[sdiy] High pass filter using the 2164

Rutger Vlek rutgervlek at gmail.com
Mon Mar 25 19:32:35 CET 2019


Hi Mike,

I second the suggestion to check the latest SSI2164 datasheet. I’ve contributed schematics for low-pass, high-pass and all-pass stages, along with a driver circuit to drive them (with V/oct scaling if needed). You can cascade stages to obtain band-pass filters, phasers, and various slopes of hp or lp functions. Alternatively, you could obtain the same variety of responses (albeit less accurate, sometimes) by ‘pole mixing’, as David already explained. I’ve used the same approach in the Dendrites multimode VCF for Eurorack format.

@David: thanks for the kind words! Inspiring each other this way is exactly what I think is the main purpose of this wonderful list. Speaking of this feedback path as a separate input for the pole-mixer, there are several ways to do it. Mathematically these are identical… until one of the stages starts to saturate and become non-linear.

On a side note: perhaps you’ve seen the attempts I made at modelling (in analog circuitry!) of various different filter topologies (Moog, Korg, Prophet, Yahama) in the Nucleus and Dendrites designs. In terms of resonance behaviour, low-end response, high-end roll-off I’m really happy with the results. But to be honest, I’ve always felt the saturation behaviour of both design was a little ‘rough’. It’s basically a combination of discrete OTA stages saturating and a Zener circuit on the input+feedback mixer that acts differently on the feedback and on the input (creating interesting interactions). I’m now experimenting with attempts to model saturation behaviour in a way that you could also switch, and decouple this from the integrator stages in the filter, such that I could achieve better tracking and lower noise at the same time. Not sure if it will work out though, the circuit is getting rather complex, and I’m wondering how musically relevant the differences are.

Regards,

Rutger


> On 25 Mar 2019, at 18:22, David G Dixon <dixon at mail.ubc.ca> wrote:
> 
> This is a good tech note, but it ignores the input+feedback as a fifth signal.  In fact, using the 5 signals available in the filter, a larger range of filter modes is possible (at least 31, 22 of which were selected for the Intellijel Polaris, which uses the 5-input pole mixing scheme).  By disabling the first stage, one gets that input+feedback as the output of stage 1, which means that the disabled modes are limited to 3-pole operation.  There are a significant number of interesting 4-pole modes which require an enabled first stage.
>  
> Also, a public and heartfelt thanks to Rutger Vlek for cluing me in to adding the feedback to the input for that fifth signal.
>  
> If you want to see the filter modes being utilized on the Polaris, please consult the Polaris user's manual which is posted on the Intellijel website:
>  
> https://intellijel.com/downloads/manuals/polaris_frequency-responses.pdf <https://intellijel.com/downloads/manuals/polaris_frequency-responses.pdf>
>  
> If you're curious, these are the gain formulae for each of the 22 responses shown in that PDF document (Note, every other value is negative):
>  
> LPA 0 1 0 0 0  HPA 1 1 0 0 0  BPA 0 1 1 0 0  NA 1 3 6 4 0    APA 1 2 0 0 0 
> LPB 0 0 1 0 0  HPB 1 2 1 0 0  BPB 0 0 1 1 0  NB 0 1 3 6 4   APB 1 4 4 0 0
> LPC 0 0 0 1 0  HPC 1 3 3 1 0  BPC 0 0 0 1 1  NC 0 3 6 4 0   APC 1 6 12 8 0
> LPD 0 0 0 0 1  HPD 1 4 6 4 1  BPD 0 0 1 2 1  ND 0 4 12 16 8 APD 1 8 24 32 16
>                               BPE 0 1 3 3 1  NE 1 4 12 16 8
>  
> So, based on this table, there are three modes where the fifth input is required.  These are the ones which require both the input signal and the output from the fourth stage, so HPD, NE, and APD.  The most important of these is the four-pole highpass mode, HPD.  Also, the double notch, NE, is a very interesting and useful mode that is only available with the five-input scheme.  Finally, the four-pole allpass mode APD is important for phasing.
>  
> Another thing which is cool which was not done in the Polaris, but which I may do when I finally get around to building a "Doc Sketchy" version of it, is to crossfade that first input.  You will notice that the first input is always either 0 (off) or 1 (unity gain).  It turns out that one can smoothly morph between various modes by fading this input.  For example:
>  
> NA --> NC
> ND --> NE
> LPA --> HPA
>  
> The most interesting of these is the notch pair ND-NE.  When the input signal is halfway on, the gain formula is {0.5 4 12 16 8}, which is half of the APD formula {1 8 24 32 16}.  This means that when these two notch responses are input-crossfaded, they go through the 4P AP response at half gain.  I don't know, it might sound cool...?
>  
> Even though they are not listed, there are many other interesting modes possible when the input+feedback signal is faded in and out against some of the other modes.  One does have to be a little bit careful about total gain, though.  For example, if one does this against the APB mode, one gets the BPA mode at a gain of 4 when the input is turned off.
>  
>  
> -----Original Message-----
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On Behalf Of Neil Johnson
> Sent: Monday, March 25, 2019 7:59 AM
> To: Mike B
> Cc: SDIY List
> Subject: Re: [sdiy] High pass filter using the 2164
>  
> Hi Mike,
>  
> > I working on a 4 pole Cascade high pass filter.  My present design uses Lm13700 and works fine, but I would like to switch to the V2164.  I cannot find anyone doing this.
> > 
> > The issue is the input resistor on the 2164 to convert the voltage to an input current were as the 13700 is a voltage input. We're would that input resistor go?
> > 
> > Hopefully this is not a really stupid question.
>  
> Olivier Gillet published a nice tech note on making filters with the SSM2164:
> https://mutable-instruments.net/archive/documents/pole_mixing.pdf
>  
> That shows how to make a single low-pass filter element, then combine
> four of them in a chain and from that synthesize a range of filter
> responses by mixing various tapped outputs together.
>  
> Cheers
> Neil
> -- 
> http://www.njohnson.co.uk
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org <mailto:Synth-diy at synth-diy.org>
> http://synth-diy.org/mailman/listinfo/synth-diy <http://synth-diy.org/mailman/listinfo/synth-diy>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20190325/c3c714b6/attachment.htm>


More information about the Synth-diy mailing list