So this project has been going on for quite a while now.
Mostly because time related issues.
Anyway, It all started with the complete frustration that there are basically two things:
Having the full flexibility of something like a DSP is all nice, but often projects simply just don't need all these options.
Yet, at the same time amplifiers seem to be cheap and easily available.
At the same time, there doesn't seem to be anything that could just basically replace a passive filter.
Also, a DSP with good THD+N and SNR can be very expensive.
So I started with this idea in mind what a general and generic (passive) filter would look like.
Basically any filter needs at least:
- shelving filter to compensate for baffle step
as well as;
Also helpful in any active filter system, is a addition 1st or 2nd order HP filter to be able to cross with something like a subwoofer (or low end woofer in a 3-way system).
So with background in such designs and a bit of puzzling, I also managed to make this very versatile to be able to connect or bypass certain sections (filter blocks) or even daisy chain all of the above for even more flexibility.
To make that a bit more visual, here is a block diagram with a couple of ideas;
As well as a schematic;
So basically this will work in a 2-way system or for something like a single system when all functions are needed (like a subwoofer).
With a bit of creativity, other filters or tweaks can be made.
A Linkwitz-Transform isn't quite possible yet, but I am a bit on the edge of just making a little adapter board for that instead.
(it kinda messes up the nice PCB design)
The param EQ is adapted a little bit to just make the design and calculations a bit easier.
(yet still working the same)
- The first stage can also be used as a (very) crude differential (balanced) to single ended amplifier.
- Any HP filter can be replaced for a LP filter or vice versa (just by swapping the capacitors for resistors and resistors for capacitors)
This way a 4th order LP (or HP ) filter can be made (see variation 2)
- In some cases phase can be flipped, which can be very simply fixed, by flipping the polarity of the speaker/driver itself again.
- opamps can be any flavor/price/performance (as long as they are unity gain stable)
PCB size for now 83.5x40mm
incl volt regulators, as well as some SMD parts on the back like bridge rectifier, decoupling etc.
Inspiration was taken from;
https://www.linkwitzlab.com/filters.htm
http://sim.okawa-denshi.jp/en/Fkeisan.htm
https://sound-au.com/articles/active-filters.htm
(and probably a lot more, I forget what I have gathered over the years )
Mostly because time related issues.
Anyway, It all started with the complete frustration that there are basically two things:
- no affordable/cheap alternatives for active filters
- the analog active filter boards that are available never make ANY sense (in my opinion).
Having the full flexibility of something like a DSP is all nice, but often projects simply just don't need all these options.
Yet, at the same time amplifiers seem to be cheap and easily available.
At the same time, there doesn't seem to be anything that could just basically replace a passive filter.
Also, a DSP with good THD+N and SNR can be very expensive.
So I started with this idea in mind what a general and generic (passive) filter would look like.
Basically any filter needs at least:
- shelving filter to compensate for baffle step
- 1st, 2nd or 3rd order HP filter for the tweeter (can also be any other kind of driver obviously )
- A simple parametric EQ for this channel (boosting as well as cutting)
- Gain/attenuation
as well as;
- 1st, 2nd or 3rd order LP filter for a woofer (again, can be something else)
- Also a simple parametric EQ (boosting as well as cutting)
- maybe also gain, just because why not. (technically not needed)
Also helpful in any active filter system, is a addition 1st or 2nd order HP filter to be able to cross with something like a subwoofer (or low end woofer in a 3-way system).
So with background in such designs and a bit of puzzling, I also managed to make this very versatile to be able to connect or bypass certain sections (filter blocks) or even daisy chain all of the above for even more flexibility.
To make that a bit more visual, here is a block diagram with a couple of ideas;
As well as a schematic;
So basically this will work in a 2-way system or for something like a single system when all functions are needed (like a subwoofer).
With a bit of creativity, other filters or tweaks can be made.
A Linkwitz-Transform isn't quite possible yet, but I am a bit on the edge of just making a little adapter board for that instead.
(it kinda messes up the nice PCB design)
The param EQ is adapted a little bit to just make the design and calculations a bit easier.
(yet still working the same)
- The first stage can also be used as a (very) crude differential (balanced) to single ended amplifier.
- Any HP filter can be replaced for a LP filter or vice versa (just by swapping the capacitors for resistors and resistors for capacitors)
This way a 4th order LP (or HP ) filter can be made (see variation 2)
- In some cases phase can be flipped, which can be very simply fixed, by flipping the polarity of the speaker/driver itself again.
