State Variable Filter Design questions

ruairioflaherty · Apr 16, 2011

Hi all,

I'm working on a version of the Porter Net EQ at the moment and have been trying to improve my understanding of active filters.

I know that a state variable topology can output HPF, BPF and LPF at the same time so why do designers not use this to give a switched shelving option on eqs? (I couldn't find any examples of it anyhow). In the attached cutout the designer has chosen to switch with a relay to an entirely separate filter for the shelving option.

What would happen if I were to tap the output of the filter from point A or C instead of B? There must be some catch that I can't see?

On an unrelated topic why does Porter choose to fire the relay K3 with a transistor rather than directly from the V+?

Thank you,
Ruairi

Low Band edited schematic

ruairioflaherty · Apr 16, 2011

Full schematic for reference, sorry about the size, this is Porter's original PDF cropped.

ppa · Apr 16, 2011

ruairioflaherty said:
Hi all,

I know that a state variable topology can output HPF, BPF and LPF at the same time so why do designers not use this to give a switched shelving option on eqs? (I couldn't find any examples of it anyhow).

(if I remember well)
because the filter has different features by using its different outs. Using the filter in BDF mode its gain does not depend by the Q pot, but the same filter in HPF or LPF mode has the gain depending by the Q pot.
The problem is that to get a good Q in HPF or LPF mode (starting from the filter set for BDF mode) to have the shelving feature, the gain is too low and the signal needs to be amplified some times, so many designer prefer use an other unity gain filter to avoid other noise.

ruairioflaherty · Apr 16, 2011

Thank you Pier.

I'm toying with the idea of leaving out the shelving bands altogether on my build, I'd rather an extra band of peaking EQ.

Do you have any idea why he uses transistors to fire the relays?

ppa · Apr 16, 2011

ruairioflaherty said:
Thank you Pier.

I'm toying with the idea of leaving out the shelving bands altogether on my build, I'd rather an extra band of peaking EQ.

Do you have any idea why he uses transistors to fire the relays?

I think (or better, I supposed) because the switch spikes are induced in the signal audio in its schematic.

JohnRoberts · Apr 16, 2011

ruairioflaherty said:
Hi all,

I'm working on a version of the Porter Net EQ at the moment and have been trying to improve my understanding of active filters.

I know that a state variable topology can output HPF, BPF and LPF at the same time so why do designers not use this to give a switched shelving option on eqs? (I couldn't find any examples of it anyhow). In the attached cutout the designer has chosen to switch with a relay to an entirely separate filter for the shelving option.

What would happen if I were to tap the output of the filter from point A or C instead of B? There must be some catch that I can't see?

On an unrelated topic why does Porter choose to fire the relay K3 with a transistor rather than directly from the V+?

Thank you,
Ruairi

Low Band edited schematic

Last question first...

Two reasons to perhaps switch the relay that way.
#1 to use a lower current switch,
#2 to keep the high current flowing in the relay coil in a tight small area, while the lower current transistor drive current can be located further away and not crosstalk into audio circuitry.
=====
WRT using SVF for shelving, I know it can be done because I did it in a Loft Console back in the late '70s/early '80s. That console had 4 band full parametric, with the top and bottom EQ bands switchable between peaking and shelving EQ. I don't have schematics, or recall the details of circuits I designed 3 decades ago, but I seem to recall some weirdness with the SVF used as a shelf EQ. IIRC I did something like combine the BP and LP outputs together to make the LF shelf, and BP + HP together to make the HF shelf. Since the BP and HP or LP are 180' out from each other this required some extra polarity inversions before combining.

My recollection is that the full parametric shelf, while much more flexible than a simple shelving EQ, was perhaps more control than easily grasped or explained. For that reason I may have defeated the Q control in shelving mode, to keep the results more in line with expectations. There weren't very many of these consoles made.

JR

ruairioflaherty · Apr 16, 2011

JohnRoberts said:
Two reasons to perhaps switch the relay that way.
#1 to use a lower current switch,
#2 to keep the high current flowing in the relay coil in a tight small area, while the lower current transistor drive current can be located further away and not crosstalk into audio circuitry.

