Phase issues and delay adjustments

[Rob B.]

Hi all, 1st time posting but been a lurker for awhile.
Years ago I built a ported stereo boombox 21"x11"x9" with 2 x 5" bass/mid & 2 x 1" silk dome tweeters, used a KAB 250 amp, passive 2500 crossovers & L-pad and was happy with the results. All drivers in the same enclosure no divider. Box is 3/4" pine, built like a tank, and been taken all over the place with me to work sites, camping, party's etc.
I've recently been trying to recreate this with the KABD-4100 and the Sigmastudio DSP but can't get the sound right and see a lot of phase issues in REW, though not really around the crossover area.
I'm new to DSP's and want to make sure I've understood what I've read as far as setting delays for the drivers, and your input would be greatly appreciated.

-Start with no x-over or eq and measure each driver individually, using one of the tweeters as a reference in the measurement.
-Should I then do the same using the other tweeter? Or will the result be the same as the other 3 drivers delay will all sync to the 1st tweeter used as reference.
-Then check all with the x-over on?
-If I then do nearfield and on axis measurements to try for a quasi-anechoic chamber plot, do I have the x-over on or off again, and by combining the NF and on axis plot will that mess with phase again?

I won't be able to take new measurements until after the holidays (tree and furniture removal), but just wanted to know if this approach works for the single enclosure w/4 drivers, or should I be doing something else.

Thanks and Happy Holiday's!

 

[sm52]

Hi. Are the tweeter and midbass connected to different amplifiers when using a DSP?

 

[Keith_W]

Re: "phase issues". Not sure what you mean. If your phase graph looks like a complete mess, right click the graph in the "Phase & SPL" window, and choose "Estimate IR Delay". But I wouldn't bother looking at the phase graph unless there is something specific you want to look at.

Re: "nearfield measurements". You have to specify what you mean by "nearfield". If you mean "microphone almost touching the driver", this is a bad idea because (1) it may capture "evanescent waves" (google Fraunhofer vs. Fresnel sound fields) and (2) you won't capture the baffle diffraction step (BDS) response. You should measure at least 2x baffle width away from the speakers. This makes it difficult to capture a quasi-anechoic measurement, particularly for long wavelengths. At some point you WILL be limited by your measurement setup and you won't be able to capture a "clean" measurement of your loudspeaker alone.

You will need to gate out reflections from the measurement you wish to correct using REW's FDW and MTW features (click on the IR windows button in REW). Use the impulse response or ETC to look for reflections, this will tell you where to gate.

My recommendation is to (1) ONLY do driver correction IF you can capture a clean quasi-anechoic measurement. This means there will be a lower limit, below which you can not correct, because of limitations of measurement, and (2) forgo low freq driver correction. Correct low frequencies together with the room at MLP.

Re: DSP procedure. This is the process:

1. Using the driver's datasheet as a guide to the operating range of the driver, measure each driver under quasi-anechoic conditions.
2. Use this information to design your XO.
3. Place your mic at a suitable distance to avoid parallax error (I usually place mic at the MLP) and sweep with a timing reference. Tweeter ref is OK but I prefer loopback. Using the impulse response, time align all your drivers.
4. Look for summation errors at the XO point. To tell the difference between a summation error and a room mode, measure at a different distance (1m closer or further away) and see if the dip shifts. Consider the usual tricks - invert polarity, different order slopes, AP filters, etc.
5. Leaving the mic at MLP, take measurements for room correction of low freqs only.

 

[Rob B.]

Hi. Are the tweeter and midbass connected to different amplifiers when using a DSP?

No, the amp is a 4 x 100W @ 6ohms unit.

 

[Rob B.]

Re: "phase issues". Not sure what you mean. If your phase graph looks like a complete mess, right click the graph in the "Phase & SPL" window, and choose "Estimate IR Delay". But I wouldn't bother looking at the phase graph unless there is something specific you want to look at.

Re: "nearfield measurements". You have to specify what you mean by "nearfield". If you mean "microphone almost touching the driver", this is a bad idea because (1) it may capture "evanescent waves" (google Fraunhofer vs. Fresnel sound fields) and (2) you won't capture the baffle diffraction step (BDS) response. You should measure at least 2x baffle width away from the speakers. This makes it difficult to capture a quasi-anechoic measurement, particularly for long wavelengths. At some point you WILL be limited by your measurement setup and you won't be able to capture a "clean" measurement of your loudspeaker alone.

You will need to gate out reflections from the measurement you wish to correct using REW's FDW and MTW features (click on the IR windows button in REW). Use the impulse response or ETC to look for reflections, this will tell you where to gate.

My recommendation is to (1) ONLY do driver correction IF you can capture a clean quasi-anechoic measurement. This means there will be a lower limit, below which you can not correct, because of limitations of measurement, and (2) forgo low freq driver correction. Correct low frequencies together with the room at MLP.

Re: DSP procedure. This is the process:

1. Using the driver's datasheet as a guide to the operating range of the driver, measure each driver under quasi-anechoic conditions.
2. Use this information to design your XO.
3. Place your mic at a suitable distance to avoid parallax error (I usually place mic at the MLP) and sweep with a timing reference. Tweeter ref is OK but I prefer loopback. Using the impulse response, time align all your drivers.
4. Look for summation errors at the XO point. To tell the difference between a summation error and a room mode, measure at a different distance (1m closer or further away) and see if the dip shifts. Consider the usual tricks - invert polarity, different order slopes, AP filters, etc.
5. Leaving the mic at MLP, take measurements for room correction of low freqs only.

Thanks Keith.
The method I used for measurements is explained in a posted link on this site "How to make quasi-anechoic measurements" originally posted in the AVR forum by Napilopez.

I don't have an ability for a loopback, and it's a portable speaker so no MLP.

I do understand the limitations of the quasi-anechoic method, and can get a good clean measurement up to around 5.5ms, or down to around 225-250hz if I recall correctly.

My main concern is with 4 drivers in the single enclosure, 2 of them bass/mids, and each set of W/T's in a stereo setup, is it possible to get a clean(ish?) phase line?
Again, phase doesn't seem to be an issue around a crossover @ 2500, but looks worse mostly between 450-950hz.

Thanks for your response, and I'll try starting with delay corrections in a few weeks.

 

[Rob B.]

Thanks Keith.
The method I used for measurements is explained in a posted link on this site "How to make quasi-anechoic measurements" originally posted in the AVR forum by Napilopez.

I don't have an ability for a loopback, and it's a portable speaker so no MLP.

I do understand the limitations of the quasi-anechoic method, and can get a good clean measurement up to around 5.5ms, or down to around 225-250hz if I recall correctly.

My main concern is with 4 drivers in the single enclosure, 2 of them bass/mids, and each set of W/T's in a stereo setup, is it possible to get a clean(ish?) phase line?
Again, phase doesn't seem to be an issue around a crossover @ 2500, but looks worse mostly between 450-950hz.

Thanks for your response, and I'll try starting with delay corrections in a few weeks.

Sorry, to be clear, I mean that with the crossover @ 2500 the worst of the phase issues are between 450-950hz.

 

[Keith_W]

Phase does not matter as long as there is no dip in the frequency response.

 

[sm52]

the amp is a 4 x 100W @ 6ohms unit.

That is, the left tweeter is controlled by one channel of the amplifier, the left midbass is controlled by the second channel. The same is true for the right channel. Right?

 

[sm52]

My main concern is with 4 drivers in the single enclosure, 2 of them bass/mids, and each set of W/T's in a stereo setup

It's not clear how this works. Could you show a photo?