The problem with passive crossovers as far as I know are power loss and a 90/180/270 degree phase delay. So if you have a high pass first order crossover on a cap, (a cap in series with a driver) it should delay all signals by 90 degrees. This means that if you feed the speaker a 100hz sine wave mixed with a 10khz sine wave, one signal would be shifted past the other. The 100hz wave would be shifted by 2.475 msec compared to the 10khz wave. I'm not sure how audible this shift is, some claim it makes all the difference
A passive low pass crossover like this one is supposed to have a variable phase shift:
Signal --> @----/\/\/\/-----[OUT]-----|(----GND
The reactance of the cap increases as frequency decreases, shorting less of the signal to ground allowing more to pass to the output. So at lower frequencies the phase shift is very low. The phase shift here I believe is supposed to increase with frequency. So If the right values were selected for the cap and resistor, is it possible that it could delay low freqencies by a few degrees, and high frequencies by more degrees so that they would arrive at the same time? 90 degrees at 20hz and 20Khz are different amounts of time. If this crossover could delay waves by a fixed amount of time regardless of frequency, it would be a perfect crossover. The two frequencies would arrive late, but they would be aligned as they were at the input, adjusted only in amplitude, no phase shift.
A passive low pass crossover like this one is supposed to have a variable phase shift:
Signal --> @----/\/\/\/-----[OUT]-----|(----GND
The reactance of the cap increases as frequency decreases, shorting less of the signal to ground allowing more to pass to the output. So at lower frequencies the phase shift is very low. The phase shift here I believe is supposed to increase with frequency. So If the right values were selected for the cap and resistor, is it possible that it could delay low freqencies by a few degrees, and high frequencies by more degrees so that they would arrive at the same time? 90 degrees at 20hz and 20Khz are different amounts of time. If this crossover could delay waves by a fixed amount of time regardless of frequency, it would be a perfect crossover. The two frequencies would arrive late, but they would be aligned as they were at the input, adjusted only in amplitude, no phase shift.
I find 6dB is still the best. 2nd and 3rd orders simply mangle the music. Have not worked enough on 4th order Linkwitz-Riley to commend though.There's some interesting notes on crossovers here:Multi-way Crossover Design
"The problem with passive crossovers as far as I know are power loss and a 90/180/270 degree phase delay"
The phase shift is what makes the crossover work.
The 'power loss' is only from the DCR of the inductor, and is usually trivial.
12dB and 18dB crossovers usually don't sound right no matter how you phase the drivers.
24dB can sound very good.
The crossover is the sum of the acoustic response of the drivers and the electrical response of the filters.
One of the best sounding small speakers that I have heard was a 6" two way with a 1" tweeter, and a 24dB crossover at 2Khz.
Electrically the low pass feeding the 6" was 12dB, the high pass feeding the 1" was 18dB. The power response of the 6" was falling at 12dB above 2Khz so the combined roll off was 24dB/oct. The Q of the 12dB low pass filter was adjusted so the combined response was -6dB at 2Khz, IOW Linkwitz Riley. The power response of the tweeter rolled off below 2Khz at 6dB so the combined roll off was 24dB. The Q of the 18dB high pass filter was adjusted so the combined response was -6dB at 2Khz.
A short conical waveguide was used with the tweeter for two reasons. One was to limit the directivity of the tweeter at the crossover point so that it matched the directivity of the woofer closer. The second was to get the voice coil of the tweeter closer to the plane of the woofer.
The HF driver was connected reverse phase through an L-pad and adjusted for maximum notch at 2Khz. The L-pad was replace with high quality fixed resistors and the tweeter was wired in normal phase.
At the time I did this, ca 1988, there was nothing commercially available similar to this. Today this is old hat.
6dB can sound good too.
But it will probably have to be a three way to have any real output capability.
A Bakgaard two way has the best transient response but is quite complex. With the phase link driver and the extra crossover parts involved the cost and complexity will be on the order of a three way.
I used to use a Bakgaard crossover for PA, and switched to 24dB for the largest venues.
The Bakgaard sounds much better, but 24dB plays louder with better driver protection.
http://164.195.100.11/netacgi/nph-P...0&s1=4031321.WKU.&OS=PN/4031321&RS=PN/4031321
The Bakgaard was also written up in the JAES. The first volume of the Loudspeaker Anthologies has this, the 24dB article by Linkwitz, and many others.
The phase shift is what makes the crossover work.
The 'power loss' is only from the DCR of the inductor, and is usually trivial.
