Royal Filter updated

[StephenGiles]

http://www.4shared.com/file/39430228/a9b5c65d/Royal_Filter_2008.html



To get you started, this is purely an idea for discussion, improvement and even making the beast work. I have no time to breadboard it as we are seeing the Lark Rise Band this evening - yes, real Lark Rise music which should have been used in the current TV series!.......and Mothering Sunday visiting tomorrow.

Just 4 ICs, and there are 3 spare gates on the 4016. What could they be used for?

[slacker]

Looks interesting. Is this the circuit where the start frequency for the filter is set by how loud the signal is?

[StephenGiles]

In theory, yes.

[StephenGiles]

 

[markusw]

Hey Stephen,

thanks for the food for thought 

I hacked your schem into LTSpice.

A couple of comments/questions:

* IC2A is an inverting gain stage? I think the +ve and -ve inputs are reversed.
*The whole thing is supposed to run from a bipolar supply?
* In the sims the trigger plus 4016 is working. However, the 2N5087 isn't.
How is the 2N5087 supposed to work? Sorry for the dumb question..
*When I replace the 2N5087 with a NPN the sim seems to work (output of IC2d first follows the the natural envelope and as soon as the trigger switches to "low", it decays according to the settings of the "stop pot" and the "decay pot"). Don't know why though ....

Please remember: just simulations so far...

Best regards,

Markus

[StephenGiles]

Thanks Markus, you are right about IC 2a  -circuit is updated.

Yes it runs from a bipolar supply, I'll put a note on the circuit.

A 5087 is used in the Crashpad circuit, I can't find my scan at the moment - it was a very bad one anyway so I'll do a new scan. I think that when turned momentarily on by a pulse from trigger ICIb it is simply used to pass the voltage at the output of IC2c.

Now I want that voltage to be proportional to the strength of the note played. Peak Follower IC1c should deal with this nicely. IC1b trigger should also turn on the 4016 switch long enough to pass the Peak Follower voltage to Sample and Hold IC1d, and buffer IC2d to quickly charge the large non-polarised capacitor to the Peak Follower voltage level. The capacitor is then allowed to discharge to the voltage held at the output of IC2b by the Filter Stop frequency pot at the rate set by the Sweep Time pot.

I have used the Bi Filter Follower filter purely because I like the sound of it, but any suitable filter will do. It's the control circuitry that bugs me at the moment.

A couple of questions - firstly I forget where the trigger circuit came from but will it do, and can it be used to switch on both the 4016 switch and the transistor? I thought maybe it would need isolating resistors of modest value, but am not sure what.



What I want to happen when energised by a note played is this:

[StephenGiles]

http://www.4shared.com/file/39556305/8587cb8c/EHcrashpad.html
http://www.4shared.com/file/39556657/10359c0e/Royal_Filter_2008ver2.html

Here is the updated circuit and the Crashpad circuit where you can see that the transister is a 5087.

[markusw]

Thanks Stephen for the Crashpad schemo  and for the update

Another observation: if I add an inverting opamp stage between the trigger opamp and the 5087 it seems to work too (like without the additional inverting stage and the NPN).
Might it be that the 5087 needs a negative pulse (voltage) to turn on? Sorry for another dumb question 

Could the 5087 be replaced by one of the 4016 switches?

BTW, I also tried to run the simulation by feeding a wav sample (from my bass).
It seems that the trigger is "on" for quite a while before the audio signal drops sufficiently to make it go low.
So probably it would be an idea to include a differentiator cap to make the trigger go on for just a short time.
On the other hand, for guitar (played with a pick) the trigger pulse should be shorter.
What do you think?

[StephenGiles]

Thanks Markus, I think that a 4016 switch in place of the 5087 is a good idea.

I've added 0.01uf differentiator caps and a 10k resistor to ground for starters, and I'll also look into length of the trigger pulses.

Rough updated schematic  is here:

http://www.4shared.com/file/39626552/56ad7866/Royal_Flter_ver3.html

I'm not at home this evening but hopefully I will bread board this tomorrow.

Any other contributions from folks would be welcome!

