[bob~]

Reply to [bob~] on Thu, 02 Sep 2021 22:33:42 GMT

jameslo — Thu, 02 Sep 2021 22:33:42 GMT

How would you configure and use [bob~] to demonstrate what the ladder filter fuss is all about? For some reason (inexperience? poor taste?) I keep returning to [vcf~].

Edit: after an hour or two of monkeys at the typewriter, I'd say that the 2 main differences from [vcf~] are the roll-off slope ([bob~] never completely mutes signal even when the cutoff freq is 8 hz) and the resonance sounds brighter and grittier (bob at resonance 3 sounds like vcf at q 22 to me). There's got to be more to it than that, right?

Reply to [bob~] on Fri, 03 Sep 2021 08:40:13 GMT

oid — Fri, 03 Sep 2021 08:40:13 GMT

@jameslo [bob~] is not very good, I would never guess it was a transistor ladder if it did not say so, never would guess it was an emulation of an analog VCF even. 90% of the transistor ladder sound is it being 4-pole, the moog sound is just as much the VCOs with their rather bent waveforms. A few of the early ARP synths used a transistor ladder which is ultimately just a slight improvement on the early moog ladder and almost identical to the later moog ladders, but no one every mistakes those ARPs for moogs and they sound nothing alike. Four [vcf~]s in series will probably get you closer to the ladder sound than [bob~], [vcf~] is a pretty good lowpass.

I wonder if you can do pole mixing in pd...

Reply to [bob~] on Fri, 03 Sep 2021 12:55:42 GMT

bocanegra — Fri, 03 Sep 2021 12:55:42 GMT

I wonder if you can do pole mixing in pd...

Pole mixing in itself should be trivial -see: https://electricdruid.net/multimode-filters-part-2-pole-mixing-filters/ and my crude implementation:

(You could replace [lop~] with [vcf~])

The problem is implementing proper feedback/resonance. See this digital implentati0n of the moog ladder (from https://www.researchgate.net/publication/241280072_POWERWAVE_-A_HIGH_PERFORMANCE_SINGLE_CHIP_INTERPOLATING_WAVETABLE_SYNTHESIZER ):

The single RC stages would be easy to to replicate but to get proper (0 latency) feedback you'd have to stuff everything into a single [fexpr~] object (or design it with a block size of 1?).

Something for @jameslo to tinker with

Reply to [bob~] on Fri, 03 Sep 2021 13:02:34 GMT

jameslo — Fri, 03 Sep 2021 13:02:34 GMT

@oid I was afraid to reach that conclusion, but you make a strong case. I wonder if [audiolab/pp.ladder~] is closer?

@bocanegra He he he....

Reply to [bob~] on Fri, 03 Sep 2021 19:22:40 GMT

seb-harmonik.ar — Fri, 03 Sep 2021 19:22:40 GMT

@jameslo @oid there's also [ggee/moog~] and [muug~] https://github.com/pcasaes/muug
I often use [cyclone/svf~]'s lowpass. not sure why [bob~] sounds so unlike a lowpass.. maybe miller overlooked something, or maybe it's not as steep as other lowpasses.. Personally I still like the sound of it tho. (although it's more like a peaking/lowpass hybrid)

Reply to [bob~] on Sat, 04 Sep 2021 21:41:38 GMT

tungee — Sat, 04 Sep 2021 21:41:38 GMT

This is the automatonism implementation of the bob~ object. This module sounds really nice moogish

Reply to [bob~] on Sat, 04 Sep 2021 13:36:41 GMT

jameslo — Sat, 04 Sep 2021 13:36:41 GMT

@tungee Thanks. What's going on with the rightmost [bob~] instance? Should it's attenuated output be connected to the outlet?

Reply to [bob~] on Sat, 04 Sep 2021 13:44:26 GMT

tungee — Sat, 04 Sep 2021 13:44:26 GMT

And here the Ladder mod of automatonism

Reply to [bob~] on Sat, 04 Sep 2021 13:47:53 GMT

tungee — Sat, 04 Sep 2021 13:47:53 GMT

In my opinion the power of the Bob object is leashed, when more then 1 Osc is feeded into the Bobcat like in a real analog synth( 2 OSc+ plus subosc)

Reply to [bob~] on Sat, 04 Sep 2021 13:49:55 GMT

tungee — Sat, 04 Sep 2021 13:49:55 GMT

@jameslo I think the output of the 0.33~must be routed also into the output

Reply to [bob~] on Mon, 06 Sep 2021 17:02:15 GMT

bocanegra — Mon, 06 Sep 2021 17:02:15 GMT

As the feedback/resonance problem has been solved in @oid 's polemixing thread, here is a simple yet somewhat CPU expensive implentation of the moog ladder with signal rate cut-off inlet (without sigmoids in the filterloops tho): 4stageLadder.pd

There is an output for each stage so you can do your pole mixing should you want to, but for pure googelymoogely sound just use the last output

Inlets: Signal, Cut-off, Resonance

Reply to [bob~] on Tue, 07 Sep 2021 10:28:13 GMT

tungee — Tue, 07 Sep 2021 10:28:13 GMT

Hi Bocanegra what are ranges of the cutoff frequency and resonance? Bytheway thank you. Nice and compact solution

Reply to [bob~] on Tue, 07 Sep 2021 12:22:16 GMT

bocanegra — Tue, 07 Sep 2021 12:22:16 GMT

@tungee cutoff clips at around 14KHz i think (samplerate/2pi) - I haven't tried out frequency above 10k. Resonance is theoretically infinite, but I haven't gone beyond 15 in my test. Above 3 you will begin to hear some distortion of the feedback signal.