Excellent, thank you.

[quote author=JohnRoberts]
WRT using SVF for shelving, I know it can be done because I did it in a Loft Console back in the late '70s/early '80s. That console had 4 band full parametric, with the top and bottom EQ bands switchable between peaking and shelving EQ. I don't have schematics, or recall the details of circuits I designed 3 decades ago, but I seem to recall some weirdness with the SVF used as a shelf EQ. IIRC I did something like combine the BP and LP outputs together to make the LF shelf, and BP + HP together to make the HF shelf. Since the BP and HP or LP are 180' out from each other this required some extra polarity inversions before combining.

My recollection is that the full parametric shelf, while much more flexible than a simple shelving EQ, was perhaps more control than easily grasped or explained. For that reason I may have defeated the Q control in shelving mode, to keep the results more in line with expectations. There weren't very many of these consoles made.
[/quote]

A quick Google turns up nothing on the Loft schematics unfortunately, sounds like an interesting topology. As per above I'm thinking of leaving out the shelves for now.

Another question if I may? As drawn this eq has 4 bands with 10K linear gain pots in the feedback loop of U2-A. If for example I'd like to add 2 bands then would scaling the "pots" to 15K be ok? I won't need much range per band, +/- 6dB will be plenty and as these will be built on rotary switches they can be any value i want them to be.

Porter also notes that the gain ranges can be adjusted by changing the value of the resistors feeding the wipers of the gain pots.

Cheers,
Ruairi

JohnRoberts · Apr 16, 2011

ruairioflaherty said:
A quick Google turns up nothing on the Loft schematics unfortunately, sounds like an interesting topology. As per above I'm thinking of leaving out the shelves for now.

Another question if I may? As drawn this eq has 4 bands with 10K linear gain pots in the feedback loop of U2-A. If for example I'd like to add 2 bands then would scaling the "pots" to 15K be ok? I won't need much range per band, +/- 6dB will be plenty and as these will be built on rotary switches they can be any value i want them to be.

Porter also notes that the gain ranges can be adjusted by changing the value of the resistors feeding the wipers of the gain pots.

Cheers,
Ruairi

There is no U2A in the schematic you posted.

SVF are generally pretty easy to scale and adjust part values for convenience.

JR

ruairioflaherty · Apr 16, 2011

Sorry John, I wasn't clear. I was referring to the complete schematic which I attached in the 2nd post.

It does seem from my studies that SVPs are very flexible in the terms of component choices. Porter worked in this design to keep nearly all of the pots 10K and mentions in his notes some necessary tradeoffs.

As an example in this design for convenience he uses a dual 10K lin pot to feed the integrators where most commercial designs seem to use a 47K rev log. The caps needed are all small so scaling is not an issue either way. If one was building this up on rotary switches is there an optimum value - my guess is that 10K is an ok load for the 1st stage and not too high a source impedance for the 2nd. Johnson noise would hardly be an issue? Bunching of frequencies will be easy to manage with a switch.

For those not familiar with the design, I've attached a copy of Porters design notes.

gyraf · Apr 17, 2011

the Calrec Eq has switched shelve/bell this way..

JohnRoberts · Apr 17, 2011

ruairioflaherty said:
Sorry John, I wasn't clear. I was referring to the complete schematic which I attached in the 2nd post.

OK, 15k should work in place of 10k pots there.

It does seem from my studies that SVPs are very flexible in the terms of component choices. Porter worked in this design to keep nearly all of the pots 10K and mentions in his notes some necessary tradeoffs.

As an example in this design for convenience he uses a dual 10K lin pot to feed the integrators where most commercial designs seem to use a 47K rev log. The caps needed are all small so scaling is not an issue either way. If one was building this up on rotary switches is there an optimum value - my guess is that 10K is an ok load for the 1st stage and not too high a source impedance for the 2nd. Johnson noise would hardly be an issue? Bunching of frequencies will be easy to manage with a switch.

A classic design issue with SVF frequency control is getting both accuracy and wide adjustment range with standard production pots. Using a switch allows great precision, and tracking between sections.