- opamps can be any flavor/price/performance (as long as they are unity gain stable)
PCB size for now 83.5x40mm
incl volt regulators, as well as some SMD parts on the back like bridge rectifier, decoupling etc.
Inspiration was taken from;
https://www.linkwitzlab.com/filters.htm
http://sim.okawa-denshi.jp/en/Fkeisan.htm
https://sound-au.com/articles/active-filters.htm
(and probably a lot more, I forget what I have gathered over the years
)
So I got a question from someone how to simulated and calculate the values.
My basic workflow is just simulate your filter design in your favorite speaker filter program (like VituixCAD, LspCAD etc).
Just use standard DSP IIR filter blocks for this.
After this, it's just a matter of translating the parameters with some math.
Luckily there are some filter calculators for these days (as already mentioned in the previous post).
I have been working on some myself, especially because the param EQ is ever so slightly different.
You do have to keep in mind that resistors and especially capacitors have tolerances.
So I would always suggest to take at 1% resistors (which is fine these days).
Getting capacitors with low tolerance is more tricky, especially with the higher values.
Kemet R82 is available in 5% and 10%.
Also, of course there are only that many E value numbers.
So sometimes you have to fiddle around a bit.
I made a spreadsheet/excel sheet that can calculate these things, as well calculates the actual parameters based on the real values.
It's not complete just yet, but will be available shortly
In the end it's a bit going back and forth with filter software program again to get a good match.
This board works with a symmetrical power supply btw, so it can be easily combined with just any amplifier with symmetrical power supply, as long as you don't overload the voltage regulators.
I am working on a little module so this can also be used with a single voltage supply as well.
My basic workflow is just simulate your filter design in your favorite speaker filter program (like VituixCAD, LspCAD etc).
Just use standard DSP IIR filter blocks for this.
After this, it's just a matter of translating the parameters with some math.
Luckily there are some filter calculators for these days (as already mentioned in the previous post).
I have been working on some myself, especially because the param EQ is ever so slightly different.
You do have to keep in mind that resistors and especially capacitors have tolerances.
So I would always suggest to take at 1% resistors (which is fine these days).
Getting capacitors with low tolerance is more tricky, especially with the higher values.
Kemet R82 is available in 5% and 10%.
Also, of course there are only that many E value numbers.
So sometimes you have to fiddle around a bit.
I made a spreadsheet/excel sheet that can calculate these things, as well calculates the actual parameters based on the real values.
It's not complete just yet, but will be available shortly
In the end it's a bit going back and forth with filter software program again to get a good match.
This board works with a symmetrical power supply btw, so it can be easily combined with just any amplifier with symmetrical power supply, as long as you don't overload the voltage regulators.
I am working on a little module so this can also be used with a single voltage supply as well.
Very nice!
What I would like to suggest (request if you like) is to make the boards suitable for an optional hat-type second deck with -say 4- gyrator circuits for heavy duty equalizing. In practice this will mostly be suck-out circuitry for dealing with peaking alu coned (or similar) midwoofers. The gyrator circuitry is not very component sensitive and can be easily optimized in VCad before being built. From what I have heard is that the sims and the real builds match very closely.
What I would like to suggest (request if you like) is to make the boards suitable for an optional hat-type second deck with -say 4- gyrator circuits for heavy duty equalizing. In practice this will mostly be suck-out circuitry for dealing with peaking alu coned (or similar) midwoofers. The gyrator circuitry is not very component sensitive and can be easily optimized in VCad before being built. From what I have heard is that the sims and the real builds match very closely.
To make them stackable?
Yeah that is always possible.
BTW, you can always use a standard param EQ circuit as a deep notch as well.
I personally would go hybrid I think.
So use this board as well as some passive filter elements.
In your case just an inductor plus capacitor (maybe a resistor).
Seems to bring down distortion as well.
Yeah that is always possible.
BTW, you can always use a standard param EQ circuit as a deep notch as well.
I personally would go hybrid I think.
So use this board as well as some passive filter elements.
In your case just an inductor plus capacitor (maybe a resistor).
Seems to bring down distortion as well.
Yes, an optional stack for those who would like to have that extra equalizing option. That really would be great.
