Some of this is going to be the natural group delay of the crossover. However first you might try reversing the polarity, you seem to have some cancellation.
Not sure whether this helps, but this what I see when I open it in REW. Screen capture provides some setting information. I am a bit confused by what I am seeing.
This thread may/may not shed some useful light on the topic of LR 2nd order electrical alignment vs acoustical alignment: https://www.diyaudio.com/community/threads/l-r-12db-oct-tweeters-reversed-or-not.101489/
I would recommend reversing tweeter phase and remeasuring, if you have got it right as is there should now be a big dip in the XO region, if not there are other phase contributors to consider and you might just want to measure the phase response of the tweeter by itself to understand what it is contributing. (No XO, just a protection capacitor in series to keep DC and LF out of the tweeter.)
As an additional thought measure the phase and amplitude response of the XO into resistive loads and check that it is what you expect. I used LR 12dB passive networks in an older iteration of my system and don't remember this as being typical.
This thread may/may not shed some useful light on the topic of LR 2nd order electrical alignment vs acoustical alignment: https://www.diyaudio.com/community/threads/l-r-12db-oct-tweeters-reversed-or-not.101489/
I would recommend reversing tweeter phase and remeasuring, if you have got it right as is there should now be a big dip in the XO region, if not there are other phase contributors to consider and you might just want to measure the phase response of the tweeter by itself to understand what it is contributing. (No XO, just a protection capacitor in series to keep DC and LF out of the tweeter.)
As an additional thought measure the phase and amplitude response of the XO into resistive loads and check that it is what you expect. I used LR 12dB passive networks in an older iteration of my system and don't remember this as being typical.
The Twitter is already reversed when I measured since the 2nd order LR as I said before shift the phase for 180 degree, which require polarity switch on one driver.
Thanks for the resource I am going to check it out.
But I am going to measure it tomorrow with normal polarity on both speakers just to see if I will have bigger cancelation to confirm that the driver is completely opposite as expected for 2nd order LR crossover. (it's already late here 1:30 AM so I can't measure lol)
Thanks for the resource I am going to check it out.
But I am going to measure it tomorrow with normal polarity on both speakers just to see if I will have bigger cancelation to confirm that the driver is completely opposite as expected for 2nd order LR crossover. (it's already late here 1:30 AM so I can't measure lol)
I see your window is ~100 msec, but looking at your impulse response you have significant reflections between 2.5 and 20 msec. If we window it down to 2.5 msec the dip at 4.25k Hz looks deeper. I'd disregard info < 1000 Hz due to reflections. Measuring with the tweeter inverted (relative to what you have already presented) may be useful, but I think your XO slopes look shallow (post 19) and we would see a wider null if polarity were the issue. Your XO looks looks like a normal 2nd order with some extra stuff I'm not sure what it's doing (attachment, post 1).
If you were able to do some ground plane measurements in an open space (and tilt the speaker to keep the mic on-axis) (both system level, and individual drivers only) for measurements with fewer reflections, I wonder if some of this unusual behavior we see would evaporate.
If you were able to do some ground plane measurements in an open space (and tilt the speaker to keep the mic on-axis) (both system level, and individual drivers only) for measurements with fewer reflections, I wonder if some of this unusual behavior we see would evaporate.
I see your window is ~100 msec, but looking at your impulse response you have significant reflections between 2.5 and 20 msec. If we window it down to 2.5 msec the dip at 4.25k Hz looks deeper. I'd disregard info < 1000 Hz due to reflections. Measuring with the tweeter inverted (relative to what you have already presented) may be useful, but I think your XO slopes look shallow (post 19) and we would see a wider null if polarity were the issue. Your XO looks looks like a normal 2nd order with some extra stuff I'm not sure what it's doing (attachment, post 1).
If you were able to do some ground plane measurements in an open space (and tilt the speaker to keep the mic on-axis) (both system level, and individual drivers only) for measurements with fewer reflections, I wonder if some of this unusual behavior we see would evaporate.
Thank you for your response!
I am going to try that and see what I'll get. I am also planning to change the 3.9uF capacitors to 6.8uF to move the XO to 2.5khz and see if that phase shift will repeat and the XO region again.
I am using DBX RTA-M omni mic. Roksan K3 amp and Icon Cube6nano as soundcard.
Check your settings on the interface. Turn the direct monitoring completely off. I’m not acquainted with Icon interfaces, do they have any sound processing chip in the digital signal path?
