Using a Fixed Filter Bank (with inductors) to achieve the sound of Eraserhead and 2001: A Space Odyssey

Hi, all - I posted this in my weekly thread for my show, but this could be its own thread. I recently took on a personal challenge to recreate the soundscape of 2001: A Space Odyssey and Eraserhead.

These tests limited the technology to what was available to both films at the time of production. Specifically, a few noise generators for testing, and a 1960s-era 907/914 Moog FFB. I didn’t allow for any additional EQ and even recorded into a 1970s Sony Microphone Mixer, so this can be considered as close to the studio options available to both films, and nothing more. The end result is pretty accurate and was a kick to figure out. You’ll hear the demo in track three of this episode:


This was a great episode, always such a fun podcast. Thank you for sharing your methods for getting a moog 907 affordably, I’m not sure it’s worth the risks but good to know it’s possible :stuck_out_tongue_winking_eye:


Holy shit; where have you been all my life.


Laughs! Yeah, I’m catching up on Lynch’s newly appeared filmography and not terribly disappointed in bringing about the end of the world. I mean, Mulholland Drive kicks all kinds of serious caboose! If it’s anything, I leaped ahead to the future to grab some of Moog’s newer FFB circuits in 2030 (they reintroduced the 914 with a spring reverb and a set of double inductors, and call it the 928) and I can confirm Trump is in jail at that point.

Thanks genuinely for the kind support and excellent in-joke!

Cheers, @niall! Thanks, friend. Enjoy the show.


This is great! I don’t know how I haven’t seen your show pop up here before but I’ll definitely be tuning in for more.

I’m about half way through part three of this so maybe you mention later on, but what noise generators are you using for your different colors before going into the filter bank?

1 Like

This is so cool! I really love the daisy demonstration at the end, one of my favorite moments from the movie.

1 Like

Thanks, @smbols! I actually don’t mention which ones used in the demo; it’s a red noise output from a minimoog — this would have been available to Lynch with a simple modification at the time, getting a red noise output from to the signal flow going into the filter, explained here, and just for reliability in the demo, I use a modern FSFX double rainbow for the purple, blue, and grey, which is a 5U mod of the Quantum Rainbow with an added dial for Quanta.

Kubrick would have had noise generators that provide those colors in test equipment if he didn’t go forward with synths, but his working partnership with Wendy Carlos suggests he probably had some analog connection — Though it was probably common to find an old Grason Stadler unit like this one. And I have to admit to immediately craving one of these Grason Stadler Audiometers. I don’t have any Grason equipment but they always intrigue me for being tube gear for noise generation. I also use the noise output of a Metasonix D2000, which is possible by playing a sustained gate into either noise drum. That’s the closest I can get to approximating a Grason tone.

Cheers, Tyler! Thank you.


Really enjoyed this. Quality stuff.

1 Like

Yes! Finally I can see the light, maybe even soon feel comfy having people over for dinner with Eraserhead noises in the kitchen.

Ive dearly missed a Moog Murf 105M I sold a couple of years back, it sounded friendlier and didnt have that low and low mid presence like its older grandpa showed here but a fun unit.

Had an eye out for a 2nd hand AJH filterbank for a while but none of the AJH modules Ive tried so far have that presence. Theres also the Verbos Bark that probably have more of that presence, but the Verbos is designed to have less crossover with steeper filters so maybe not as useful for these kind of tasks. Still curious tho. Probably overthinking things as usual but my Delorian was stolen, and these guys arent cheap investments.

Anyways, thanks for a great episode!