I've never pondered ideal values for SVF but in general I would also focus on the capacitors, as they are the more difficult and expensive parts to get in good dielectric and precision. Obviously use values large enough that stray circuit capacitance is not an issue, but small enough to be practical in size and availability. I would think hundreds to thousands of pF.

If using very low noise opamps for the integrators you could consider going toward the larger cap values to keep resistor values lower and quieter. While I wouldn't take this to extremes. The noise of these integrator sections is BP filtered by the filter, so only noise in the bandpass is output.

JR

For those not familiar with the design, I've attached a copy of Porters design notes.

ruairioflaherty · Apr 17, 2011

JohnRoberts said:
I've never pondered ideal values for SVF but in general I would also focus on the capacitors, as they are the more difficult and expensive parts to get in good dielectric and precision. Obviously use values large enough that stray circuit capacitance is not an issue, but small enough to be practical in size and availability. I would think hundreds to thousands of pF.

If using very low noise opamps for the integrators you could consider going toward the larger cap values to keep resistor values lower and quieter. While I wouldn't take this to extremes. The noise of these integrator sections is BP filtered by the filter, so only noise in the bandpass is output.

Well as this will be a one off eq for my mastering work, the cost of capacitors is thankfully not an issue, nice rotary switches will be the biggest cost I face.

With that in mind I'm thinking of sticking with 10K as a value for the frequency pots, the cap values are still very reasonable. Going lower than 10K doesn't seem necessary to me but thinking about noise sources in design is really above my pay-grade, I'm only just starting to become aware of it, the trick being to know when and where it matters. In this case going to sticking with 10K yields a 7dBV drop in Johnson noise from say a 47K. I do plan to use low noise opamps which leads me to my next question...

I'm trying to learn why a certain opamp would be chosen in a given design (if as in this case we could remove cost as a factor). Porter specified all 5532 again for ease but I'm in a position to optimise my choice per section. With that in mind here's my best guess as to the specs that really matter in the filter -

- It would be nice to stick with duals for space reasons but I could pair the filter in/out amp and integrators together if using different opamps made sense. I'm thinking bipolar rather than FET inputs.

THD - obvious enough, we'd like it low and none of the amps in the filter are being asked to drive too heavy a load or being fed from too high a source impedance removing those 2 challenges at least.
Noise - None of the source impedances are too high so I think the voltage noise is dominant issue here?
Input offset voltage - Porter deals with all of the offset in one go with a 100uF electrolytic after the filter but obviously if this goes high high we'll get clicking on the switches/pots. I guess I'm safe going for anything that specs at or below 5532 levels.

Not as important here
CMRR - I'm not sure about this (I'm studying Samuel's doc and he mentions CM distortion when using non-inverting inputs, I need to work on this more)
Slew - I think anything modern will be fast enough.
Voltage swing - I'd like to aim higher here to give me options for the main signal path
Power consumption - doesn't matter, can spec the supply as needed

I'm looking at some of the National parts right now like the LM49860.

Apologies for the length of the post.

Cheers,
Ruairi

JohnRoberts · Apr 17, 2011

ruairioflaherty said:
JohnRoberts said:

I've never pondered ideal values for SVF but in general I would also focus on the capacitors, as they are the more difficult and expensive parts to get in good dielectric and precision. Obviously use values large enough that stray circuit capacitance is not an issue, but small enough to be practical in size and availability. I would think hundreds to thousands of pF.

If using very low noise opamps for the integrators you could consider going toward the larger cap values to keep resistor values lower and quieter. While I wouldn't take this to extremes. The noise of these integrator sections is BP filtered by the filter, so only noise in the bandpass is output.

Click to expand...

Well as this will be a one off eq for my mastering work, the cost of capacitors is thankfully not an issue, nice rotary switches will be the biggest cost I face.

With that in mind I'm thinking of sticking with 10K as a value for the frequency pots, the cap values are still very reasonable. Going lower than 10K doesn't seem necessary to me but thinking about noise sources in design is really above my pay-grade, I'm only just starting to become aware of it, the trick being to know when and where it matters. In this case going to sticking with 10K yields a 7dBV drop in Johnson noise from say a 47K. I do plan to use low noise opamps which leads me to my next question...

