Why does output invert at 2300khz?

[slobmw]

Hi all, I'm playing around with active crossover in a set of Pioneer FS52's I had lying around. I'm trying to measure the raw driver output in the tower - passive crossover removed and amplifier power direct to driver. For both the woofers and the mid, seperately, I get a phase shift in the middle of the frequency sweep. I have a minidsp nanoavr as the dsp device, but all peq and xover is bypassed.

See the attached jpeg for the mid graph. Nicely linear-ish up to 2300khz or so, then changes 360 deg.

These are crossed over at 2000khz in the factory xover.

I'm a noob, so have little doubt there is a simple answer, I just don't see a reason for this shift,

Thanks for any help

slobmw

 

[scott wurcer]

Hi all, I'm playing around with active crossover in a set of Pioneer FS52's I had lying around. I'm trying to measure the raw driver output in the tower - passive crossover removed and amplifier power direct to driver. For both the woofers and the mid, seperately, I get a phase shift in the middle of the frequency sweep. I have a minidsp nanoavr as the dsp device, but all peq and xover is bypassed.

See the attached jpeg for the mid graph. Nicely linear-ish up to 2300khz or so, then changes 360 deg.

These are crossed over at 2000khz in the factory xover.

I'm a noob, so have little doubt there is a simple answer, I just don't see a reason for this shift,

Thanks for any help

slobmw

Phase wrapping, phase goes from 0 to 2pi and back again (360 degrees in a circle). Think of going round and round in a circle. SPL and Phase Graph

 

[jackinnj]

You can mathematically "unwrap" phase. My VNA allows for this with the click of a mouse.

 

[eriksquires]

In traditional multi-way crossovers you often have to invert a driver so that the phase through the crossover region aligns ideally.



That is, you may have the W in + phase, then reverse the mid, then have the tweeter in the same phase as the woofer.



It sounds like you are measuring that.

 

[slobmw]

Thanks for the replies, but I admit to still being unclear as to what is happening.

Phase wrapping, phase goes from 0 to 2pi and back again (360 degrees in a circle). Think of going round and round in a circle.

I think this is unwrapped... To me this isn't a wrap, but a decision by the driver to do a quick loop d'loop, and then return to trend. So I think I know what it did, I'm trying to figure out why a single driver - no other drivers driven - would do such a thing.

In traditional multi-way crossovers you often have to invert a driver so that the phase through the crossover region aligns ideally.

This is measuring a single driver, so that shouldn't be an issue here. I did invert that particuar driver, again driven by itself, and the drop disappears. But its unclear to me why running a single driver "in reverse" as it were, would eliminate that shift.

 

[scott wurcer]

I think this is unwrapped... To me this isn't a wrap, but a decision by the driver to do a quick loop d'loop, and then return to trend. So I think I know what it did, I'm trying to figure out why a single driver - no other drivers driven - would do such a thing.

This is physically unreal 0 = 360 mathematically there is no difference, reversing phase = 180 BTW. There is possibly a quirk in the computation. What is the orange plot, the sharp discontinuity could be an indicator of the problem spot?

 

[charlie2]

Hi all, I'm playing around with active crossover in a set of Pioneer FS52's I had lying around. I'm trying to measure the raw driver output in the tower - passive crossover removed and amplifier power direct to driver. For both the woofers and the mid, seperately, I get a phase shift in the middle of the frequency sweep. I have a minidsp nanoavr as the dsp device, but all peq and xover is bypassed.

See the attached jpeg for the mid graph. Nicely linear-ish up to 2300khz or so, then changes 360 deg.

These are crossed over at 2000khz in the factory xover.

I'm a noob, so have little doubt there is a simple answer, I just don't see a reason for this shift,

Thanks for any help

slobmw

I may be wrong but you are only connecting one driver at a time for measuring

If not make sure the minidsp is deactivated in all channels not used

 

[JohnPM]

Is there a sharp dip in the amplitude response there? 360 degree steps are typically of the phase trace as it passes through a null, a non-minimum phase part of the response.

 

[slobmw]

Is there a sharp dip in the amplitude response there? 360 degree steps are typically of the phase trace as it passes through a null, a non-minimum phase part of the response.

Thanks, yes. When I look at the minimum phase graph I see a sharp spike at that point. That seems to be consistent across all my measurements.

That actually fits with something I just observed as I'm playing with putting the pieces together. Having put all the drivers together with Bessel crossovers at 250 and 3k, I have what looks to me like a nice smooth-ish phase to about 6.3 khz. I see the 360 degree step, and the only apparent null, where, as you suggest I would, I see a spike in the minimum phase.

So... do I say, yep. null at 6.3khz and move on, or is that evidence that something can be fixed?

 

[bolserst]

If you saved the measurement data, you might check to see what you have the impulse response window size set at. By default, REW usually sets the right hand window to something large like 500mS...which is good for measuring rooms. But in your case, you are wanting to see the response of the driver. Setting the right hand window to something in the range of 3mS to 5mS will remove reflections from the impulse response that can cause odd looking phase wrap behavior. Don't forget to click on the "Apply Windows" button after changing the window size, then go back and look at the frequency/phase response to see what affect it had.

Attachment take from here: Loudspeaker measurements