Electronic x-over ... Ideas

[a.wayne]

Hello Gents,

Looking to do an 2 or 3 way active x-over, not against using op-amps but would prefer not to. Looking for, Variable gain , frequency and slopes 6/12/18/24db for each section any suggestions as to schematics , Kit, or brand available to look at .....


Regards,

 

[lineup]

Linkwitz is a real authority on loudspeaker filters.
At this link you will find some filter circuits:
Active Filters

Of course he is using OP-amps. Like most do for Active filters.

12 dB Linkwitz-Riley
12 dB/oct Linkwitz-Riley crossover is perhaps the best and most used crossover filter. It has got some very nice qualities.

 

[Conrad Hoffman]

I've been using Marchand boards for many years and am totally satisfied with them. IMHO, the requirements of higher order filters make opamps the only way to go and there's no shortage of excellent opamps these days.

 

[a.wayne]

Thanks gents , I will give both sources a look.....

Conrad , I used to use marchand stuff , wow , decades ago i guess , never thought of looking that way again. I need to have variable slope 6/12/18 I'm not aware of marchand doing such, his and most others are fixed at 12db or 24 db.

regards,

 

[Calvin]

Hi,

the easiest way to replace OPamps by discrete transistors is to use filters whith buffers, i.e.unity-gain-Sallen-Key filters. I use ccs-loaded sourcefollowers made from NJFETs. Here a pair of matched transistors replaces the OPamps. To my ears those filters sound much better than any OPamp filter.
Actually I don´t really understand the need for different paths (2- or 3-way) and different filter steepness.
If You intend to have a ´don´t worry about it´-filter for test purposes I´d suggest to build with OPamps. If You intend a filter for one certain speaker or a final version You could take something different into consideration.
Analog active filters are almost always a series of cascaded building blocks.
I nearly always build my filters as single blocks on small PCBs which connect to a motherboard that holds the power supply and an array of paralleled multi-pin-connectors. Each connector may hold one distinct filter- or equalizer-block. The number of connectors per stereo-channel depends on the degree of into how many different functions You split up the signal path. Channel-wise the connectors are mostly contacted in parrallel fashion in a kind of Bus-system. A motherboard for a stereo-2-way-CR could e.g hold 4 or more multipin-connectors. Small PCBs holding the filter-circuitry and/or EQs can be pushed into the connectors. This way there´s a lot of flexibility. Testing different flter topologies and parts is easily accomplished.

This design has the great advantage that EQs may be included.
Nearly each and every speaker needs equing. While this equing is typically incorporated into a passive filter (no textbook-designs) of a loudspeaker, it is often not the case with active filters.
But if a speaker needs equing the most elaborate and theoretically well designed filter topology is off of the optimum and as such rather low usabilty.
You could add dedicated EQ-stages and this has been done quite regularly so and lead to CRs with excessively high OPamp- and parts-number counts and crappy sound. Or You do something similar to the passive filter designer and choose Your own filter which incorporates the filter and the eq at once. The latter way means less parts number count and imho better sound.
Simulation programs like SwitcherCAD et al are Your powerful friend here ;-)
Anyway, using a motherboard You could test different topologies and strategies easily. It is a bit more of effort in the beginning but soon pays back, because it´s one of the easiest ways to change something and to compare results.

jauu
Calvin

 

[Baldin]

Hi Calvin

Great idae with a mother boadr and an a number og building blocks.
I'm using a similar approach with my class-d amp, as I want to try out different topologies etc.

Could you show a picture of the board??

Guess it's configured as a series chain of "slots" for filter blocks. How do you handle "slots" which is not used?
I would thing a "PC style" jumper would be the way to go .....

Best regards Baldin

 

[AndrewT]

It has all been done for you.
http://www.diyaudio.com/forums/group-buys/53152-mox-active-crossover-buy.html#post593224
and there are a few preceding and follow up threads with simpler and more complex versions .