Well you've got me confused... If you are using switches, why are you talking about frequency pots?

I'm trying to learn why a certain opamp would be chosen in a given design (if as in this case we could remove cost as a factor). Porter specified all 5532 again for ease but I'm in a position to optimise my choice per section. With that in mind here's my best guess as to the specs that really matter in the filter -

- It would be nice to stick with duals for space reasons but I could pair the filter in/out amp and integrators together if using different opamps made sense. I'm thinking bipolar rather than FET inputs.

you're guessing, aren't you? Significant input bias current could cause DC issues around a SVF.

THD - obvious enough, we'd like it low and none of the amps in the filter are being asked to drive too heavy a load or being fed from too high a source impedance removing those 2 challenges at least.
Noise - None of the source impedances are too high so I think the voltage noise is dominant issue here?

noise current is multiplied times feedback network and source impedance in parallel, it could matter if design is quiet enough.

Input offset voltage - Porter deals with all of the offset in one go with a 100uF electrolytic after the filter but obviously if this goes high high we'll get clicking on the switches/pots. I guess I'm safe going for anything that specs at or below 5532 levels.

most modern parts should be lower than 553x

Not as important here
CMRR - I'm not sure about this (I'm studying Samuel's doc and he mentions CM distortion when using non-inverting inputs, I need to work on this more)

inverter sections don't need any input CMRR, other sections do...

Slew - I think anything modern will be fast enough.

anything fast enough will do... The LF and probably BP section of SVF will have relaxed slew rate needs, but these days there isn't much cost benefit to using slower parts.

Voltage swing - I'd like to aim higher here to give me options for the main signal path

Power consumption - doesn't matter, can spec the supply as needed

I'm looking at some of the National parts right now like the LM49860.

Apologies for the length of the post.

Cheers,
Ruairi

JR

ruairioflaherty · Apr 17, 2011

JohnRoberts said:
Well you've got me confused... If you are using switches, why are you talking about frequency pots?

Sorry John, I'm trying to generalise so that the thread is useful to others, hence the State Variable thread title rather than a specific "My Porter EQ". My build will be entirely on rotary switches but DC offsets is still an issue I want/need to be aware of.

you're guessing, aren't you? Significant input bias current could cause DC issues around a SVF.

Well probably one step up from guessing but barely...
- As a starting point Porter specified bipolar inputs so I'm guessing if I can stay at or below the 5532 input bias current spec I should be ok.
- None of the source impedances feeding the opamps in question is too high which is one instance when FET input opamps are a better choice.

[quote author=Ruairi]
Noise - None of the source impedances are too high so I think the voltage noise is dominant issue here?
[/quote]
[quote author=John]
noise current is multiplied times feedback network and source impedance in parallel, it could matter if design is quiet enough.
[/quote]

That will be tonight's study topic, thank you. These are the pitfalls and tradeoffs I'm trying to understand. In this case I do strive for the design to be as low noise as possible.

For anyone following, I found this video to be helpful to illustrate John's point above (the host calculates current noise starting at 4.33)
http://www.youtube.com/watch?v=W0vfALQ_n54

[quote author=Ruairi]
Input offset voltage - Porter deals with all of the offset in one go with a 100uF electrolytic after the filter but obviously if this goes high high we'll get clicking on the switches/pots. I guess I'm safe going for anything that specs at or below 5532 levels.
[/quote]
[quote author=John]
most modern parts should be lower than 553x
[/quote]

Indeed, I don't believe that there's any low enough to avoid using DC blocking electrolytic on the output of the filter so any advantages in newer parts re offset spec are mostly moot?

inverter sections don't need any input CMRR, other sections do...

I'm working on my understanding of this one, it may have seemed obvious but I never thought of non inverting input use introducing CM elements...

