Active Subtractive XOs

Jump to Latest
#101
Hi Charles,

I hadn’t seen Pass's paper before. I didn't read it closely but after looking at you plots I searched his site for it as the link you have on page one didn't work. Does he give any reference to Small? Anyway, it seems the mixed order crossover he presents as a subtracted active, as you have plotted, is the same thing I came up with from the transfer function approach.

I don't think the circuits themselves, as pass gives them, are all that useful. The transfer functions should not be used as the filter response but for the acoustic targets. I guess that has been covered somewhere in this thread as well?

Anyway, looking back it seems I just sort of got to the same place you are through a different path.

But I still think there are some twists on TP design that I haven't seen discussed. I'll bring them up during the ICTA development. But you do seem to be ahead of the game.
 
#103
phase_accurate said:


The steeper slopes may well reduce IMD and FMD but it is open to what extent this advantage is again given up by the increased power demand around the crossover frequency.
Click to expand...


I guess maybe the Pass paper isn't really the same thing what I was lookign at as it seems limited to a few asymmetric crossovers. Quoting form the Pass paper, "Unfortunately I don't know of a way the complimentary output can be made toroll off at a slope greater than 6dB/octive..."

The 2nd order LP, 3rd order HP crossover I was looking at can be tuned to have less peaking;

An externally hosted image should be here but it was not working when we last tested it.


at the expensive of the initial slope of the HP roll off. I chose the peaking level for the ICTA based on the need to have the 3rd order HP roll kick in as quickly as possible without overly extreme peaking. The requirements for the ICTA are somewhat different since there is a need for the 3rd order roll off of the mids in the open baffle format. I may actually go to a symmetric 3rd order.

An externally hosted image should be here but it was not working when we last tested it.


I also prefer the higher order roll off on the woofer since it is difficult to find a driver suitable for a dipole woofer which would work well with a 1st order acoustiic LP response.

Power requirements are greater but I certainly don't want to be blowing the cones out of the mids.
 
#104
Power requirements are greater but I certainly don't want to be blowing the cones out of the mids.
Click to expand...

Hi John

Keep in mind that you might have to take the phase response of the woofer rolloff into account as well in order to achieve a good (or maybe an even better than an already good) SPL summation. But this is not your foe but actually your friend as you can see here with an older verison of my crossover:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=104958&highlight=

BTW:The idea for the "truncation" of the lowpass is originating form one of your papers. :cheers: Altough I use a different approach in the meantime to take the woofer's "upper rolloff" into account.

Regards

Charles
 
#105
phase_accurate said:


Hi John

Keep in mind that you might have to take the phase response of the woofer rolloff into account as well in order to achieve a good (or maybe an even better than an already good) SPL summation. But this is not your foe but actually your friend as you can see here with an older verison of my crossover:
Click to expand...

Well that's the trick, isn't it. If you want the woofer and mid to sum correctly then all the contributions to the phase response as well as amplitude response need to be accounted for. Everything has to be consistent. There are a number of ways to do that using pole shifting and/or all pass filters. Just book keeping. Makes me wonder about all those so called TP systems that use 1st order x-o with nothing else.
 
#106
john k... said:
Well that's the trick, isn't it.
Click to expand...

Indeed, and that is the fatal flaw of the crossovers without the
context of the drivers. I have spent the last year playing with
versions of subtractive crossovers for woofers in baffles for
FR drivers without what I would call a general satisfactory
result. It's easier to simply fool around with filters until you get
what you like...

😎
 
#107
Nelson Pass said:


Indeed, and that is the fatal flaw of the crossovers without the
context of the drivers. I have spent the last year playing with
versions of subtractive crossovers for woofers in baffles for
FR drivers without what I would call a general satisfactory
result. It's easier to simply fool around with filters until you get
what you like...

😎
Click to expand...

Hi Nelson,

You may want to keep an eye out on my ICTA development. I've been doing the transient perfect thing for, well before the amp under the table was produced. Look familiar? Works flawless after 30 years (a couple of tune up along the way.)

An externally hosted image should be here but it was not working when we last tested it.
 
#108
John that does indeed look cool !

I have spent the last year playing with versions of subtractive crossovers for woofers in baffles for FR drivers without what I would call a general satisfactory result.
Click to expand...

Well it should at least be possible to get frequency response and step response correct on-axis. Off axis one has to rely on driver configurations like John did if you have to use steep crossovers.
If the drivers are less picky then non-symmetrical FR and woofer topologies are possible.
The two main parameters that must be fulfilled to do so are: The crossover frequency must be where the wavelength is large compared to driver distance. And the drivers must be able to cope with flat crossover slopes.

In this case it is easy to try what I posted in the following thread:

http://www.diyaudio.com/forums/showthread.php?s=&threadid=104958&highlight=

My latest topology is again some type of subtractive crossover that contains some compensation tricks.
But in the end one has to achieve an overall "equal numerator and denominator" function in order to work properly.

Regards

Charles
 
#109
Nelson Pass said:
Indeed, and that is the fatal flaw of the crossovers without the
context of the drivers. I have spent the last year playing with
versions of subtractive crossovers for woofers in baffles for
FR drivers without what I would call a general satisfactory
result. It's easier to simply fool around with filters until you get
what you like...

😎
Click to expand...

the attached PDF (first page) shows an approach for a midrange bandpass filter, which seems to be a mixtur between an active subtractive XO and two 12 db high pass stages.
1) IC1 and IC2 forms together with C401/301 and C402/302 together with VR202 a variable high pass filter stage according Sallen-Key topology (50-545Hz)
2) IC5/3 and IC5/4 together with C403/303 and C404/304 so as VR203 fulfills the same purpose but with a range for adjusting between 250 Hz and 7200 Hz.

Now to the mixer IC4/3:
The inverted mixer input PIN 13/9 receive the signals of both mentioned high pass stages
The non inverted input PIN 12/10 don't receive a full range signal as usual on subtractive XO's. Instead this it receive a highpass filtered signal of only 6db slope from high pass filter under 2) from PIN 8/14 of IC5/3.

what does this special approach do ?

BTW - the output PIN 14/8 of this mixer fits both the levelable line stage IC7 and the high pass filter under 2) itself

Thank you for an information.

P.S.: under
Subtractive Crossover Networks
and
Subtractive Cross-overs
dont provide the wanted advice.

This topology I have found in the HiFonics Callisto VIII - go to
Hifonics Callisto VIII dual 3 way/4 way electronic crossover - schematic wanted
 

Attachments

#113
A subtractive crossover is one where one or more output signals are derived by subtraction of the output signal(s) of different transfer funcions from each other or from the input signal.
The simplest version is the constant voltage crossover where the lowpass output is achieved by subtraction of the highpass-filtered input signal from the unfiltered input signal (and vice versa).

Regards

Charles
 
You must log in or register to reply here.
Top Bottom