Here's my first try. Looks a little wonky, probably need to re-do the polars. I scaled the 60 degree polars down by 8dB to match the on axis SPL's, then loaded them into VituixCAD. The X,Y,and Z values are all zeros on the schematic screen because we are using usb style single channel measurements. The tweeter offset of 39 microseconds is entered on the tweeter driver frequency response tab (not shown).
That's fancy stuff, Bill. I am certain that tweeter can handle a lower crossover point without distortion. I last listened to it with a first order electrical XO at 1450 and it sounded great but I was concerned about overworking them.
OK, I'll go back and give a 1st order type filter a try and see what happens. In the meantime, since you are experimenting with lazy susan type polar measurement techniques, I was wondering if you could re-measure and post a new set of polars, keeping the amp volume and mic position constant. Take all measurements pointing at the tweeter at a distance of about 1 meter and record 7 separate FRD's for tweeter alone at 0, 15, 30, 45, 60, 75, and 90 degrees horizontal. Then, with the same volume and mic position, record 0, 15, 30, 45, 60, 75 and 90 degrees for the woofer alone. No need to re-do the tweeter+woofer combined, since we already know the offset is 39 microseconds. This will clean up the polars in VituixCAD a bit. The resulting power response, in-room response, and directivity curves will be slightly skewed and inaccurate, per VituixCAD measurement rules, but this additional data should help clean up the 6 pack screens a bit and give us a rough idea as to how smooth and downward (or upward) sloping these curves are in the critical 2 to 7kHz region.
I used the FRD files that @Eggguy previously posted on 6/2 above.
1) "Chubby eggs tweeter.frd"
2) "Chubby eggs woofer.frd"
3) "Chubby eggs combined.frd"
I loaded the #1 and #2 frd files into separate XSim driver icons, then paralleled the driver icons to the amp icon with direct wire connections, then imported the #3 frd file into XSim using the "get file" drop down function. I got a pretty good match between the two curves by dialing the "mod delay" to a negative -0.53 inches (which is equal to about 39 microseconds). The curves are not a perfect match, but they come pretty close. You can also do this calculation using VituixCAD's ---> Tools ---> Auxiliary ---> Time Align. But I always do it using XSim because that is the way I have always done it in the past (force of habit).
So then we should be good to go in terms of model delay. I loaded the new polars into VituixCAD and the model is looking much better now. There will be a small amount of modeling error caused by the fact that I am using a single model delay for all the polar FRD angles. This single model delay of 39 microseconds will only be correct for the two on-axis curves. This is because, as you rotate the speaker to capture the polar angles, the distance from the microphone to each driver changes (or wobbles) by a small amount. This change (or wobble) is automatically corrected by a 2 channel measurement system. A single channel USB mic, however, will introduce a small amount of phase error. Also, the overall slope of the power response and In-room response curves will be reduced slightly because we are not measuring the full 180 degrees to the rear wall behind the loudspeaker. Keep this in mind when looking at the models.
Still working on a 1st order model. Wow, first order is NOT an easy task! I've printed Wolf's tips from earlier in this thread and am using them as a guide to navigate the squigglies! Should a first order electrical xover have a deep reverse null, or is that something only applicable to 2nd order electrical LR filters?
According to my limited understanding the reverse null will be slight. It seems to me that it is a slight shift up or down of the main lobe at XO. Not certain, just slightly opinionated.
@Eggguy said:
Bill has come a long way since his days of the Beer budget window rattler.
Ya, that was done using the original Radio Shack SPL meter and a Stereo Review warble tone test record. I still have both of them. The speakers, however, are long gone.
Comments
Just kidding. I do understand the importance of Z offset. Because time arrival is a big deal. I arrived 60 years ago and that is significant.
So on tweeter axis, on woofer axis, and combined on mid axis? That seems like cheating.
Sorry, I meant to say something better
I am in construction, I have rough edges but the are plumb and square
No all 3 from the same measuring point. Don't move the mic or change the volume unless you are doing a nearfield measurement.
Even my edges have edges.
Nothing you make has square anything.