You certainly have a point in going hybrid, but going active is going active for most here. Furthermore, passive filter caps and coils are outrageously expensive nowadays. But I am happy to have built my stash in more affordable days, and so have you I presume..
Plus: with all the time& energy you spend on such a project, you might als well go literally all the way.
You certainly have a point in going hybrid, but going active is going active for most here. Furthermore, passive filter caps and coils are outrageously expensive nowadays. But I am happy to have built my stash in more affordable days, and so have you I presume..
Plus: with all the time& energy you spend on such a project, you might als well go literally all the way.
Well, since the bulk is done by the active filter, you need just something simple.
Like 0.47mH or so (guesstimated value) and design your super notch accordingly.
Which are cheap and have low DC resistance.
Which I mostly do by putting a cap parallel to the inductor (incl a bit of resistance)
The truth is, that I haven't build many passive filters anymore the last few years, lol
Active is just so easy these days.
I created something like this once before. I can see a small problem in your design approach. You say:
I once designed a 2-channel board that had very flexible component placement so that both HP and LP configurations were possible and it was about 3x5 inches IIRC. At the time my idea was to stack these boards, otherwise the 2D real estate they occupy gets too large to be practical for all but the simplest and low order designs. Instead of making the board re-configurable for both HP and LP, make two boards and dedicate each to either LP or HP. In that way only the cap slots need to be sizable. Or you can do an SMD design, but they are not DIY friendly.
I went through lots of analog PCB designing and building about 15 years ago before getting into DSP and never looked back. I recently tossed a large number of the PCBs that I had developed back then since I never expect to use them in the future. I still have boxes of components that I was hoarding at the time that I should probably try to sell...
But what I see on the pic of your proto board in post #1 only has spaces for TH resistors. Capacitors for active filters do not come in this size format and the typical radial MKP and MPP type caps that would typically be used have a variety of pin spacing depending on value and voltage rating. Not every combination is available or affordable. How will the user be able to solder on a 5mm pin spacing WIMA cap that is wider than the slot you have assigned to the cap?
I once designed a 2-channel board that had very flexible component placement so that both HP and LP configurations were possible and it was about 3x5 inches IIRC. At the time my idea was to stack these boards, otherwise the 2D real estate they occupy gets too large to be practical for all but the simplest and low order designs. Instead of making the board re-configurable for both HP and LP, make two boards and dedicate each to either LP or HP. In that way only the cap slots need to be sizable. Or you can do an SMD design, but they are not DIY friendly.
I went through lots of analog PCB designing and building about 15 years ago before getting into DSP and never looked back. I recently tossed a large number of the PCBs that I had developed back then since I never expect to use them in the future. I still have boxes of components that I was hoarding at the time that I should probably try to sell...
Kemet R82 or similar will fit just fine.
Keep in mind that the voltage rating can be very small.
I have tested this as well.
I used to work and design active filters boards like this for a company I worked for before.
In that case I think we had one size bigger, which was enough to fit those orange drop caps in there.
Meaning there is space for a whole range of capacitor types.
Obviously your favorite fancy MKP of high-end brand "something" won't fit.
But it was also never the goal of this project.
Well, unless that's what people want, but that will make the board much bigger.
In any case it will always be fine to swap LP filters for HP filters.
A DSP is just easier, but not cheaper for the same performance.
In fact, getting a certain kind of performance is close to impossible with a DSP + ADC/DAC.
Personally I just like to get those boards assembled in SMD, I have done this for quite a few companies.
SMD if perfectly DIY friendly these days, especially when you go to 1206 or so.
I have a similar experience in designing all kinds of PCB's, analog as well as digital/DSP.
They all have their pros and cons.
Again, the goal here was just to provide a very affordable alternative for those who want to make a nice 2-way system (or subwoofer) but don't want to spend $$$ for either a passive filter or a DSP.
Kemet PFR are available in 1%, although the range of values is limited. Typically you have to use series and parallel combinations of caps to get specific values for an active filter.
PPS are the caps to go for in SMT, since PP and PS melt at soldering temperatures. PPS is nearly as good and very temperature resistant. Also C0G/NP0 ceramic can be used (but not any other ceramic type).
PPS are the caps to go for in SMT, since PP and PS melt at soldering temperatures. PPS is nearly as good and very temperature resistant. Also C0G/NP0 ceramic can be used (but not any other ceramic type).