[quote author=Ruairi]
Slew - I think anything modern will be fast enough.
[/quote]
[quote author=John]
anything fast enough will do... The LF and probably BP section of SVF will have relaxed slew rate needs, but these days there isn't much cost benefit to using slower parts.
[/quote]

In this case the opamps are more or less cost no object. I do want to stick entirely with modern IC opamps as I don't believe discrete options will bring anything to the table - I don't need huge load driving capability and I'm not looking for mojo/tone/vibe/magical distortions etc.

Thank you,
Ruairi

pstamler · Apr 18, 2011

I think you should look seriously at the OPA132 (or OPA 2132 for a dual package). It's a FET-input opamp, meaning no problems with bias current meeting pots and turning them scratchy. FET-input opamps tend to be dirtier than bipolar in non-inverting mode, but this one, at least, is very clean in inverting mode -- see Groner's tests. As long as you're not trying to drive a 600-ohm load with it, you should get very clean results.

Peace,
Paul

ruairioflaherty · Apr 18, 2011

pstamler said:
I think you should look seriously at the OPA132 (or OPA 2132 for a dual package). It's a FET-input opamp, meaning no problems with bias current meeting pots and turning them scratchy.

Thanks Paul. What I'm really trying to do here is get an understanding of why I might choose one opamp over another. Right as you posted I have Jung, Self and Horowitz and Hill open on the bench in front of me. Not to mention about 47 webpages

I'm being beaten with the learning club bigtime.

I will add the OPA132 to my list for sure. Samuel's document is absolutely fantastic and certainly an aid in choosing the right part (at least wrt distortion specs). Again for the me the trick is understanding when it matters and when it's a non issue.

Thank you,
Ruairi

abbey road d enfer · Apr 18, 2011

Re. choice of values in SVF, pls note that the input resistor to the integrator loads significantly the pot, resuting in law-steering; in particular, that's what allows the use of Lin pots to get an almost Log frequency range. With these values, you can get good spread with sensible values.
I had designed a unit with almost perfect 1/3 octave markings and 1:10 range; it had x0.1, x1 and x1 frequency multiplicators via DG308's.
The main reason why the LP and HP outputs are difficult to use for shelving is that they are 2nd-order, so the curves are not really shelving, they have the same shape as the Pultec's combined low cut and low boost, i.e. a dip before the boost or a hump before the cut.
The circuit may easily become unstable, because of the increased phase rotation.
In addition, the frequency of the cut/boost in bell mode doesn't translate well in the shelf mode. Typically, it takes a 1kHz bi-quad to produce a 100Hz low shelf or a 10kHz high shelf.
Designers have addressed these issues in different manners, either by combining outputs or inhibiting one of the integrators.

ruairioflaherty · Apr 18, 2011

abbey road d enfer said:
Re. choice of values in SVF, pls note that the input resistor to the integrator loads significantly the pot, resuting in law-steering; in particular, that's what allows the use of Lin pots to get an almost Log frequency range. With these values, you can get good spread with sensible values.
I had designed a unit with almost perfect 1/3 octave markings and 1:10 range; it had x0.1, x1 and x1 frequency multiplicators via DG308's.

That's interesting. On a side note why a DG308 rather than a normal switch? Was the project digitally controlled in some way?

The main reason why the LP and HP outputs are difficult to use for shelving is that they are 2nd-order, so the curves are not really shelving, they have the same shape as the Pultec's combined low cut and low boost, i.e. a dip before the boost or a hump before the cut.
The circuit may easily become unstable, because of the increased phase rotation.
In addition, the frequency of the cut/boost in bell mode doesn't translate well in the shelf mode. Typically, it takes a 1kHz bi-quad to produce a 100Hz low shelf or a 10kHz high shelf.
Designers have addressed these issues in different manners, either by combining outputs or inhibiting one of the integrators.

Yes, John Roberts mentioned earlier in the thread having to combine the HP and BP sections in this instance. Before I posted I searched through quote a few commercial SVF implementations and couldn't find an instance of anyone using the individual HP or LP outputs, this confused me as every textbook constantly touts this as a big advantage of the SVF!