Here's my first try. Looks a little wonky, probably need to re-do the polars. I scaled the 60 degree polars down by 8dB to match the on axis SPL's, then loaded them into VituixCAD. The X,Y,and Z values are all zeros on the schematic screen because we are using usb style single channel measurements. The tweeter offset of 39 microseconds is entered on the tweeter driver frequency response tab (not shown).
And the reverse null:
Looks fairly decent, albeit an LR4 alignment.
InDIYana Event Website
That's fancy stuff, Bill. I am certain that tweeter can handle a lower crossover point without distortion. I last listened to it with a first order electrical XO at 1450 and it sounded great but I was concerned about overworking them.
OK, I'll go back and give a 1st order type filter a try and see what happens. In the meantime, since you are experimenting with lazy susan type polar measurement techniques, I was wondering if you could re-measure and post a new set of polars, keeping the amp volume and mic position constant. Take all measurements pointing at the tweeter at a distance of about 1 meter and record 7 separate FRD's for tweeter alone at 0, 15, 30, 45, 60, 75, and 90 degrees horizontal. Then, with the same volume and mic position, record 0, 15, 30, 45, 60, 75 and 90 degrees for the woofer alone. No need to re-do the tweeter+woofer combined, since we already know the offset is 39 microseconds. This will clean up the polars in VituixCAD a bit. The resulting power response, in-room response, and directivity curves will be slightly skewed and inaccurate, per VituixCAD measurement rules, but this additional data should help clean up the 6 pack screens a bit and give us a rough idea as to how smooth and downward (or upward) sloping these curves are in the critical 2 to 7kHz region.
Hope these are helpful
Perfect. I'll load these up and give my model a first order spinorama spin!
Bill where was the offset determined?
I used the FRD files that @Eggguy previously posted on 6/2 above.
1) "Chubby eggs tweeter.frd"
2) "Chubby eggs woofer.frd"
3) "Chubby eggs combined.frd"
I loaded the #1 and #2 frd files into separate XSim driver icons, then paralleled the driver icons to the amp icon with direct wire connections, then imported the #3 frd file into XSim using the "get file" drop down function. I got a pretty good match between the two curves by dialing the "mod delay" to a negative -0.53 inches (which is equal to about 39 microseconds). The curves are not a perfect match, but they come pretty close. You can also do this calculation using VituixCAD's ---> Tools ---> Auxiliary ---> Time Align. But I always do it using XSim because that is the way I have always done it in the past (force of habit).
OK, I wasn't sure if the mic was moving
But we're those done all on tweeter axis without moving mic?
Yes
Yes means I did not move the mic
I know the importance of a steady reference
Keeps me from thinking too highly of myself
So then we should be good to go in terms of model delay. I loaded the new polars into VituixCAD and the model is looking much better now. There will be a small amount of modeling error caused by the fact that I am using a single model delay for all the polar FRD angles. This single model delay of 39 microseconds will only be correct for the two on-axis curves. This is because, as you rotate the speaker to capture the polar angles, the distance from the microphone to each driver changes (or wobbles) by a small amount. This change (or wobble) is automatically corrected by a 2 channel measurement system. A single channel USB mic, however, will introduce a small amount of phase error. Also, the overall slope of the power response and In-room response curves will be reduced slightly because we are not measuring the full 180 degrees to the rear wall behind the loudspeaker. Keep this in mind when looking at the models.
Still working on a 1st order model. Wow, first order is NOT an easy task! I've printed Wolf's tips from earlier in this thread and am using them as a guide to navigate the squigglies! Should a first order electrical xover have a deep reverse null, or is that something only applicable to 2nd order electrical LR filters?
According to my limited understanding the reverse null will be slight. It seems to me that it is a slight shift up or down of the main lobe at XO. Not certain, just slightly opinionated.
Bill has come a long way since his days of the Beer budget window rattler.
May I say, I source of inspiration?
Being the new guy, I think that I will eventually get over my obsession with first order electrical crossovers.
Ya, that was done using the original Radio Shack SPL meter and a Stereo Review warble tone test record. I still have both of them. The speakers, however, are long gone.
Easier to put a full ranger with a whizzer in a big box and call it good . . . Keeps phase flatter too . . .
Yes it is
Point well taken