This was based on the ESP (rod elliott) design and he pretty much advocates this way. But I guess it could be done either way.
Among the many simulations that I've done, there has to be one that was done that way. I didn't compare that, but I guess that needs to be done.
What I compared more is the supposedly "phase-linear" scheme, with the subtractive method and the all-pass/delay cells, and I fail to see any advantages from the sims. I don't understand why the final result doesn't really show anything different regarding the phase. So I don't think in the end, there is any real advantage to this phase-linear approach and it only costs more in parts, opamps and whatever.
I will try the approach of splitting in 2 first and then split that again afterwards.
The "ESP" way, the high channel only has one filter cell in the path, and the others have more. If we split in 2 and then 2 again, all channels have 2 filter cells in the path, so not sure this is better, as the most susceptible to noise is the highest channel, and having only one cell definitely helps minimizing the effects.
Modularizing is definitely something to do, but not by multiplying the number of pcbs and not by adding more wiring between all that stuff.
I prefer minimizing all wiring as much as possible, so I aim to put everything on a single pcb, psu and all, including connectors and all that. So when doing assembly, the bulk of the work is just stuffing/soldering on the pcb, and then the rest is mostly mechanical, with hardly any wires.
With the input/output connectors, power plug and all electronics on a single pcb, we can arrange to have virtually zero external wiring.
And for modular design, I've been thinking about how to make the parts of the filter cells, all fit on small modules that plug into the main pcb (no wires). With the caps selected first and then the resistor values calculated, only those resistors change with different xover frequencies, so those can fairly easily be put on small modules, that can be interchangeable to change the xover points.
I've been digging for possible connectors to make it reliable yet as cheap as possible, and as much DIY doable as possible.
I looked into having small pcbs with the pcb edges as connectors, but the female connectors to receive them are a bit pricy. I also looked at plain headers, through hole/board, so the small pcb modules could plug into the main pcb from above, with the pcbs parallel. That could be a way to go, as it's cheap, easy and those types of connectors can (should) be found anywhere on this planet.
But finally, I think it might be even simpler, since only resistors need to be put on interchangeable modules, to put them on an IC DIL socket, so all that's needed on the main pcb are such IC sockets, and the resistors would be on their own sockets/modules that would plug into that. This would make it quite modular and configurable, and make it DIY doable anywhere. By "anyone", including myself, who can't tackle SMT.
Actually for the xover, I wasn't thinking (yet) about microcontrollers, but since the ultimate goal is to fully integrate this into a much more complex system, using a microcontroller is a must.
For the testing, that sounds good. I didn't plan on having muting on the xover, because I was thinking more about the larger project that will have other muting elsewhere. But now that I'm thinking about making this xover available as a stand alone device, this could be added to make it a full blown device.
Having looked at many commercial offerings, and how they put forth some specs that I find are likely exaggerated, in view of how many circuits and especially the large number of coupling caps along the signal's path. I just don't see how they can reconcile their specs for noise, and especially distortion, with so much stuff adding noise and the numerous coupling caps adding loads of distortion. They must be lying.
I want to minimize what's in the signal's path, especially the coupling caps.
The ready made xovers are most expensive and are stand alone devices. They add to the wiring and they're not as good as they could be.
They use design schemes that don't favor the sonics and performance. At least most of them, from what I've seen.
I'm planning to made this design as a stand alone device for DIY sharing, but using the principals used towards the larger goal, and that could never be achieved with anything on the market.
At the moment, I can't afford getting all kinds of gear to do all the measuring and pre-design stuff. I can't get all that stuff like the miniDSP, plus that microphone, plus, plus, plus.... It's too much..
Once designed, what was learned will serve for the larger goal, and in the mean time, I'll make this into a usable and shareable DIY project, that others can build, regardless of what country they're located in.
So many can't build all they want because they can't find certain things, or those things are too expensive where they are, so keeping it simple helps this.
In any case, I don't see why anyone needs to get all bent out of shape over xovers, because this has been done for many years, with success, although perhaps not to today's standards, but still, it worked before, so why not now? Why say it won't work or can't be done?