Thanks again,
Ruairi

JohnRoberts · Apr 18, 2011

ruairioflaherty said:
JohnRoberts said:

Well you've got me confused... If you are using switches, why are you talking about frequency pots?

Click to expand...

Sorry John, I'm trying to generalise so that the thread is useful to others, hence the State Variable thread title rather than a specific "My Porter EQ". My build will be entirely on rotary switches but DC offsets is still an issue I want/need to be aware of.

When finished this will be the Ruairi EQ...

DC offsets around the SVF are interesting since there is different AC and DC gain around the multiple feedback paths because of the integrator stages. Further the frequency control pots which look like a divider to the forward signal, to increase the effective resistance seen by the integrator multiplies the DC offset for stable operating point.

you're guessing, aren't you? Significant input bias current could cause DC issues around a SVF.

Click to expand...

Well probably one step up from guessing but barely...
- As a starting point Porter specified bipolar inputs so I'm guessing if I can stay at or below the 5532 input bias current spec I should be ok.
- None of the source impedances feeding the opamps in question is too high which is one instance when FET input opamps are a better choice.

Sorry my comment was a little mean...

Porter designed with the best tools he had at the time... In an optimal design using modern parts, you might find different parts optimal for different sockets. For example, the integrator sections need to be unity gain stable parts, while other opamps around the circuit may be operating at higher noise gains and benefit from decompensation (something Porter could manage with size of compensation cap in 5534 (while I'm too lazy to look up what he actually used).

[quote author=Ruairi]
Noise - None of the source impedances are too high so I think the voltage noise is dominant issue here?

[quote author=John]
noise current is multiplied times feedback network and source impedance in parallel, it could matter if design is quiet enough.
[/quote]

That will be tonight's study topic, thank you. These are the pitfalls and tradeoffs I'm trying to understand. In this case I do strive for the design to be as low noise as possible.

For anyone following, I found this video to be helpful to illustrate John's point above (the host calculates current noise starting at 4.33)
http://www.youtube.com/watch?v=W0vfALQ_n54
[/quote]
Sorry I didn't watch your video link.

Noise contributions in a SVF will be different for each section. The wide band noise from the HP section output gets added to the BP section noise and both get a one pole rolloff. The LP section noise which is LPF, goes though the HP stage before adding to the BP output.

Impedances seen by each stage are perhaps similar but different, and changing with frequency settings.

[quote author=Ruairi]
Input offset voltage - Porter deals with all of the offset in one go with a 100uF electrolytic after the filter but obviously if this goes high high we'll get clicking on the switches/pots. I guess I'm safe going for anything that specs at or below 5532 levels.

[quote author=John]
most modern parts should be lower than 553x
[/quote]

Indeed, I don't believe that there's any low enough to avoid using DC blocking electrolytic on the output of the filter so any advantages in newer parts re offset spec are mostly moot?
[/quote]

I don't know, there are some excellent modern parts, and perhaps other ways to avoid DC blocking caps for a very high performance path, while properly applied blocking caps can be pretty low distortion.

inverter sections don't need any input CMRR, other sections do...

Click to expand...

I'm working on my understanding of this one, it may have seemed obvious but I never thought of non inverting input use introducing CM elements...

OK, this is a pretty esoteric curve ball, but in SVF the way the opamps are used is quite different between the integrator sections and the others. When designing the circuitry inside an opamp, it is difficult work to ignore common mode voltages present at both inputs. Most opamps are very good but not perfect. For the inverting integrator stages, the + inputs are grounded, so these opamps don't have to worry about CM. In fact they don't even need to be two input opamps, while this would be more of a concern if designing with DOA. In practice, if selecting optimized opamps for each socket, the integrator sections could trade CM for some other characteristic.

[quote author=Ruairi]
Slew - I think anything modern will be fast enough.

[quote author=John]
anything fast enough will do... The LF and probably BP section of SVF will have relaxed slew rate needs, but these days there isn't much cost benefit to using slower parts.
[/quote]

In this case the opamps are more or less cost no object. I do want to stick entirely with modern IC opamps as I don't believe discrete options will bring anything to the table - I don't need huge load driving capability and I'm not looking for mojo/tone/vibe/magical distortions etc.

