Hi guys
Pretty simple question, when I need to change the Z location of one of the drivers in a 2 way crossover?
For the crossover I'm currently working the measurements resulted in a 33 mm offset between the midbass and tweeter at the crossover frequency
What doesn't make sense to me is that since the microphone and speaker are completely still across all measurements, wouldn't this offset already be included in the phase measurement?
Thanks
Pretty simple question, when I need to change the Z location of one of the drivers in a 2 way crossover?
For the crossover I'm currently working the measurements resulted in a 33 mm offset between the midbass and tweeter at the crossover frequency
What doesn't make sense to me is that since the microphone and speaker are completely still across all measurements, wouldn't this offset already be included in the phase measurement?
Thanks
Whether or not the offset is in the measurements or not depends on the measurement method/software.
You either need to do dual channel measurements or have a way of locking time zero and do all measurements with the mic in the same positon to get measurements with the offsets (and hence phase data) baked in.
Tony.
You either need to do dual channel measurements or have a way of locking time zero and do all measurements with the mic in the same positon to get measurements with the offsets (and hence phase data) baked in.
Tony.
It is, when you play both at the same time, the interference patterns are horrible.
OK that was kind of a joke. The problem is that phase is relative. So if you measure a tweeter in an anechoic room at 1 meter, and again at 3, the phase should be identical (excluding variations due to room interactions). There is no relative distance in the measurement with most tools.
So we use interferometry to measure the relative distance between them to a very high resolution. NOW we can measure relative phase, and proper crossover design.
Best,
E
Thanks
Yes, I have measured off axis at 10° intervals
But, even if I consider only the on axis response, there is a bit of difference between including the offset or not
With offset (the crossover was optimized using the offset):
Without offset:
Yes, I have measured off axis at 10° intervals
But, even if I consider only the on axis response, there is a bit of difference between including the offset or not
With offset (the crossover was optimized using the offset):
An externally hosted image should be here but it was not working when we last tested it.
Without offset:
An externally hosted image should be here but it was not working when we last tested it.
Which program did you use to do the measurements? Did you follow Kimmo's recommended procedure using dual channel measurements?
If you followed his procedure then the Z offset should already be included, and should NOT be put into the driver z offset. If you for instance converted all of the measurements to minimum phase, then you would need to put in the X Y and Z offsets.
If you need to accrurately work out the Z offset then Jeff Bagbies procedure should give it to you. https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=2ahUKEwjz3YqR7crcAhVOKewKHWqSAcoQFjABegQIABAC&url=http%3A%2F%2Ftechtalk.parts-express.com%2Ffiledata%2Ffetch%3Fid%3D1149302&usg=AOvVaw17azVPpIXj6hirNa7YYlU9
edt: actually if you measured each driver on axis (doing the dual channel) then the Y offset should also be entered (and X offset if the drivers are not all in line vertically), if you measured all drivers on the same axis (eg the tweeter) using dual channel, then NO offsets should be entered.
edit2: The best way to check is to do the measurements and also take measurements with a simple crossover in place (same setup) then check with overlays whether your simulated response matches up with the measured response of that simple crossover. If they don't (within the tolerances expected of the components) then there is something not right in your sim inputs or your measurements.
Tony.
If you followed his procedure then the Z offset should already be included, and should NOT be put into the driver z offset. If you for instance converted all of the measurements to minimum phase, then you would need to put in the X Y and Z offsets.
If you need to accrurately work out the Z offset then Jeff Bagbies procedure should give it to you. https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=2ahUKEwjz3YqR7crcAhVOKewKHWqSAcoQFjABegQIABAC&url=http%3A%2F%2Ftechtalk.parts-express.com%2Ffiledata%2Ffetch%3Fid%3D1149302&usg=AOvVaw17azVPpIXj6hirNa7YYlU9
edt: actually if you measured each driver on axis (doing the dual channel) then the Y offset should also be entered (and X offset if the drivers are not all in line vertically), if you measured all drivers on the same axis (eg the tweeter) using dual channel, then NO offsets should be entered.
edit2: The best way to check is to do the measurements and also take measurements with a simple crossover in place (same setup) then check with overlays whether your simulated response matches up with the measured response of that simple crossover. If they don't (within the tolerances expected of the components) then there is something not right in your sim inputs or your measurements.
Tony.
Last edited:
Invert a driver in your sims, and see.
Yeah, I had tried that, without the offset I get the deep notch, with I get a notch but way shallower
No, I used REW and since I have an UMIK I can't do two channel measurements
I measured both drivers with the microphone in the same position, rotating the speaker on a jig that I build, them measured near field for the midbass and passive radiators and used the merger to arrive in the full range response
The procedure I used is basically the one you posted, measuring each driver and the sum and finding the Dz using any crossover software (Time-Alignment)
Thanks for the tip about using the crossover, I still have to order the parts for this one but I have an old one that was used with the same drivers in another box, the response will not be perfect but as you said will shine some light if what I'm simulating actually happens
OK You might want to consider using holm impulse to do the measurements. It will give you relative phase if you use the "lock time zero" option (without dual channel, but you need to do a sanity check to make sure the latency hasn't changed during the measurements, just do the same measurement as the first and make sure they still line up for time zero). However, I found that I did get much better results with my off axis measurements once I changed to using REW dual channel.
There were some oddities in the sim with vituixcad with my holmimpulse time zero locked measurements that went away with the REW dual channel. However, There was a second variable, in that the holm meausrements were taken all on the tweeter axis (with one measurement for boths M's in my MTM) and the REW measurements were taken on axis with each individual driver, so I can't definitively say whether REW dual channel works better than Holm impulse time zero locked.