12dB and 18dB crossovers usually don't sound right no matter how you phase the drivers.
24dB can sound very good.
The crossover is the sum of the acoustic response of the drivers and the electrical response of the filters.
One of the best sounding small speakers that I have heard was a 6" two way with a 1" tweeter, and a 24dB crossover at 2Khz.
Electrically the low pass feeding the 6" was 12dB, the high pass feeding the 1" was 18dB. The power response of the 6" was falling at 12dB above 2Khz so the combined roll off was 24dB/oct. The Q of the 12dB low pass filter was adjusted so the combined response was -6dB at 2Khz, IOW Linkwitz Riley. The power response of the tweeter rolled off below 2Khz at 6dB so the combined roll off was 24dB. The Q of the 18dB high pass filter was adjusted so the combined response was -6dB at 2Khz.
A short conical waveguide was used with the tweeter for two reasons. One was to limit the directivity of the tweeter at the crossover point so that it matched the directivity of the woofer closer. The second was to get the voice coil of the tweeter closer to the plane of the woofer.
The HF driver was connected reverse phase through an L-pad and adjusted for maximum notch at 2Khz. The L-pad was replace with high quality fixed resistors and the tweeter was wired in normal phase.
At the time I did this, ca 1988, there was nothing commercially available similar to this. Today this is old hat.
6dB can sound good too.
But it will probably have to be a three way to have any real output capability.
A Bakgaard two way has the best transient response but is quite complex. With the phase link driver and the extra crossover parts involved the cost and complexity will be on the order of a three way.
I used to use a Bakgaard crossover for PA, and switched to 24dB for the largest venues.
The Bakgaard sounds much better, but 24dB plays louder with better driver protection.
http://164.195.100.11/netacgi/nph-P...0&s1=4031321.WKU.&OS=PN/4031321&RS=PN/4031321
The Bakgaard was also written up in the JAES. The first volume of the Loudspeaker Anthologies has this, the 24dB article by Linkwitz, and many others.
I am sure it sounded great. This reminds me of the early 90's that I worked on a similar concept, harnessing the driver's natural roll-of and combining it with the electrical crossover. I was using LMS then.
6dB is fine for home listening, but for PA, I agree 24dB is best. In 'live' applications, the power levels and dynamic range is very different from home hi-fi. I was once shown a Peavy 15" cone torn cleanly from the center to the surround--amazing.
Hi,
Phase and delay are not the same, although they are most time related to each other. Concerning delay: What counts is that all frequencies of interest are delayed the same amount in the total sum-up of the various branches. A 1st order is the easiest way to get there but it is not the only way. With some effort it is possible to get there with higher order filters but still then it works by wasting acoustic energy.
1st order filters have also many disadvantages. They severely limit power handling of the drivers. And they give an undesirable radiation pattern from non-coincident drivers, although the M-T-M configuration relaxes this somewhat.
Phase and delay are not the same, although they are most time related to each other. Concerning delay: What counts is that all frequencies of interest are delayed the same amount in the total sum-up of the various branches. A 1st order is the easiest way to get there but it is not the only way. With some effort it is possible to get there with higher order filters but still then it works by wasting acoustic energy.
1st order filters have also many disadvantages. They severely limit power handling of the drivers. And they give an undesirable radiation pattern from non-coincident drivers, although the M-T-M configuration relaxes this somewhat.
if anyone wants to measure "phase":
I had some boards burned for a DIY Phase-Meter. It's based upon an Intersil application note, uses a couple current-feedback opamps to feed a high speed comparator. The output voltage is a linear function of phase. Since the Intersil specified opamps are not easy to obtain, I used a Texas Instruments part. If you don't need "high speed" you can substitute parts, in the configuration shown, however, it should work to 10MHz . (btw, the TI and Intersil parts have a "disable" function which reduces the amount of switching greatly.) If anyone needs boards contact me at jack@tech-diy.com
Here's the schematic as it appears on Intersil's site.
I had some boards burned for a DIY Phase-Meter. It's based upon an Intersil application note, uses a couple current-feedback opamps to feed a high speed comparator. The output voltage is a linear function of phase. Since the Intersil specified opamps are not easy to obtain, I used a Texas Instruments part. If you don't need "high speed" you can substitute parts, in the configuration shown, however, it should work to 10MHz . (btw, the TI and Intersil parts have a "disable" function which reduces the amount of switching greatly.) If anyone needs boards contact me at jack@tech-diy.com
Here's the schematic as it appears on Intersil's site.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.