1 Like

Thanks @Above! I never have had an opportunity to use a Murf but they look awesome, especially the option to play a pattern through the EQ! I don’t want to dismiss AJH for filter banks, but it’s likely any eurorack module is physically too small for it to have the right components. Inductors themselves are each about the width of a thimble per frequency, so when you look at a 907 or 914 in person and flip it around you’ll notice the inductors themselves are about the size of the dials. There’s a nice cross-section of this in the Synth-Werks reissue. I’d love to get a Synth Werk model 10 or model 15 one day. Here’s their FFB which is component spec to an actual 914, image from the web:

I think the AJH filterbank is actually pretty dishonest in their title, calling it “inductor based” - Not sure why they need to bullshit there, many people misinterpret this to mean inductors are included. I’m carbon-based, for example, as are you. But if we’re trying to be martian I think we’d call ourselves martian-like or martian-esque, which is more accurate. The AJH might not sound exactly right to your ears because it just doesn’t have all the right components. Here’s how they say that in the description:

For the FFB 914 we have used Gyrator based active inductor circuitry which exactly replicates the vintage passive inductor

Not sure that’s true, just because inductors are not just a sound but also a dirty technology, they pick up feedback and EMF and other noise from electronics around them. I’ve noticed pairing a spring reverb tank next to my 907 actually incorporates feedback of the spring itself into the fixed filter bank. That’s not easily replicated. I’d say if you’re saving up for an AJH, just get a small power supply for 5U and invest in a properly sized inductor clone from Synth Werk. But avoid the 5U ones that don’t include inductors, otherwise you’re just getting a very fancy EQ with no crosstalk. For example, the dotcom FFB isn’t exactly well regarded, even though so much of their other modules are perfect. The YuSynth without inductors is CLOSE, as is the mos-lab. But sound is half of this. You want instability.

Now the Verbos Bark Filter, however, that’s something nice! To my ears it’s really very close to a vintage FFB. Not sure what they’re doing there with that one but it’s pretty damn perfect as a eurorack filter with resonant bands.

I’ll also say as much as I’m tempted by this wonderful model 15 reissue from Aion, my guess is their FFB 907 lacks the goods too. - inductors themselves run about $200 just for the parts needed and the depth of the module would need to be about six to seven inches in the back, and I think these modules are intended for skiffs. I don’t enjoy Behringer and Aion possibly changing the understanding of some of these classic modules with missing pieces that are so important to their sound. Especially since they’re offering a $3,500 model 15 in miniature now, too: - At that price, you should get a model 15 reissue cabinet with the few modules that can’t be recreated at small scale.

Appreciated! Thank you for a great comment!


This is all fascinating stuff. Do you mean here that if one were to run an enveloped saw wave (or whatever) sequence into the spring reverb and a different sequence into the FFB there would be some elements of what’s going through the springs “mixed” (for lack of a better term) with what’s going through the FFB present on the FFB output? I know that spring reverb circuits themselves are subject to picking up surrounding noise, too.

I was curious if there was a feasible diy version of this (with inductors) and it looks like Random Source has NOS Haible pcbs of this circuit with the option of using real inductors and an analogue realities build that does the same… but, alas, getting your hands on the actual inductors may be more time and money than I’m willing to spend. Interestingly, there is a euro version of the Yusynth version of this by antumbra here. Looks like a pretty big build and I wonder if it’s worth a closer look at. I’ve already got the slider version of the ResEQ, so the idea seems a little insane… yet… yet…

1 Like

Great question, and yes and no - It’s completely unpredictable. I mean the vibrations of the spring itself, carrying whatever audio signal, can show up in the fixed filter audio path. Radio does this too. I had an LFO disrupt some of the frequency bands. You move a module and it disappears. It’s partly because, I think, inductors are sort of like wound up antennas, though an engineer would likely disagree. I just mean they pick up feedback and cross talk from what modules are nearby, and spring reverbs do the same thing so they can be played off of one another (or drive you annoyed if you want a perfectly clean signal path) - benefits of older technology, really.

Looks like you found the Analog Realities link! Yeah there are a few opportunities to get an inductor FFB built new. Your mileage may vary but I’ve found it very satisfying to pursue. They’re like acoustic guitars of synthesis. Good research on your part!

1 Like

Checks out fine to me. A coil of wire (aka inductor) is the classic method of turning magnetic flux into electromotive force and vice versa.

1 Like

I once spent hours + hours troubleshooting (unwanted) noise and hum from a dual spring reverb module I built when the problem was so obvious - tanks or connecting cables were too close to beefy linear power supply. Kicked myself over that one. I’m not sure a full inductor FFB is in my near future but I will be keeping my eyes open for the slim chance of a group buy in the future, in which case I will be all over it! Any idea if transformers can also be used to exploit stray voltages and/or audio like this?