[markusw]

Thanks for the update! 
Will change the simulation accordingly this evening and report back.
Might be that the differentiator will need another comparator to turn on the 4016.
Will check it in the simulations.

I'm very curious about your breadboarding results!
Simulation is one thing, reality another.
On the other hand, so far almost all of the simulations were at least very close to reality.
Let's see....

[markusw]

Hey Stephen,

I changed the simulation according to your schem.

Some observations from the simulations (again fed with a wav file):

* the 100u cap after IC1c was too large, made the envelope too slow
* a resistor was added before the differentiating cap. seems Harry Bissell did this mod in his envelope follower to reset the three peak followers, see his "Muffy".
Without it the second switch doesn't really work.
* The two 10µ decay caps were replaced by a 100n cap. Otherwise the switching pulse was too short to charge the decay caps. According to the sims the 100n/1Meg combination should give a useful decay time.
* The resistor in front of the sample cap was removed. Otherwise it was charged too slow and the "decay sample/leak" did sample just a very low value.

See attached the  screenshot http://smg.photobucket.com/albums/v21/markusw/?action=view&current=Clipboard01.jpg


The traces:

Green: voltage at the 10u cap after IC1c
Blue: the final envelope
Red: the switching pulses

It seems that the decay cap samples too early and doesn't really get the peak.
Could the switching pulse be delayed for or stretched to let's say 10ms? This should be enough to make the decay cap catch the peak (or at least a voltage close to peak).


I'm keen to see/hear how this transfers to your breadboard 

What do you think?

[markusw]

Hey Stephen,

I changed the simulation according to your schem.

Some observations from the simulations (again fed with a wav file):

* the 100u cap after IC1c was too large, made the envelope too slow
* a resistor was added before the differentiating cap. seems Harry Bissell did this mod in his envelope follower to reset the three peak followers, see his "Muffy".
Without it the second switch doesn't really work in the sims.
* The two 10µ decay caps were replaced by a 100n cap. Otherwise the switching pulse was too short to charge the decay caps. According to the sims the 100n/1Meg combination should give a useful decay time.
* The resistor in front of the sample cap was removed. Otherwise it was charged too slow and the "decay sample/leak" did sample just a very low value.

See attached the  screenshot http://smg.photobucket.com/albums/v21/markusw/?action=view&current=Clipboard01.jpg


The traces:

Green: voltage at the 10u cap after IC1c
Blue: the final envelope
Red: the switching pulses

It seems that the decay cap samples too early and doesn't really get the peak.
Could the switching pulse be delayed for or stretched to let's say 10ms? This should be enough to make the decay cap catch the peak (or at least a voltage close to peak).


I'm keen to see/hear how this transfers to your breadboard 

What do you think?

[StephenGiles]

Thanks Markus, I'll have another think and blow the dust off my breadboard tomorrow!

[puretube]

how about replacing the 2 optoes with the 2 spare switches,
and PWM them for frequency control, M*R-style, Sir?

[StephenGiles]

how about replacing the 2 optoes with the 2 spare switches,
and PWM them for frequency control, M*R-style, Sir?

Possibly

[puretube]

BTW:
so the unwanted side effect of anything envelope-controlled
is: "the Ripple"
(a smallish processed derivate of the original signal, riding on the [wanted] largerish envelope contour...).

So:
how about adding a little oppositely phased ripple to the "signal in process",
such that the ripple content cancels out???

(without wanting to go further into details, with the Frankfurt MusikMesse 2008
with all it`s innovations coming up very soon...  )

[markusw]

how about adding a little oppositely phased ripple to the "signal in process",
such that the ripple content cancels out???

Sounds interesting!

So we just have to wait until Frankfurt Musikmesse 2008 for you revealing some more details? 

[puretube]

no - the waiting time will rather be
until some socalled "freeze tomp oxes"
have revenge-e-neared certain future pedals... 

[StephenGiles]

no - the waiting time will rather be
until some socalled "freeze tomp oxes"
have revenge-e-neared certain future pedals... 

Ah, I'm with you now Plenty of goop on your's I hope!

[StephenGiles]

It's looking like next week now before I have enough clear time to breadboard this - and a little further research is required.