 

[Calvin]

Hi,

depending on the needed signals I usually use the DIN41617 21-pin or 31-pin connectors (Assmann). Every pin of a connector gets a certain signal asigned, like powersuppl + and minus, gnd, HP-signal-out, etc, etc. The same pin numbers of each connector is asigned to the same signal. All signals are just routed parallel through all connectors (Signal-in and Signal-out may be different if the free-pin count number becomes too small). This way the position and kind of a filter-PCB maybe totally free. Usually I assign the Input- and output-PCB to certain connectors and leave the rest of the connectors free to choose.

jauu
Calvin

 

[a.wayne]

Hi,

the easiest way to replace OPamps by discrete transistors is to use filters whith buffers, i.e.unity-gain-Sallen-Key filters. I use ccs-loaded sourcefollowers made from NJFETs. Here a pair of matched transistors replaces the OPamps. To my ears those filters sound much better than any OPamp filter.
Actually I don´t really understand the need for different paths (2- or 3-way) and different filter steepness.
If You intend to have a ´don´t worry about it´-filter for test purposes I´d suggest to build with OPamps. If You intend a filter for one certain speaker or a final version You could take something different into consideration.
Analog active filters are almost always a series of cascaded building blocks.
I nearly always build my filters as single blocks on small PCBs which connect to a motherboard that holds the power supply and an array of paralleled multi-pin-connectors. Each connector may hold one distinct filter- or equalizer-block. The number of connectors per stereo-channel depends on the degree of into how many different functions You split up the signal path. Channel-wise the connectors are mostly contacted in parrallel fashion in a kind of Bus-system. A motherboard for a stereo-2-way-CR could e.g hold 4 or more multipin-connectors. Small PCBs holding the filter-circuitry and/or EQs can be pushed into the connectors. This way there´s a lot of flexibility. Testing different flter topologies and parts is easily accomplished.

This design has the great advantage that EQs may be included.
Nearly each and every speaker needs equing. While this equing is typically incorporated into a passive filter (no textbook-designs) of a loudspeaker, it is often not the case with active filters.
But if a speaker needs equing the most elaborate and theoretically well designed filter topology is off of the optimum and as such rather low usabilty.
You could add dedicated EQ-stages and this has been done quite regularly so and lead to CRs with excessively high OPamp- and parts-number counts and crappy sound. Or You do something similar to the passive filter designer and choose Your own filter which incorporates the filter and the eq at once. The latter way means less parts number count and imho better sound.
Simulation programs like SwitcherCAD et al are Your powerful friend here ;-)
Anyway, using a motherboard You could test different topologies and strategies easily. It is a bit more of effort in the beginning but soon pays back, because it´s one of the easiest ways to change something and to compare results.

jauu
Calvin

Hello Calvin,

An unity gain discrete type filter is exactly what i was looking for , i had one years a go with variable slopes and freq. I would use op-amps only if no choice , but marchand has a fully passive setup that is interesting .....Should be easy peasey to change and have the 6/12db slopes

XM46 Passive Line Level Crossover Network, PLLXO, 24 dB/oct

It has all been done for you.
http://www.diyaudio.com/forums/group-buys/53152-mox-active-crossover-buy.html#post593224
and there are a few preceding and follow up threads with simpler and more complex versions .

Thanks for the link Andrew , this one might work ................

 

[BobEllis]

I have 6 MOX boards available if you're interested

 

[a.wayne]

Spoke to Phil Marchand and he recommends his XM44, it has the features i'm looking for with variable slope and all ....

Nice piece of Kit :

XM44 Electronic Crossover Network 2-way, 3-way or 4-way, 24 dB/oct or 48 dB/oct

Not sure how this compares to the MOX, just starting to give that one a look over , but before i go off listening to Op-amps 😛 i would like to give the passive deal another shot ! so i'm still looking at that route , simple RLC circuit in the front end of the amplifiers is what's in mind ...


cheers ....