When some people go to some store, buy some speakers, some amps, some xovers, some equalizer and whatever else, they get this put together and it works. And they don't go through all the soul searching...
And besides, this isn't so different from previous projects, like for example the one Jens Rassmussen was making many years ago. It was just 2way, and then later 3way, and many people were interested enough to have a group buy for that. And not that long ago, there were still some inquiring about getting more of those pcbs. But Jens pretty much disappeared from the forums, including his own web site and anything that used to be available back then. So that's basically gone now.
Interest still exists, and not many are able to go through buying all that gear to do it "right". Of course, eventually, having that measurement microphone and using a miniDSP might be a huge improvement to find out what is needed, measure the speakers and whatever else to find the right xover frequencies and all that, but in the mean time, the simple xover can still be made to work, be useful, and affordable.
I haven't seen any designs on the forums of any 4way xovers like that, so I'm working on one, because it can then be used in my own larger plan.
It's too bad Jens has disappeared like that, he was doing great work.
I'm looking into what would be best as an adjustable delay. I'm simulating a few, but I wonder if someone has some good ideas about this out there.
I'm doing one with 2 opamps, but that has a 2gang pot, and that would be a good thing to avoid such 2gang pots, if possible. Anyone has any good ideas about that?
The delay would be best if it could go both ways, ahead or behind, to suit situations where the physical speaker alignments are way off. For example, most hi-fi speakers have their tweeters "ahead" of the rest, but in my 4way speaker system, my compression horn would be the one being far behind everything else, while the tweeter would be the farthest ahead and the low end speakers somewhere in between..
Being able to compensate for the phase can partially be done by moving some speakers, but then it looks weird, so doing it at the xover is better. It needs to be able to go both ways, ahead and behind of the 0degrees.
I'm doing one with 2 opamps, but that has a 2gang pot, and that would be a good thing to avoid such 2gang pots, if possible. Anyone has any good ideas about that?
The delay would be best if it could go both ways, ahead or behind, to suit situations where the physical speaker alignments are way off. For example, most hi-fi speakers have their tweeters "ahead" of the rest, but in my 4way speaker system, my compression horn would be the one being far behind everything else, while the tweeter would be the farthest ahead and the low end speakers somewhere in between..
Being able to compensate for the phase can partially be done by moving some speakers, but then it looks weird, so doing it at the xover is better. It needs to be able to go both ways, ahead and behind of the 0degrees.
Alas you are simply wrong - analog filters work by having lots of capacitors in the signal path, its absolutely fundamental. Use the wrong sort of capacitor, yes, you'll have capacitor distortion, but use polyproylene or polystyrene and you don't (come to that use the wrong sort of resistor and you have resistor distortion). DC blocking caps, if generously sized, don't have to be anything special for ultra low distortion, as there are only mV of signal voltage across them. Even 1% distortion on a 10mV signal across a DC blocking cap is equivalent to 0.001% distortion at full signal swing, and capacitor distortions at mV levels are always way below that level anyway, even cheap electrolytics, so a cheap electrolytic can easily give immeasurable distortion as a DC-blocker.
You can put a signal through 100 opamps and capacitors and have it come out fine, if you understand the properties of the devices and chose the right ones. Basically that's what an old analog mixing desk did...
As for noise, simply raise the signal level up in the first stage and make sure the first stage is low-noise - problem basically solved.
The idea that reducing the number of components in the signal path improves things is completely flawed - no single transistor amp stage, for instance, can compete against a good opamp for linearity. The best opamps can get 0.000015% distortion. 1.5 parts per million. There is no substitute for cunning in getting performance from a circuit.
I wasn't referring to those necessary caps. I saw so many designs, mostly commercial offerings, with far too many dc blocking caps, all over the place, and they're not really even that big, so I suspect they do introduce much more distortion than they should.
In one of those designs, I counted 5 dc blocking caps, which were all after the filter cells, and I think there must be a way to limit that to maybe just one.
I really don't see how this works there. 1% is 1%, no matter what the signal level is. When amplified, whatever distortion is there, is also amplified, and 1% is still 1%. I don't get it.
