Mastodawn

Graham Sutherland / Polynomial

one hundred internet points if you can successfully guess what this circuit does, and how.

three hints:

1: it's pretty weird.

2. the inputs and outputs are no more than a couple tens of kilohertz (and are not a fixed frequency - the values shown in the sim below are arbitrary)

3: it only works in a real circuit and if you simulate this you will not see it doing the weird thing.

UPDATE: we have a winner. solution is here: https://chaos.social/@gsuberland/113921571918206983

Graham Sutherland / Polynomial (@[email protected])

I'm not even sure how to properly describe this. "nonlinear modulating low-pass filter", I guess? the core of it is that the MOSFET is a big power FET with a ton of Cgs. combined with the 220k resistor, you get a low-pass filter. the channel A voltage is adjusted to keep the MOSFET sub-threshold. except the really evil part of this is that the MOSFET's gate capacitance is dependent on Vds. so channel B modulates Vds, adjusting the filter's cutoff frequency.

chaos.social

Graham Sutherland / Polynomial Jan 31, 2025

anyone who asks an LLM is instantly disqualified and also loses my respect.

Graham Sutherland / Polynomial Jan 31, 2025

ok, a few of you have gotten very close so I'm going to provide a bit more help

Graham Sutherland / Polynomial Jan 31, 2025

alright, let's go with one final hint that's REALLY going to bake your noodle.

the MOSFET never turns on. at all.

Graham Sutherland / Polynomial Jan 31, 2025

we have a winner!

https://aus.social/@nickzoic/113921516097962178

nick (@[email protected])

@[email protected] @[email protected] uh, dang, I was working on the idea that the power rail was sagging ... ok so this has to be something to do with the capacitance of the depletion region varying with V_DS and thus turning that mwahaha bit into a swept filter

Aus.Social

Graham Sutherland / Polynomial Jan 31, 2025

I'm not even sure how to properly describe this. "nonlinear modulating low-pass filter", I guess?

the core of it is that the MOSFET is a big power FET with a ton of gate capacitance (Cgs). combined with the 220k resistor, you get a low-pass RC filter. the channel A voltage is adjusted by the opamps to keep the MOSFET sub-threshold.

except the really evil part of this is that the MOSFET's gate capacitance is dependent on Vds. so channel B modulates Vds, adjusting the filter's cutoff frequency.

Graham Sutherland / Polynomial Jan 31, 2025

I have literally no idea what this sounds like but I'm going to build it and find out.

Graham Sutherland / Polynomial Jan 31, 2025

another fun detail is that if you use a trench FET the Cgs has a very nonlinear drop as Vgs increases, whereas if you use a planar FET the Cgs is a fair bit more linear.

Graham Sutherland / Polynomial Jan 31, 2025

stop reading my mind, bot

https://chaos.social/@weirdunits/113921206654668130

Weird Units (@[email protected])

coulombs per MOSFET

chaos.social

Graham Sutherland / Polynomial Jan 31, 2025

this isn't even the most cursed idea I have for abusing MOSFET properties. you can get into far weirder territory with SiCFETs, for example: https://chaos.social/@gsuberland/113924986156295201

Graham Sutherland / Polynomial (@[email protected])

one of the other evil tricks I want to try abusing with lesser-known FET behaviours is utilising subthreshold hysteresis in SiCFETs as a distortion effect. SiC has this odd property where the "history" of charge flow on the gate modifies the Vgs behaviour, moving the Miller plateau and threshold voltage depending on whether you were charging or discharging the gate, and how deep into the subthreshold region you went. the magnitude of the effect is also altered by Vds.

chaos.social

gaytabase Jan 31, 2025

Graham Sutherland / Polynomial Jan 31, 2025

@dysfun I do have bagels and, at least in my mind, they are glowing like a pickup in a game.

jaseg Jan 31, 2025

@gsuberland using something like IXTN660N04T4 you’d even be able to get a useful amount of apparent capacitance there. IXTN660N04T4 has almost a microcoulomb of gate charge.

Graham Sutherland / Polynomial Jan 31, 2025

@jaseg yeah infineon have some chunkybois too, and some wide SOA ones with weird transfer curves (presumably because they're combination trench and planar)

Fi 🏳️‍⚧️Jan 31, 2025

oh that's gonna sound real funky

✧✦Catherine✦✧Jan 31, 2025

@gsuberland yeah i figured that you're using the FET as a modulator but it wasn't completely clear for what purpose

ms_nonbinary_flag

@gsuberland so it should just be a scaled and shifted 50hz sine? the mosfet looks like its doing, well, not much since all you'd see there is gate capacitance? or is that *exactly* what its doing and the parasitic capacitance is making it act as a filter? would be very scuffed but i only have a basic understanding of electronics so uhhh not sure

ms_nonbinary_flag

@gsuberland oh the answer toot did not federate, but it was capacitance afteralll, crazy circuit! didn't know that much about depletion region stuff but am glad to have learnt a bit more about it :>

nick Jan 31, 2025

@gsuberland SUS

emily, blinkenlight witch Jan 31, 2025

@gsuberland this is my favorite nethack level

zl2tod Jan 31, 2025

@gsuberland
Bloody ring modulator ...

Bloody phase comparator ...

You are trying to make me buy that arbitrary waveform generator ...

Lasse Beyer Jan 31, 2025

@gsuberland oh, I thought you meant this in general, not just specific to this thing here 😅

Fi 🏳️‍⚧️Jan 31, 2025

hm. that looks like some modular synth nonsense.