Thank you,
Ruairi

[/quote]

Agreed no very heavy lifting here, but I could imagine a design using a few different parts.

It is pretty common these days to use canned solutions for input/out like THAT chip sets. Porter invested some effort as I recall in making an output driver that didn't suck. He didn't have the options we do today, to use simple solutions.

I am still dealing with broad strokes here in my answers, but there are many here who can help contribute to this project.

JR

ruairioflaherty · Apr 18, 2011

JohnRoberts said:
When finished this will be the Ruairi EQ...

Well it won't be because of my ability but rather my bloody mindedness!

DC offsets around the SVF are interesting since there is different AC and DC gain around the multiple feedback paths because of the integrator stages. Further the frequency control pots which look like a divider to the forward signal, to increase the effective resistance seen by the integrator multiplies the DC offset for stable operating point.

Excellent, thank you.

Sorry my comment was a little mean...

Porter designed with the best tools he had at the time... In an optimal design using modern parts, you might find different parts optimal for different sockets. For example, the integrator sections need to be unity gain stable parts, while other opamps around the circuit may be operating at higher noise gains and benefit from decompensation (something Porter could manage with size of compensation cap in 5534 (while I'm too lazy to look up what he actually used).

No worries, you were pretty accurate.

That's the stuff that's of real interest to me, both in the practical sense of "what will I put in this box" but also in learning why a designer might choose one part or another. You and Abbey have illustrated many times over the years the reality of a real designers life in that they often need to use the parts they have in inventory and minimise using different packages etc. One of the nice things about DIY is to be free from those constraints.

In this design Porter went all 5532 so nothing was externally compensated AFAIK.

Sorry I didn't watch your video link.

Well you don't need to but the host breaks down the noise calculations for a simple opamp stage in a very clear way.

I don't know, there are some excellent modern parts, and perhaps other ways to avoid DC blocking caps for a very high performance path, while properly applied blocking caps can be pretty low distortion.

Indeed the blocking caps are not the evil enemy, as you say once properly done I'm not worried about them. Once I get a build up and running I can jumper these filter caps for some listening tests and see what they really do to the sound, I expect very little.

OK, this is a pretty esoteric curve ball, but in SVF the way the opamps are used is quite different between the integrator sections and the others. When designing the circuitry inside an opamp, it is difficult work to ignore common mode voltages present at both inputs. Most opamps are very good but not perfect. For the inverting integrator stages, the + inputs are grounded, so these opamps don't have to worry about CM. In fact they don't even need to be two input opamps, while this would be more of a concern if designing with DOA. In practice, if selecting optimized opamps for each socket, the integrator sections could trade CM for some other characteristic.

One might well be getting into diminishing returns but that's the kind of refinement that I find so interesting. I've learned so much since the start of this thread about opamp theory. Of course ideally I'd have that theory as my base and work from there but I find it very difficult to absorb with real world examples. I started a schematic last night that pairs the in/out amp and integrators together along with some other changes that Porter suggests in his readme file.

Agreed no very heavy lifting here, but I could imagine a design using a few different parts.

It is pretty common these days to use canned solutions for input/out like THAT chip sets. Porter invested some effort as I recall in making an output driver that didn't suck. He didn't have the options we do today, to use simple solutions.

I have the THAT in/out chipsets in a few units here and they are indeed very fine sounding. In conceiving this new build as part of my dream chain I'm pretty certain I'd like it to be unbalanced. I have a couple of things on my side

- My chain is (and will be) short, 4 or 5 units in close proximity.
- Al source and load impedances are predictable (and easy to work with).
- the broad goal is to change the sound as little as possible when run flat.

I am still dealing with broad strokes here in my answers, but there are many here who can help contribute to this project.

And your answers are all the more useful to me and others for it.

Cheers,
Ruairi

State Variable Filter Design questions

Help Support GroupDIY Audio Forum:

Well-known member

Attachments

Well-known member

Attachments

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Attachments

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Similar threads

Join the conversation!

Join today and get all the highlights of this community direct to your inbox. It's FREE!