Plus whatever a stage adds as distortion, it adds to whatever was previously added, and this happens all the way along the whole chain from end to end, so anything that introduces distortion into the signal's path, gets compounded. It adds up. I don't see how it could be otherwise.
0.000015% would be more like 0.15ppm, not 1.5
Which opamp would that be?
I think it can be very helpful, with the right microphone, and hooked up to a computer with the appropriate software.
Eventually I'll have to try that out. It's not super cheap, and it's not the only thing to get, so unless you have a big enough wallet, it's not that easy to come by for some of us.
Especially since most of that stuff is to be used only for measurement and testing, but then be shelved once it's all done. It's still quite a bit of money sleeping in gear that isn't used very often.
Basically all that really needs to be decided in the design of the xover, is the xover frequencies, signal levels, and delays. I wouldn't bother with things like linkwitz transform, which is supposed to be aimed only at closed boxes, and other things like those shelving or notch/peaks or whatever.
I think if an equalizer is used, there should be hardly any need for notches/peaks and shelving stuff.
Besides, it's also dependent a lot on the type of speakers, room/venue, and preferences.
I don't deal with the kind of speakers used in living room hifi, and when a sound system needs to be able to be used in many places, and outdoors eventually, then anything pertaining to a specific room is pointless.
So basically, what I want in the xover, is choosing the xover frequencies, adjusting delays and levels, and that's pretty much it.
I think xovers and the amps should be right next to the speakers, with the shortest speaker cables possible, and then a balanced link back to the mixer/equalizer, which in the kind of setting that I'm looking at, could be at least a distance of several meters, potentially 10-20meters, or even more. So a balanced link is a must. The link from the xover to the amps don't really need to be balanced if they're all together, like in a rack or some stack, with very short cables.
I think if an equalizer is used, there should be hardly any need for notches/peaks and shelving stuff.
Besides, it's also dependent a lot on the type of speakers, room/venue, and preferences.
I don't deal with the kind of speakers used in living room hifi, and when a sound system needs to be able to be used in many places, and outdoors eventually, then anything pertaining to a specific room is pointless.
So basically, what I want in the xover, is choosing the xover frequencies, adjusting delays and levels, and that's pretty much it.
I think xovers and the amps should be right next to the speakers, with the shortest speaker cables possible, and then a balanced link back to the mixer/equalizer, which in the kind of setting that I'm looking at, could be at least a distance of several meters, potentially 10-20meters, or even more. So a balanced link is a must. The link from the xover to the amps don't really need to be balanced if they're all together, like in a rack or some stack, with very short cables.
I really don't know enough about how that minidsp works and how this works with a microphone hooked to a computer.
There should be a pink noise source, and I'm wondering how that's generated.
Where to inject the pink noise also needs to be properly considered. Should be at least before equalization, to make sure the equalizer is in the chain.
Roughly, my overall scheme would be something like on that attached diagram.
The longest link would be between the equalizer and the xover, which would have to be balanced.
If the pink noise source also comes from the computer that runs the analyzer software, then that requires that computer to be fairly near the equalizer/mixer, and that could be pretty far from the xover and speakers, so there are considerations to take into account there.
There should be a pink noise source, and I'm wondering how that's generated.
Where to inject the pink noise also needs to be properly considered. Should be at least before equalization, to make sure the equalizer is in the chain.
Roughly, my overall scheme would be something like on that attached diagram.
The longest link would be between the equalizer and the xover, which would have to be balanced.
If the pink noise source also comes from the computer that runs the analyzer software, then that requires that computer to be fairly near the equalizer/mixer, and that could be pretty far from the xover and speakers, so there are considerations to take into account there.
Attachments
Yes, you don't get it.
Blocking cap has 10mV across it, signal through it is 1V, capacitor distortion can only affect the 10mV across the cap, not the 1V signal going through it. Blocking cap is in a divider circuit with the load, so long as the cap's impedance is much lower than the load then its effect is much smaller, same as for any impedance in the signal path.