Graham Sutherland / Polynomial Jan 31, 2025

@munin you are within the ballpark, but the devil is in the details :)

Allan Chow Jan 31, 2025

@gsuberland I am part of the problem
https://g.co/gemini/share/67165a2cc43c

Before you continue

SA6JKK - Johan Jan 31, 2025

@grumpasaurus @gsuberland There is no frequency dependent components in the op-amp feedback. So I don't think a filter is correct.

I only see resistors and gain stages so it might do something to add signals together

SA6JKK - Johan Jan 31, 2025

@grumpasaurus @gsuberland Well I was wrong. The FET was the filter capacitor.

greg 🇿🇦Jan 31, 2025

@gsuberland two wild guesses:

1. Microphone preamp with built in 'push to talk'

2. Microphone preamp with built in ANR

Graham Sutherland / Polynomial Jan 31, 2025

emily, blinkenlight witch Jan 31, 2025

@sk44p @gsuberland ANC (automatic noise creation)

Lunaphied Jan 31, 2025

@gsuberland hopefully you will share both the answer and how you came up with the damn thing eventually?

Graham Sutherland / Polynomial Jan 31, 2025

@Lunaphied naturally. it's too cursed not to talk about :D

(I'm also going to build it)

zl2tod Jan 31, 2025

@gsuberland
Something about letting bursts of 1kHz through at 50Hz using the gate protection diodes in the FET ...

Graham Sutherland / Polynomial Jan 31, 2025

@zl2tod no but you are looking in the right area

1.3.6.1.4.1.61513 Jan 31, 2025

@gsuberland out on a limb here, voice inversion scrambler?

Graham Sutherland / Polynomial Jan 31, 2025

K.C. Budd Jan 31, 2025

@gsuberland two distinct stages of ... negative feedback-loop op-amps with some sort of clipper ahead of it and a mixer after it and I get lost after that. But the real question is what happens when you take the weird morphed signal output of each of these similar-but-different-frequency circuits and connect them together via that N-channel MOSFET looking thing (?) at the output stage of the "B" circuit.

OOf. I wanna know what it sounds like! Probably some weird envelope of the high freq circuit being modulated by the same but 1/50th slower pattern of the "B" circuit via the FET.

BAH!

Graham Sutherland / Polynomial Jan 31, 2025

@phreakmonkey it is indeed an N-channel MOSFET.

Graham Sutherland / Polynomial Jan 31, 2025

@phreakmonkey and you are on the right lines there.

Cat 🐈🥗 (D.Burch)

@gsuberland Noise Gate?

Graham Sutherland / Polynomial Jan 31, 2025

Ronny Pries Jan 31, 2025

@gsuberland I don't know why (I can't really read circuits), so this is a deduction:

It somehow looks like a really feed-forward, feedback slew limiter. With the purpose of noise or distortion reduction?

Which could be useful for FM modulation?

Graham Sutherland / Polynomial Jan 31, 2025

Ronny Pries Jan 31, 2025

@gsuberland I'm looking forward to the solution! :)

William Jan 31, 2025

@gsuberland is it a modulator of some sort? Coupled via the Cgs on that fet?

Graham Sutherland / Polynomial Jan 31, 2025

@phenidone ooooooh you're VERY close there.

William Jan 31, 2025

@gsuberland I dunno. Looks kind of like a quadratic multiplier ("mixer" in the RF sense not the audio sense) around the fet but that signal doesn't get coupled out in the way I would expect.

Maybe a more dramatic effect like a robot-voice vocoder kind of thing?

I am limited by 5 hours of sleep and blinking at this over hotel breakfast, and also being dumb.

Graham Sutherland / Polynomial Jan 31, 2025

@phenidone "audio effect" would be in the right range.

I have posted a further hint:

https://chaos.social/@gsuberland/113921423149098863

Graham Sutherland / Polynomial (@[email protected])

Attached: 1 image ok, a few of you have gotten very close so I'm going to provide a bit more help

chaos.social

timoth‽Jan 31, 2025

@gsuberland It plays the solo from "Cherub Rock."

NosirrahSec 🏴‍☠️ guillotine enthusiast Jan 31, 2025

@gsuberland my complete uninformed opinion trying to brute force it by sight is a feedback loop? But...it's asynchronous.

At least it looks like it's not 1:1, so increases incrementally?

Graham Sutherland / Polynomial Jan 31, 2025

@NosirrahSec nope.

kwayk42 Jan 31, 2025

@gsuberland are those resistors with two values trim pots? Also the values that read as just 10, are those 10 ohm or a typo for 10kOhm?

Graham Sutherland / Polynomial Jan 31, 2025

@kwayk42 just 10R, and yes they're potentiometers.

kwayk42 Jan 31, 2025

@gsuberland I'm going to guess that this circuit is what gave Furby a soul

tyx Jan 31, 2025

@gsuberland
Not an answer (and I'm not an analog magic sorcerer), but the key thing is probably in these constant (and small) voltage buffers which will behave quite different depending on their type (RtR? Unipolar power? Which type of output?). Could be quite a lot of things including oscillations possible only with the power bus feedback/crosstalk.

Graham Sutherland / Polynomial Jan 31, 2025

@tyx nope, opamps are jellybean RRIO and don't do anything magic.

Aaron Jan 31, 2025

@gsuberland My first thought is, someone has been reading Douglas Self again.

Graham Sutherland / Polynomial Jan 31, 2025

Aaron Jan 31, 2025

@gsuberland Okay what the actual f***is up with that mosfet?

This isn't a Douglas Self meticulously engineered audio circuit, you're doing something utterly eldritch with the inherent substrate diode and/or gate capacitance.