HI all,
I am James. Whilst being still very noob with Pd, it is quite a while that I am pulling thread between atoms
I would like to know if anyone implemented a low pass gate (like the Buchla) in Pd ? My web search just drove me to a MaxMSP patch ( a maxpat file) which I cannot use.
Or could someone explain me how to implement this ? My trials do not sound right.

Many thanks for your help.

@JMC64 This should work.......
David.

If it doesn't attenuate enough as the control signal to [unsig~] falls then you could do something like this....

Or if that is too steep put a [sqrt] in the cord from [unsig~] to the [*~]

Hi David,

Thank you for your patch. There is something I am not sure I understand well: People say that low pass gate is not an envelope applied to a filter. In you patch it seems to me that I should apply an envelope (i.e. ADSR or else) to the cutoff of the filter ... Is that it ?

Looking forward to learn more from you.
Thank you a lot

@JMC64 Maybe some people use the same term "Low Pass Gate" to describe something other than the Buchla design?
The original definitely had a control signal. Here is a description of its use in early analogue synths...... https://electronicmusic.fandom.com/wiki/Lowpass_gate
It seems that what made it useful was the non-linearity of its response, which is why I suggested the [sqrt]. If you do more research on the Buchla design then you might get some clues about that.

The ADSR envelope you described was most likely the origin of the signal applied to the control, and probably triggered from the keyboard. So fast/slow ramps up and down to create percussive effects on the source signal.

If you are not really talking about the Buchla design then of course what I posted is not what you are looking for.... so a different description is necessary.
Pd is verstatile...... so we often say that "anything can be done"........
I am sure you can get what you want to achieve.
David.

@David, Thank you for your explanations. The Buchla design is what I am looking or, indeed.
Looking on internet , I found http://electro-music.com/forum/phpbb-files/vactrolsimpleresponse_5_195.jpg showing the response of a vactrol
I end up with doing the following using a sigmoid function.
osc~ line~ 1 55
| |
| expr~ 1 - 1/(1+exp(6*(2*$v1-1)))
| |
| expr~ if($v1 > 0.995, 1, $v1)
| |
| expr~ if($v1 < 0.005, 0, $v1)
| |
| *~ <---Cutoff frequency
| |
svf~

However (as I do not own a LowPass gate hardware, I don't know what happens when you apply an attack which last longer than the response time of the "vactrol", what happens?

Is it simply an envelope -> Sigmoid or is is something like Envelope * Sigmoid . The latter being trigger by its gate (line~)

Any idea? thank you .

@JMC64 I have not checked it, but I assume you have created a ramp with the shape of the brown/red curve.
In the middle (to the end of) analogue olden days the way to control a voltage with another was through an opto-isolator.
The green line is its input signal. A voltage of 0.5V for 100ms.
0.5V because that was the necessary voltage to set "on" and 100ms because that was the time necessary to get the output side to "full".
Once the input voltage was removed it took about another 75ms before the output returned to zero.

So you have found the non-linearity of the control.

In the synth application the input would have been either a voltage from a keyboard key, or a square/sawtooth lfo wave probably for a repeated pulse, if that was what was selected. Well actually it could have been any waveform that the synth could produce...... depending on where it was patched by the operator.

The problem is that the led in the opto-isolater takes time to ramp down the resistance of the light sensitive resistor (so that the output voltage rises), and so very short control pulses will not get to full voltage on the output. Longer ones will be fine...... the Led can be held "on" for as long as you like......

The low pass filter cutoff and resonance might be harder to evaluate, but this paper....... 44.dafx2013_submission_56.pdf has all the information you will need.
David.

@whale-av : thank you a lot. The paper you are referring is really good but a bit to math oriented for someone who gave up on Z transforms 30 years ago.
In fact, you are exactly hitting the difficult point I face with pd.

In pd, I do not see how I can interrupt a slope while it is rising. My understanding is that the response of a vactrol is somehow fixed. I.e. it will take 100 ms to reach 100% of its response. If I apply a pulse of 50 ms sec, then , it will only reach #50% of the response.
Taking this into account, the frequency of the cut-off will be roughly 50% of the maximum opening (this being if the response is was purely linear).
But in Pd, if I apply a line~ object, the line will always reach 1 after its duration time.
How do I make in pd a function which is time dependent (reaching 100% of its value after i.e. 100 ms and which would stop at 70% if the impulse duration would be i.e. 50 ms ?
My understanding of pd is that user only has access to signal y = f(t) and not to time (t). But maybe I dunno the trick ?

@JMC64 Me too. I would have great difficulty recalling anything math from my 1970's uni days.
But no. [line~] will not reach 100% if it receives a new message taking it down before it gets there.
You can easily prove that like this......

The easiest way to mimic the opto-isolator will be with [curve~] from the cyclone library, rather than [line~]..
Something like........

The [curve~] help-file helps visualize the curves that are being produced.
Play with the 3rd parameter until you get the desired curves.
You will probably need a [change] before [select 0 1] for a practical application, or a [moses 0.5] instead of [select] and a $1 in each message if you will expect all values between 0 and 1.
David.

@Dave, thank you a lot for your help... I think I manage to make something out of this.
As I felt that the trigger (or gate should be a signal) I used a threshold~
Please find enclosed a simple demo patch which requires the cyclone lib. It can make sone nice drums sounds. Comments, advices for improvement are welcome.
test_lowpassgate.pd

Hello! Kind of late to this thread.
Thought I might post my attempt on doing a low pass gate using vanilla objects.
Replaced threshold using >=~ from the heavy library.
And used lop~ since it has similar curves like the vactrol, only fixed attack and release.
Finally, I used a resonant-lowpass filter from the heavy library.
Lots of great info here, thanks guys.
LPG.pd