Well, in any case, I think minimizing the coupling caps as much as possible is a better way. Using them only where strictly required and making them as big as practically possible.
I finally made an alternate arrangement sim for the 4 way xover that I posted about earlier, as it's proposed by Rod Elliott, for comparison.
I first split in two, then each half split again, instead of the cascading scheme he proposes.
And I ran the same sim on each and put both plots next to each other for better comparison (attached). The plot on the left is the ESP proposed arrangement.
I'm not really sure it would be any better by splitting this way as opposed to cascaded.
There is a larger dip in the summed output with the split scheme, and that dip happens at a bit lower frequency. The lowest point in that dipped output summed response is somewhere near 6.5khz with the ESP cascaded scheme, while that deeper dip with the split scheme is at a little less than 1.8khz, with almost 0.25db of dipping with the split scheme, compared to cascaded.
The phase response looks somewhat similar, except maybe a tad more chaotic with the split scheme.
I don't think splitting this way brings much improvement, if any, except perhaps for the fact that all 4 bands have the same exact number of filter cells in the signal's path.
The cascaded scheme leaves only a single filter cell in the signal path for the highest band, which would be the one most affected by having more circuitry involved.
Weighing the pros and cons...
I finally made an alternate arrangement sim for the 4 way xover that I posted about earlier, as it's proposed by Rod Elliott, for comparison.
I first split in two, then each half split again, instead of the cascading scheme he proposes.
And I ran the same sim on each and put both plots next to each other for better comparison (attached). The plot on the left is the ESP proposed arrangement.
I'm not really sure it would be any better by splitting this way as opposed to cascaded.
There is a larger dip in the summed output with the split scheme, and that dip happens at a bit lower frequency. The lowest point in that dipped output summed response is somewhere near 6.5khz with the ESP cascaded scheme, while that deeper dip with the split scheme is at a little less than 1.8khz, with almost 0.25db of dipping with the split scheme, compared to cascaded.
The phase response looks somewhat similar, except maybe a tad more chaotic with the split scheme.
I don't think splitting this way brings much improvement, if any, except perhaps for the fact that all 4 bands have the same exact number of filter cells in the signal's path.
The cascaded scheme leaves only a single filter cell in the signal path for the highest band, which would be the one most affected by having more circuitry involved.
Weighing the pros and cons...
Attachments
Having compared the behavior between the ESP proposed cascaded arrangement vs the split twice arrangement for the filter cells, in view of how much deviation there is from the flat response, the ESP cascaded version is flatter, so I retained that option.
I am putting the resistors setting the xover frequencies on component sockets, which in turn are on IC sockets, so easily interchangeable in case a different xover frequency needs to be used.
I put all-pass/delays and inverter that can be disabled with jumpers, in case they're needed or not, and their frequency setting parts are also on component sockets in case they need changing, plus those sockets can be rotated to change their delay/leading action.
After that, before outputting via balanced line drivers, the levels are adjustable.
So, xover frequencies can be chosen, delay/leading/inverting can be added, or not, and level adjusted on each channel. So it can be adjusted as needed, from measurements.
I decided to xover at 120Hz instead of 150Hz for the Lo channel, but that can easily be changed as needed.
Having the parts on component sockets allows putting in whatever we want and calculated. If measurements require something different, it's easy to implement.
Using XLR PCB mounted also eliminate any wiring, as well as using a PCB mounted Power Inlet module and the transformer, so everything can all be fitted on a single PCB with zero lose wiring.
I entered the schematics in Eagle and the attachement is the current schematic for the filters. The PSU and all decoupling caps are on a second sch sheet, which I'll post later.
I've been thinking about what would be required at the input before the balanced line receivers, and I don't see what I could add that wouldn't cause some side effects, except perhaps for some kind of EMI filter, maybe with ferrite beads and some kind of caps that wrap around the lines...
I haven't found the right type of such EMI input filter, so for now I left the balanced line going straight into the line receiver.
Just in case, I put both a male and female XLR input plug, which allows a pass through.
