Im building a small PA-speaker with a short horn and 6"midbasses above and below the horn. The horn is approx 14cm deep and the voice coil on the 6" is approx 35mm below the front plate. So there is approx 100mm difference and the sound from the 6"will arrive first. I plan to use a passive crossover at around 2500hz and will start with 12db/oct for bass and 18db/oct for horn.
How much will the time difference affect the freq response? When doing impulse-sweeps with Arta it's clear that the horn is delayed and I don't think is realistic to implement a passive delay on the midbass.
The wavelength at 2500hz is close to 14cm so with a small additional delay on the horn the drivers will be in phase for a continuous sine wave, but horn will still be full cycle delayed?
Does that make any sense to you guys? What's the best approach here? Not looking for perfect, just OK, - and maybe learning a little along the way.
Kind regards TroelsM
How much will the time difference affect the freq response? When doing impulse-sweeps with Arta it's clear that the horn is delayed and I don't think is realistic to implement a passive delay on the midbass.
The wavelength at 2500hz is close to 14cm so with a small additional delay on the horn the drivers will be in phase for a continuous sine wave, but horn will still be full cycle delayed?
Does that make any sense to you guys? What's the best approach here? Not looking for perfect, just OK, - and maybe learning a little along the way.
Kind regards TroelsM
Hi TroelsM,
If you look at the older designs like Pioneer/Tad they use a 12dB highpass end a 36dB lowpass for time alignment.
So you need a higher order on the woofer to get the delay you need.
A gues like 12dB highpass and 24 dB lowpass.
Rob
If you look at the older designs like Pioneer/Tad they use a 12dB highpass end a 36dB lowpass for time alignment.
So you need a higher order on the woofer to get the delay you need.
A gues like 12dB highpass and 24 dB lowpass.
Rob
The woofer's voice coil former is connected to the 6" cone.
There is no time delay between the movement of the former and the cone.
The acoustical point of origin is somewhere around the midpoint of the cone/dustcap.
Each electrical "order" imparts a 90 degree phase shift in addition to the phase shift of the driver.
Two drivers with the same acoustical point of origin each with second order (12dB/oct) acoustical crossover require one of the driver's polarity to be reversed (180 degree phase shift) for phase alignment.
If each "in phase" source is -6dB at the acoustic crossover point, the two will combine to "flat", "0dB".
Each "leg" of a LR crossover is -6dB at the crossover point, BW -3dB.
If the phase is 90 (or 270) degrees off, there will be a -3dB summation, a dip in response at the crossover point.
Rob, that's phase, not time alignment.
Without resorting to passive "all pass" filters (lots of components) for actual time alignment, with passive crossovers and different planes of acoustic point of origins, the best approach is getting the phase alignment as close as possible through the crossover region, which results in a smooth frequency response.
Between a single "order" and a polarity reverse, 270 degrees of phase change is available, "close enough" for rock & roll.
Art
Use a simulator and line up the slope of phase.
Do you accept delay as the slope of phase? The origin for both drivers is going to be held together in the calculations.
Probably, but I prefer cash 😉.
I don't.
Using the slope of phase aka group delay of asymmetric crossover orders, as a substitute for any needed fixed time delay, is a kludge.
Frequency dependent phase/group delay, and fixed delay, are not very good substitutes for each other.
But I can see why passive xover designers use the technique.
Apart from sloping baffles or physically offsetting drivers to effect fixed delay, there's really not any other pragmatic passive alternatives for implementing fixed time delays...that I know of...
So better a kludge than nothing...
The OP said he plans to use passive, and wondering about fixed delay offesets.
And you seemed to post a LCR example of something ....(that I couldn't understand what you were saying as no fixed delay appeared to be involved)
With DSP, do you use group delay as a fixed time substitute? I don't think you do...???
And you seemed to post a LCR example of something ....(that I couldn't understand what you were saying as no fixed delay appeared to be involved)
With DSP, do you use group delay as a fixed time substitute? I don't think you do...???
I think the short answer is it will affect your crossover, in the same way it always does when working on passive crossovers where acoustic origins are not aligned. Yes, the Z offset is a little larger than typical, but once you're over about 1.5 inches at 2500 Hz, you have a similar set of problems to deal with. You can still get a flat frequency response summation through multiple approaches even without time alignment.
Passive designs like Klipsch tend to focus on frequency response and ignore alignment of acoustic origins. Given your use case, you can certainly use the same approach. If you get all the data in a simulator, it'll be more obvious what affects what and which solution you want to implement.
https://www.stereophile.com/content/klipsch-la-scala-al5-loudspeaker-measurements
Passive designs like Klipsch tend to focus on frequency response and ignore alignment of acoustic origins. Given your use case, you can certainly use the same approach. If you get all the data in a simulator, it'll be more obvious what affects what and which solution you want to implement.
https://www.stereophile.com/content/klipsch-la-scala-al5-loudspeaker-measurements
Wow. Some of that is beyond my understanding, but thanks anyway. I've simulated and tried a few different options and I think the result looks ok. Measurements are indoors with plenty of reflections. Mic is placed at horn level. 12dB/18dB for bas/horn. Impedance correction on both (r-c).
4 graphs: overall, with horn reversed, bass alone and horn alone.
Ok for rock and roll?
( The highs drop a lot, but I can live with that og correct it with EQ)
4 graphs: overall, with horn reversed, bass alone and horn alone.
Ok for rock and roll?
( The highs drop a lot, but I can live with that og correct it with EQ)
Assuming your vertical axis is 5 dB/division, it looks like you're down around 5 or 6 dB at the cross point. If that's the case, it might be worth adjusting the crossover a bit. If the vertical scale is something else, let us know.
That would introduce a time of flight offset between the two drivers, the hypotenuse length to the woofer relative to the horn's right angle greater the closer the mic distance.
At one meter that might be ~ 1/4 wavelength (90 degrees) at the acoustical crossover of ~2100Hz. Half of a polarity reversal..
The mic should be equidistant between the center of the woofer and center of horn.
https://www.audiosciencereview.com/...ents-spinoramas-with-rew-and-vituixcad.21860/
Summation is poor in the acoustical crossover region, without knowing the measurement distance (determining the baked-in hypotenuse length offset..) and the vertical scale, hard to tell how poor.
The 4kHz woofer peak flattens the horn dip, but the response would be better off without it.
Art
Maybe my description was bad, but I have 2 woofers. One over the horn and one under. Therefore I placed mic between the woofers at horn level. I do see some rather large differences in response when I move the mic up and down. I assume it's better to measure at a larger distance (outdoors), to minimize the difference in distance to the individual drivers?
Do you rather do more steep IIR filters than introducing a real delay when you have a physical distance offset?
//