I would be introducing potential inaccuracies when moving the microphone (in my case since I was doing an MTM, everything was symetric around the tweeter axis, which made the decision easy).
Tony.
Thanks for the reply.
Mic position was constant centered on the tweeter axis, no moving of mic, so should include offsets in measurements.
I see what you mean about the small gating . But i don't quite understand how measure at 1m and transfer the offsets to say 3m where i want to model. I have been reading the forums trying to figure that out. It is the part i am struggling with. Any help would be appreciated.
Mic position was constant centered on the tweeter axis, no moving of mic, so should include offsets in measurements.
I see what you mean about the small gating . But i don't quite understand how measure at 1m and transfer the offsets to say 3m where i want to model. I have been reading the forums trying to figure that out. It is the part i am struggling with. Any help would be appreciated.
For crossover modelling you really need clean measurements. The main issue will be whether or not 1M is far enough back for proper Intergration between the drivers to occur. If it is then you don't have a problem.
If it is not then you could consider taking measurements outside (if that is possible). Putting the speaker up higher (say at 2M) with nothing above and plenty of space around should allow you to get quite good measurements at a reasonable distance provided it is a still and quiet day.
If you really want to get your teeth into something on this have a look at this thread http://www.diyaudio.com/forums/multi-way/212394-real-farfield-distance.html
I tried doing measurements at 2M because I was worried about integration of the drivers, I went back to 1.2M and it worked fine. The inter-drivers distances for my speaker are quite small though. approximately 4" from tweeter centre to each M's centre.
edit: measuring on each speakers axis and then putting the vertical offsets into your sim program (if it supports it) may be your other option. Then you can just take 1M measurements of each driver (moving the mic up and down in the vertical plane being careful to keep it at the same distance in the horizontal plain).
Tony.
If it is not then you could consider taking measurements outside (if that is possible). Putting the speaker up higher (say at 2M) with nothing above and plenty of space around should allow you to get quite good measurements at a reasonable distance provided it is a still and quiet day.
If you really want to get your teeth into something on this have a look at this thread http://www.diyaudio.com/forums/multi-way/212394-real-farfield-distance.html
I tried doing measurements at 2M because I was worried about integration of the drivers, I went back to 1.2M and it worked fine. The inter-drivers distances for my speaker are quite small though. approximately 4" from tweeter centre to each M's centre.
edit: measuring on each speakers axis and then putting the vertical offsets into your sim program (if it supports it) may be your other option. Then you can just take 1M measurements of each driver (moving the mic up and down in the vertical plane being careful to keep it at the same distance in the horizontal plain).
Tony.
Last edited:
I can get clean clean measurements at 1m, 1.5m down to about 350hz gate and can determine offsets at these distances. If i model the xover based on these distances it won't be correct at say 3m where my listening area is because the phase shifts (or at least that is how i understand it). So if i do a measurement at say 1m how do i then figure out the offsets at say 3m where i wish to model the xover? Sorry this is probably very painful for some.
Alan
Alan
ah ok, I see what you are asking. I wouldn't worry about that. There will be so many room interactions by the time you get to 3M that it will be irrelevant If you can get clean measurements at 1.5M down to 350 Hz then use those.
Once the sound has integrated from the drivers (ie it has travelled far enough that path length differences are minimal) it will essentially stay the same no matter what the distance. If the sound waves are in phase at 1.5M they will still be in phase at 3M as the speed of sound is constant
The only thing that will change over distance is that the path length difference from driver to mic will get smaller with respect to the measuring distance.
Tony.
If you can get clean measurements at 1.5M down to 350 Hz then use those. Once the sound has integrated from the drivers (ie it has travelled far enough that path length differences are minimal) it will essentially stay the same no matter what the distance. If the sound waves are in phase at 1.5M they will still be in phase at 3M as the speed of sound is constant
The only thing that will change over distance is that the path length difference from driver to mic will get smaller with respect to the measuring distance.
Tony.
Once the sound has integrated from the drivers (ie it has travelled far enough that path length differences are minimal) it will essentially stay the same no matter what the distance. If the sound waves are in phase at 1.5M they will still be in phase at 3M as the speed of sound is constant
Thankyou so much. That was doing my head. I spent hours reading stuff on this forum etc trying to figure that out. You are a life saver and probably a marriage saver. I am forever grateful to you for that information, now i can model away and measure my results and see how i go.
Cheers
Alan
Thankyou so much. That was doing my head. I spent hours reading stuff on this forum etc trying to figure that out. You are a life saver and probably a marriage saver. I am forever grateful to you for that information, now i can model away and measure my results and see how i go.
Cheers
Alan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.