I thought I'd try a decent PSU design with all discrete stuff, and simulated one that works nicely with a good ripple rejection, better I think than IC regulators like the 78xx/79xx or even LM317/337. But the part count was a bit high and I wanted to add a tracking of the voltages so both polarities would mirror each other as much as possible, so I ended up going back to the LM317/337 option, with an opamp added to track the negative side to the positive, so it's not too bad I think.
Power requirements are pretty low, and I picked a 5VA toroidal transformer for the PSU, PCB mounted. It takes a small amount of PCB real estate to have it all on a single PCB, but then there are no external wires whatsoever afterwards, no even the power cord, which plugs right into the power inlet module, which has a fuse, a switch and an EMI filter. I opted for the Schurter DD22.1121.1111 inlet module, which I wish could have a lighted switch and input voltage selection, but at least it does have most of the wanted features.
The inlet module uses the standard C14 plugs, so anyone anywhere in the world can use the appropriate power cord with the plugs in use locally. That module is rated for 1Amp, which seems low but not for the purpose here, as with a 5VA transformer, we're nowhere near that.
The PSU has 2 trimmers, one to adjust the +15V, which gets tracked automatically to the negative side, and one to adjust the tracking so it can be properly balanced to equal voltages on each side.
I'm inserting jumpers at the PSU +-15V outputs to disconnect the PSU from the rest while it's being adjusted, so at assembly time, before the first power up, those jumpers shouldn't be fitted before the proper adjustments have been made with the trimmers. A simple operation with a digital meter.
I'm planning everything to fit with space to spare on a PCB sized at 150x300mm. I plan to add an automatic muting, to prevent any plops or whatever at power up or down, but that remains to be designed and added.
One of the things the muting system would look for could be the proper balance of the supply voltages, so as to prevent unmuting if something isn't right there. Maybe that could be visualized with small LEDs, along with LEDs indicating when a channel is muted.
The muting system could have jumpers to force channels to be muted, so while doing measurements and adjustments, only the channel chosen can be operated.
Still a work in progress, but getting there now...
I am putting the resistors setting the xover frequencies on component sockets, which in turn are on IC sockets, so easily interchangeable in case a different xover frequency needs to be used.
I put all-pass/delays and inverter that can be disabled with jumpers, in case they're needed or not, and their frequency setting parts are also on component sockets in case they need changing, plus those sockets can be rotated to change their delay/leading action.
After that, before outputting via balanced line drivers, the levels are adjustable.
So, xover frequencies can be chosen, delay/leading/inverting can be added, or not, and level adjusted on each channel. So it can be adjusted as needed, from measurements.
I decided to xover at 120Hz instead of 150Hz for the Lo channel, but that can easily be changed as needed.
Having the parts on component sockets allows putting in whatever we want and calculated. If measurements require something different, it's easy to implement.
Using XLR PCB mounted also eliminate any wiring, as well as using a PCB mounted Power Inlet module and the transformer, so everything can all be fitted on a single PCB with zero lose wiring.
I entered the schematics in Eagle and the attachement is the current schematic for the filters. The PSU and all decoupling caps are on a second sch sheet, which I'll post later.
I've been thinking about what would be required at the input before the balanced line receivers, and I don't see what I could add that wouldn't cause some side effects, except perhaps for some kind of EMI filter, maybe with ferrite beads and some kind of caps that wrap around the lines...
I haven't found the right type of such EMI input filter, so for now I left the balanced line going straight into the line receiver.
Just in case, I put both a male and female XLR input plug, which allows a pass through.
I thought I'd try a decent PSU design with all discrete stuff, and simulated one that works nicely with a good ripple rejection, better I think than IC regulators like the 78xx/79xx or even LM317/337. But the part count was a bit high and I wanted to add a tracking of the voltages so both polarities would mirror each other as much as possible, so I ended up going back to the LM317/337 option, with an opamp added to track the negative side to the positive, so it's not too bad I think.