I'm not sure if REW has a way to effectively set time zero with single channel measurements. It always (as far as I can tell) sets time zero to the peak of the impulse... Just checked and it has two methods, one as I stated, and the other it estimates it. However I do not believe that the estimate will give you accruate Z offset information.
I'm also not sure whether with time zero set to the impulse peak that the offset will work properly either.
another sanity check you can do (which may be even better than using measuring a crossover) is to Measure each driver individually and then measure both playing together (do it twice once inverted)..
Then use the measurements of both drivers running together as an overlay and just do a crossover with both drivers connected to source with no other components and check if it matches.
I found that I got similar results in the sim when using Holm impulse time zero locked meaurements (with no offets entered for drivers) and minimum phase versions of the same measurements with all driver offsets entered.
Tony.
There were some oddities in the sim with vituixcad with my holmimpulse time zero locked measurements that went away with the REW dual channel. However, There was a second variable, in that the holm meausrements were taken all on the tweeter axis (with one measurement for boths M's in my MTM) and the REW measurements were taken on axis with each individual driver, so I can't definitively say whether REW dual channel works better than Holm impulse time zero locked.
I'm not sure if REW has a way to effectively set time zero with single channel measurements. It always (as far as I can tell) sets time zero to the peak of the impulse... Just checked and it has two methods, one as I stated, and the other it estimates it. However I do not believe that the estimate will give you accruate Z offset information.
I'm also not sure whether with time zero set to the impulse peak that the offset will work properly either.
another sanity check you can do (which may be even better than using measuring a crossover) is to Measure each driver individually and then measure both playing together (do it twice once inverted)..
Then use the measurements of both drivers running together as an overlay and just do a crossover with both drivers connected to source with no other components and check if it matches.
I found that I got similar results in the sim when using Holm impulse time zero locked meaurements (with no offets entered for drivers) and minimum phase versions of the same measurements with all driver offsets entered.
Tony.
Last edited:
Enter correct Y position for each driver (if it's not coaxial driver) and try to improve power response by tuning crossover. With that much S-curve in power response sound is probably harsh in practice. Issue will be clearer after you enter correct Y mm for drivers. Do not forget to check Half space in Options window if measurements cover just 0-90 deg (though 0-180 deg is recommended).
Typically axial pressure of 2-way should be a bit high below x/o frequency and low above x/o frequency to get balanced sound in real room. Too flat axial may sound harsh but dead at the same time.
Decent measurement gear for dual channel mode is not much more expensive than X/O parts for few projects. Please remember that timing of all off-axis responses is shifted compared to axial with REW and Umik-1 in single channel mode, and therefore off-axis simulation is not very reliable. Could be quite okay but we can't be sure.
Typically axial pressure of 2-way should be a bit high below x/o frequency and low above x/o frequency to get balanced sound in real room. Too flat axial may sound harsh but dead at the same time.
Decent measurement gear for dual channel mode is not much more expensive than X/O parts for few projects. Please remember that timing of all off-axis responses is shifted compared to axial with REW and Umik-1 in single channel mode, and therefore off-axis simulation is not very reliable. Could be quite okay but we can't be sure.
Last edited:
You shouldn't have to input a Z offset if both drivers are measured on the finished baffle with the microphone remaining in the same place. The offset will be inherent in the phase information.
I always measure one driver, then the other, then plug both drivers into the amplifier in parallel and do a 3rd measurement. Summing both of the individual measurements in software should achieve the same result as the 3rd measurement. If it doesn't then the delay is off on one of your individual measurements (due to unreliable latency in your measurement system) and you need to correct it in software by increasing/decreasing the delay/offset until the sum of the individual measurements matches the 3rd measurement.
I always measure one driver, then the other, then plug both drivers into the amplifier in parallel and do a 3rd measurement. Summing both of the individual measurements in software should achieve the same result as the 3rd measurement. If it doesn't then the delay is off on one of your individual measurements (due to unreliable latency in your measurement system) and you need to correct it in software by increasing/decreasing the delay/offset until the sum of the individual measurements matches the 3rd measurement.
Last edited:
This applies to semi-dual and dual channel measurement gear and mode, or with constant latency gear. REW + Umik-1 without acoustical reference is "normalizing random latency" gear which could produce unpredictable timing/phase differences for each measurement.
REW has two options for setting IR timing in single channel mode (without electrical or acoustical reference). Checking of Set t=0 at IR peak in Preferences:Analysis tab normalizes timing to highest point of IR peak in every single measurement. That setting totally destroys possibilities to catch timing/phase differences between different radiators/ways no matter how stable mic location is. This setting damages also timing differences between different off-axis measurements because shape of IR changes. Here is W+T sum of ideal LR24 at 500 Hz:
This is total timing mess compared to reality.
Timing differences are much smaller if Set t=0 at IR peak is unchecked, but latency variations and some undocumented normalisations may still exist. The same LR24 at 500 Hz:
An externally hosted image should be here but it was not working when we last tested it.
Quite okay but not the fact.
The same crossover measured in dual channel mode with loopback as timing reference:
This is not recommended gear (and software without better file naming for off-axis measurements).
An externally hosted image should be here but it was not working when we last tested it.
This is not recommended gear (and software without better file naming for off-axis measurements).
An externally hosted image should be here but it was not working when we last tested it.
Thanks, now I understand the problem of not having a reference input, will take into consideration for my next project
As you said, I managed to improve the power response quite a bit, this week I will order the crossover parts, measure everything and see how it sounds
As you said, I managed to improve the power response quite a bit, this week I will order the crossover parts, measure everything and see how it sounds
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.