Reply to [bob~] on Tue, 07 Sep 2021 16:50:04 GMT

jaffasplaffa — Tue, 07 Sep 2021 16:50:04 GMT

I also started a thread here, asking for help for optimising a Karlsen Ladder filter. I think we ended up with a decent optimised low pass filter. The high pass is not good, it could be a lot better. But I like the sound of the low pass filter.

https://forum.pdpatchrepo.info/topic/13202/looking-for-help-for-optimisation-of-karlsen-24-db-ladder-filter/29

It has oversampling implemented too.

Reply to [bob~] on Tue, 07 Sep 2021 18:42:57 GMT

tungee — Tue, 07 Sep 2021 18:42:57 GMT

@ bocanegra
i used 20hz to 14000hz for the cf and 0-15 for the resonance... but the resonance affects the sound very subtle. Is it so wanted?

@jaffasplaffa
i do like the karlsen Lp very much

Reply to [bob~] on Tue, 07 Sep 2021 19:02:09 GMT

bocanegra — Tue, 07 Sep 2021 19:02:09 GMT

@tungee - my mistake. I just took the guts from my version of @oid 's polemix which I had been doing all my testing on and thus forgot to invert the feedback.

Fixed version: 4stageLadder.pd

Basically just made the averaging filter in the feedback loop invert the signal. Weird because there is no inversion of the feedback signal in oid's polemix

Reply to [bob~] on Tue, 07 Sep 2021 19:24:01 GMT

tungee — Tue, 07 Sep 2021 19:24:01 GMT

I tested it again bocanegro, but the resonance is far from selfresonance. I miss the aggresiveness of the resonance somewhat

Bildschirmfoto 2021-09-07 um 21.26.51.png

test4stageladd.pd.zip

Reply to [bob~] on Tue, 07 Sep 2021 19:27:18 GMT

bocanegra — Tue, 07 Sep 2021 19:27:18 GMT

@tungee - the feedback chain has a softclipper (tanh function) that makes it overdrive at >3 resonance but never hit the 1/-1 clipping wall. Try cranking it up to 25 and see. Or open it up and bypass the [expr~ tanh($v1/5)] object in the feedback path - but then your should be careful with resonance values > 1...

Reply to [bob~] on Tue, 07 Sep 2021 19:34:13 GMT

bocanegra — Tue, 07 Sep 2021 19:34:13 GMT

@tungee Judging from your screenshot you still haven't inverted the feedback path. Do like this:

Reply to [bob~] on Tue, 07 Sep 2021 19:56:03 GMT

oid — Tue, 07 Sep 2021 19:56:03 GMT

@bocanegra said:

Weird because there is no inversion of the feedback signal in oid's polemix

Sure there is, look harder, I multiply the resonance paths mix ratio's by -1 when the filter mode/resonance path combination requires inversion . Part of the trick which keeps the resonance morph from sounding blah.

Reply to [bob~] on Tue, 07 Sep 2021 20:03:33 GMT

bocanegra — Tue, 07 Sep 2021 20:03:33 GMT

@oid - you are right! I have been starring myself blind at the lower left of the circuitry and missed out on all the mixing trix

Reply to [bob~] on Tue, 07 Sep 2021 20:45:12 GMT

tungee — Tue, 07 Sep 2021 20:45:12 GMT

@bocanegra

ok that dit it Bocanegra! I replaced also the tanh expr with an analog clipper . wonderfully harsh now

Thank you

Reply to [bob~] on Tue, 07 Sep 2021 22:55:10 GMT

tungee — Tue, 07 Sep 2021 22:55:10 GMT

replacing expr with the analog cliping has reduced cpu consumption of the patch also!

Reply to [bob~] on Tue, 07 Sep 2021 21:12:47 GMT

bocanegra — Tue, 07 Sep 2021 21:12:47 GMT

@tungee Nice one! I wonder if the anaclip algorithm is cheaper on CPU than [expr~]...

Comparison of the two functions:

They should sound the same. Except for the divide by 5 currently in the expr~

EDIT: "Anaclip" indeed shaves off a significant amount of CPU use

Reply to [bob~] on Tue, 07 Sep 2021 21:21:16 GMT

tungee — Tue, 07 Sep 2021 21:21:16 GMT

Yes i have 60% CPU saving Boca! expr~ sucks also more details in the sound in my opinion!

Reply to [bob~] on Tue, 07 Sep 2021 21:43:18 GMT

oid — Tue, 07 Sep 2021 21:43:18 GMT

@tungee nice, makes the [lop~] version of my morphing filter barely even register on the cpu use. Will rework it to suit and include it in my next go around. What you hear as a difference is just that you shifted the q range which gives you more control in the area you are interested in, playing with the division in the [expr~] will yield the same results at. The /5 was picked for the morphing filter since high distortion self oscillation range emphasizes the morphing nicely, but at cost of more traditional resonance sounds, putting the sliders max range at 5 makes it easier to get the more traditional sounds.

Reply to [bob~] on Tue, 07 Sep 2021 22:10:38 GMT

tungee — Tue, 07 Sep 2021 22:10:38 GMT

Build a automatonism modul out of your abstraction Bocanegra. with modulation inputs preset saving midi controlling etc.

I really like it how it sounds

Here is the automatonism module with install instructions and the modul itself:
Boca-Ladder.zip

Thank you for your explanations oid