Power requirements are pretty low, and I picked a 5VA toroidal transformer for the PSU, PCB mounted. It takes a small amount of PCB real estate to have it all on a single PCB, but then there are no external wires whatsoever afterwards, no even the power cord, which plugs right into the power inlet module, which has a fuse, a switch and an EMI filter. I opted for the Schurter DD22.1121.1111 inlet module, which I wish could have a lighted switch and input voltage selection, but at least it does have most of the wanted features.
The inlet module uses the standard C14 plugs, so anyone anywhere in the world can use the appropriate power cord with the plugs in use locally. That module is rated for 1Amp, which seems low but not for the purpose here, as with a 5VA transformer, we're nowhere near that.
The PSU has 2 trimmers, one to adjust the +15V, which gets tracked automatically to the negative side, and one to adjust the tracking so it can be properly balanced to equal voltages on each side.
I'm inserting jumpers at the PSU +-15V outputs to disconnect the PSU from the rest while it's being adjusted, so at assembly time, before the first power up, those jumpers shouldn't be fitted before the proper adjustments have been made with the trimmers. A simple operation with a digital meter.
I'm planning everything to fit with space to spare on a PCB sized at 150x300mm. I plan to add an automatic muting, to prevent any plops or whatever at power up or down, but that remains to be designed and added.
One of the things the muting system would look for could be the proper balance of the supply voltages, so as to prevent unmuting if something isn't right there. Maybe that could be visualized with small LEDs, along with LEDs indicating when a channel is muted.
The muting system could have jumpers to force channels to be muted, so while doing measurements and adjustments, only the channel chosen can be operated.
Still a work in progress, but getting there now...
Here is the PSU schematic.
The tracking opamp can be anything similar, I chose the old faithful 741, but many others have the same footprint and pinout, so wide choice there.
Probably all, if not most, of the parts can be obtained from Mouser, or perhaps Digikey or whatever (Arrow? Future Electronics? Farnell?). I'm hoping nothing would be difficult to get in some countries. And nothing should be overly expensive..
Working on a PCB layout. Everything already on the schematics can easily fit on the single 150x300mm PCB, using the bottom layer as ground plane. The goal being to try to avoid as much as possible to route anything on the ground plane, only very few and very short traces at the most, to avoid breaking that plane.
I've been thinking a small shield might be useful to limit the transformer's influence. It shouldn't have a far reaching influence, so maybe a little distance might suffice, but just in case, I figured adding a small shield would be good to try. Will put out visuals for all this soon...
The tracking opamp can be anything similar, I chose the old faithful 741, but many others have the same footprint and pinout, so wide choice there.
Probably all, if not most, of the parts can be obtained from Mouser, or perhaps Digikey or whatever (Arrow? Future Electronics? Farnell?). I'm hoping nothing would be difficult to get in some countries. And nothing should be overly expensive..
Working on a PCB layout. Everything already on the schematics can easily fit on the single 150x300mm PCB, using the bottom layer as ground plane. The goal being to try to avoid as much as possible to route anything on the ground plane, only very few and very short traces at the most, to avoid breaking that plane.
I've been thinking a small shield might be useful to limit the transformer's influence. It shouldn't have a far reaching influence, so maybe a little distance might suffice, but just in case, I figured adding a small shield would be good to try. Will put out visuals for all this soon...
What kind of speakers was this done with?
I'm not going to bother with shelf and step filters. That stuff could be managed at the equalizer level.
My speakers aren't like the hi-fi type. Definitely not all in a single box and low frequency ones aren't in closed boxes. Quite different from most living room hi-fi stuff.
When the time comes, when I'm able to set something up to be measured, and I can afford getting a minidsp with microphone and software, I'll see who those speakers behave.
For the time being, I can get a usable xover ready for that time.
I updated the schematics, with more info added.
The design is done using chosen caps and then resistors are calculated from there. With combinations of 2 1% or better resistors, the values can get very close to what's calculated. The caps should be chosen with the tightest tolerance, but it's kind of hard to get as close tolerances with caps as we can get with resistors.
Using component sockets to house all the calculated parts allows easy change if needed. So each xover frequency can be adjusted if need be, as well as the time delays/leading.