I was also interested in your thoughts on obtaining various noise flavors. My understanding in the past was that the other colors of analog noise were basically filtered versions of white noise and therefor avoided any dedicated colored noise generator but seeing people going ga ga over the quantum rainbow on the internet has me thinking there’s more to it. Am I missing out by just using white noise + filter? I’m having some serious noise FOMO over here :wink:

Once again, really enjoyed the show and will check out others as soon as I can :+1:


Hey @addamm - Thanks for the cool question. I don’t think you’re missing out really, except for considering the actual amplitude envelopes of different noise frequencies. The wikipedia article on Noise Color is really well made and documents this nicely:

You’re not missing out with a decent white noise generator. The Quantum Rainbow itself is a single noise circuit with different power curves applied to each color, so it’s really just a white noise circuit (damn good one, though) - This allows you to hook them all up into a mixer and dial between colors cleanly. What gets interesting for filtering, though, is having two completely separate noise sources. That allows for harmonics between the two. That’s when it gets pretty interesting, as you combine different noise and send each into a slowly change the volume between each you’ll find some great whistles and wind and other stuff with two separate noise circuits together in harmony. And yeah, you do get completely different percussion from white noise compared to blue or even pink. No FOMO needed, just experiment. And then there’s noise added into sample and hold or using as control voltage into filters and VCAs.

I’m not sure those transformer-based ones give the antenna-like interference you can get with an inductor, which is why I’m skeptical of the AJH gyrator one. They should call it gyrator-based. I would love to learn more about what a transformer can do in the way you’re proposing, too. “Exploit stray voltages and/or audio” sounds wildly fun.

Thanks for the kind comment about the show!


Thanks again for the thoughtful reply. I had not considered the above before. In my mind I’ve been thinking of white noise as “all the frequencies, all the time” when this isn’t true, it’s all the frequencies rapidly following and overlapping with each other… or some such :sweat_smile: I’ve just assumed white noise + white noise = same white noise.

I’ve gotten nice whistling tones by running white noise through a filter with some resonance but will have to mess around with mixing noise sources, too. Maybe noise into filter with separate noise source into filter fm… HMMM… I clearly have not given noise the respect it deserves :wink:


You don’t get the animation, but you can get some pretty fantastic deals on old White Instruments crossover EQs and they sound fabulous. I’ve owned a MuRF, an SND FB-14 and a couple similar products, but the White Instruments EQs always felt like a secret weapon.

1 Like

Thank you so much for this!

Now I’m scouring for noise making devices…

Out of curiosity, how useful do you think something like a Neve 1073 clone would be to approximating some of these sounds (specifically, the AML ez1073 which uses Carnhill inductors in the EQ circuit)?

1 Like

You mentioned a few times also over at muff, that from your tests the yusynth comes pretty close in tone and response to the non-linearities. Do you have an idea about if this holds true for the antumbra version aswell? Or is further compromise to be expected by the slimming down of parts?

Concerning the transformer question, the good ones are usually mu-metal shielded and the sweetnes lies rather in frequency dependent signal response, not unlike impulse responses when driven and further taken with the right gain staging. I love good transformers but i dont think i want stray noise wich would end up being hum i suppose.

1 Like

Hey Brad: Sorry I’m so late with this reply! The Neve stuff mystifies me. I think of Neve gear (or pine for it) the way others do about Buchla stuff. I don’t have any Neve gear at all, which might add to its mystique with me but they have two pieces, the tape emulator and the inductor EQ that really catch my mind often:

Probably will just commit to a 500 box at one point to get them.

These are US-based Neve products that I link (Neve himself lives here in Austin TX, where I live), I’ve also pined for the UK-based ones (like the 1073, or either company’s summing mixer) — with all of them I do wonder if their high-end means a lot of the charm of the inductor circuit has been engineered out of its sound. Would love an opportunity to answer your question, though.

If anyone would like to help, I’ll gladly take your Neve donations!