I am using DBX RTA-M omni mic
Hey, maybe you already know but in case you don't, the DBX RTA-M has a big HF peak on-axis, and is made to be used pointed at the ceiling, I guess.
See:
So point the speaker at the mic, but point the mic 70-90 deg off from the speaker.
Response (on-axis) is something like this, according to the link...
I had one for a bit and after I heard about this I clamped it in place near a USB dayton mic with a cal file (made a fixture to hold them quite close) and the delta I saw was similar to the above, I saw the DBX as +9.5 dB @ 14.5 kHz.
Today I changed the first 2 capacitors from 3.9uF to 6.8uF which shifted the XO at around 2.5Khz. I had no time to measure the speaker but it sound WAY better then before.
This is what XSim is telling me:
I am going to post the new REW measurements once I get chance to do that so we'll see how that will looks like.
Before I had the crossover at around where I had the phase issues...
My drivers are placed at 13.916cm above each other from the center point of the mid-bass which is = on 2500hz wave length which was my intention so I can corelate that with the XO (Not sure if this is important at all...)
This is what XSim is telling me:
I am going to post the new REW measurements once I get chance to do that so we'll see how that will looks like.
Before I had the crossover at around where I had the phase issues...
My drivers are placed at 13.916cm above each other from the center point of the mid-bass which is = on 2500hz wave length which was my intention so I can corelate that with the XO (Not sure if this is important at all...)
There seems to be a big overlap between the drivers before they roll off, not sure if this is intended.
Just as a hint, usually the impedance has a peak at the crossover frequency.
Also the caps c4 and c5 (and r2!) don't do anything but reduce the impedance at the top end. Does this have any purpose?
Just as a hint, usually the impedance has a peak at the crossover frequency.
Also the caps c4 and c5 (and r2!) don't do anything but reduce the impedance at the top end. Does this have any purpose?
Should I make steeper slope there? btw, the c4 and c5 (and r2) are there so I can hit the desired 2.5khz crossover + to attenuate the twitter since it's more sensitive and producing more SPL. When they're not there the crossover is around 1.2khz
Do you still have the phase anomalies in your measurements?
Anyway, go and have a look at other crossovers. A lot preferably. Pay attention to tweeter attenuation, impedance flattening tricks, phase matching in the summing signal and baffle step correction setups.
Anyway, go and have a look at other crossovers. A lot preferably. Pay attention to tweeter attenuation, impedance flattening tricks, phase matching in the summing signal and baffle step correction setups.
You should match the crossover frequency of both drivers. In your graph it seems the woofer has 4 kHz crossover frequency with some slight downward slope before.Should I make steeper slope there?
And the tweeter has about 1200 Hz crossover frequency, just as you state.
You should be able to create one common x-over frequency for both drivers.
You need an effective series element to attenuate signal by parallel elements. But you only have a cap tha thas near 0 ohm impedance at high frequencies.btw, the c4 and c5 (and r2) are there so I can hit the desired 2.5khz crossover + to attenuate the twitter since it's more sensitive and producing more SPL.
Yes!!A lot preferably.
With the last screenshot I was just demonstrating what those set of components do in my design since you asked. If you notice I've disconnected them to show that they're needed so I can have the XO at 2.5K.Does this have any purpose?
I second the recommendation to look at a lot of other crossover designs. 😀 The crossover components you have added should not be necessary and reflect a very low load impedance back to the amplifier at high frequencies. You can use a fixed (or adjustable) LPAD to pad down the tweeter output and that may also result in more consistent crossover behavior due to a relatively constant load impedance on the crossover over frequency and attenuation.
Ok, so It's better to have slightly higher impedance then to have lower impedance. Is that right? How that reflects to the freq. response, & is it possible to harm the amp in some way?very low load impedance back to the amplifier at high frequencies
no, that's not generally true.Ok, so It's better to have slightly higher impedance then to have lower impedance.
but it's bad to uselessly burn amplifier power (which your crossover does).
it may be beneficial to keep the impedance constant, in case of high output impedance amplifiers (generally tube amps).
I see. Thanks for the tip. It's logically that the impedance should be constant so I was just wondering what it'is more preferred.
However that isn't the case and that's not the reason it was asked.It's logically that the impedance should be constant
Could I ask how it is that your individual responses are smoothed and your combined one is not?
- Home
- Loudspeakers
- Multi-Way
- Crossover help needed - too fast phase shift around 3.7-4-7Khz