I gave up on trying to have a VCA or similar system to do the output level control, and just went with the simpler buffered method, with trimmers. Those are adjusted at the time of tweaking and then not moved again, so they're not quite like pots that get turned all the time...
So, xover frequencies and delays are determined by calculated parts on sockets, easily changeable, and each output has level adjustment. That's all that's needed to make it usable and tweakable when measurements can be made.
Small things that would be tweaked using shelf/notch/step would be tweaked with the equalizer, which can also likely do some tweaks for some irregularities due to the xover/speakers combinations, so the xover doesn't need to be more complex that this.
I'm compiling a BOM in excel, which I will post soon after. And I'm working on a layout with eagle. All-on-one pcb, with no lose wires at all, and absolutely no SMT, so it's 100% easy DIY material.
I also made an excel sheet for the xover calculations, with a system to pick and compare the resistor values and combinations, with a percentage deviation from nominal according to calculations. This shows how much closer to calculated values we can get when using different resistor combinations.
The design is done using chosen caps and then resistors are calculated from there. With combinations of 2 1% or better resistors, the values can get very close to what's calculated. The caps should be chosen with the tightest tolerance, but it's kind of hard to get as close tolerances with caps as we can get with resistors.
Using component sockets to house all the calculated parts allows easy change if needed. So each xover frequency can be adjusted if need be, as well as the time delays/leading.
I gave up on trying to have a VCA or similar system to do the output level control, and just went with the simpler buffered method, with trimmers. Those are adjusted at the time of tweaking and then not moved again, so they're not quite like pots that get turned all the time...
So, xover frequencies and delays are determined by calculated parts on sockets, easily changeable, and each output has level adjustment. That's all that's needed to make it usable and tweakable when measurements can be made.
Small things that would be tweaked using shelf/notch/step would be tweaked with the equalizer, which can also likely do some tweaks for some irregularities due to the xover/speakers combinations, so the xover doesn't need to be more complex that this.
I'm compiling a BOM in excel, which I will post soon after. And I'm working on a layout with eagle. All-on-one pcb, with no lose wires at all, and absolutely no SMT, so it's 100% easy DIY material.
I also made an excel sheet for the xover calculations, with a system to pick and compare the resistor values and combinations, with a percentage deviation from nominal according to calculations. This shows how much closer to calculated values we can get when using different resistor combinations.
In any case, even before doing any kind of tweaking from measurements, going with this basic 4 ways with multi-amps beats the heck out of what was done before, which was to not use the mid-lo speakers at all and use a passive filter with only one amp.
No measurements for sure were done and no tweaking whatsoever was possible with those passive filters. Still, as is, it worked just fine, although I'm sure the response could not be very flat. The bare bones very rough setup was functional and used that way for years.
That is until I was gone away for a long time and those speakers (and amps) have been stored away all this time and are pretty much out of easy reach and unusable at the moment. No room to set them up for now. They'll have to be dug out from their buried situation before doing anything with them. I hope no mice or rats have been in there eating away at the drivers...
No measurements for sure were done and no tweaking whatsoever was possible with those passive filters. Still, as is, it worked just fine, although I'm sure the response could not be very flat. The bare bones very rough setup was functional and used that way for years.
That is until I was gone away for a long time and those speakers (and amps) have been stored away all this time and are pretty much out of easy reach and unusable at the moment. No room to set them up for now. They'll have to be dug out from their buried situation before doing anything with them. I hope no mice or rats have been in there eating away at the drivers...
Some views of the board by itself.
I had planned on having a muting system, but ended up not including it for now. There is room on the pcb for extra stuff to do this. Maybe later I'll look into adding that, but for now, hoping the PSU with tracking won't cause too much of a glitch at powerup/down, it'll have to do without a muting system.
Power amps can be powered up after the xover anyway, so glitches can be there and not heard.
I had planned on having a muting system, but ended up not including it for now. There is room on the pcb for extra stuff to do this. Maybe later I'll look into adding that, but for now, hoping the PSU with tracking won't cause too much of a glitch at powerup/down, it'll have to do without a muting system.
Power amps can be powered up after the xover anyway, so glitches can be there and not heard.
