In honor of the new Vult Jorus (@modlfo’s new Juno6 chorus module), I present you with a 24-bit recording of the Roland Juno60’s chorus noise. The Juno6/60 chorus is notoriously noisy, and most gearheads who bought a Juno60 back before VSTs almost immediately bought an audio gate to get rid of the chorus noise when no notes were playing.
Now, in the digital future, we cherish the imperfection of analog gear, including the noise. So I recorded the noise my Juno60 chorus makes, filtered out the 60 cycle hum, and made three samples: Chorus I, Chorus II, and Chorus I+II.
I also crossfaded the ends of the loops so that you can play it looped.
Right, and it’s great to have the noise isolated so nicely; many thanks! I’m actually going to try it on a track soon.
buuuuut Jorus has adjustable speed and these samples are (necessarily) locked to the set I/II/I+II LFOs, so there’s that… Plus there may be interesting interactions between the incoming signal and the noise floor. And (if this is what @modlfo meant by “quantization”) I’d love a 256BBD (MN3009) option; Jorus, per the excellent docs, is modeled after a Juno-circuit pedal using 1024BBD (MN3007 or MN3207) which is going to have rather different–in some ways “better”–characteristics.
Obviously you’re right and modeling this stuff isn’t necessary! But I’m with @TroubledMind philosophically; the more weird, imperfect, surprising options that analog-modeled plugins have the better, as far as I’m concerned. I love how noisy and unpredictable BBDs are; about half of my (small) hardware modular rack is BBDs, so hearing that Leonardo had considered modeling the weirder aspects of a BBD chorus made my ears pick up.
In the (possibly eternal) meantime, Jorus is awesome and the looped noise samples are great to have, so thanks again to you both!
Here’s the full story. I have a full model of the BBD using 256 stages. But as you may imagine, a full model consumes quite some CPU. So I had an idea on how to get a close solution that would approximate the effect of the signal going through the BBD, considering that the signal is being “re-sampled”.
I had two options, make the chorus from scratch and taking some weeks designing, building, fixing the board, plus waiting for weeks until the MN3009 and MN3101 chips arrive… or buying a chorus pedal and use it to make the measures and get the BBD chips out.
I ended up buying a guitar pedal. But it turns out that the pedal used BBD chips of 1024 stages from CoolAudio. After making the measurements of the BBD I found practically no re-sampling effects in the audio signal. This is because the signal is being run through the BBD in a range from of ~100kHz to ~300Khz. After running the network analyzer, only signals above ~50 kHz were being attenuated. The only effect I could measure was saturation of large signals. The BBD chip it quite good, which from the modeling point of view is a bit underwhelming.
I ordered some MN3009 and MN3101 pairs. But they will take some weeks to be shipped. Once they arrive, I will try to measure the effects and check how my efficient solution stands.
For now, it’s time to move to other projects. Being Jorus a free plugin I’m not sure if I will spend many days more on it.
Thanks for the writeup, @modlfo! I find analog modeling at the level you do it endlessly fascinating.
I have a few of the Doepfer A-188-1s (all with MN300xs, not MN320xs) and I’ve found it very interesting to play with the aural differences of different chips clocked to the same delay time (i.e. a 256 at 1/4 the clock speed of a 1024). I’ll be very interested to see your rather more disciplined analysis if you get a chance to do it! It amuses me that better/more linear is often “worse” for modeling…
From context I’m guessing that the pedal was the TC June60? I think I remember reading that they were using V3207s instead of MN3009s. The trend these days definitely seems to be avoiding shorter BBD lines even for chorus and flange. I’ve sometimes wondered if shorter lines were used in older circuits because they were easier to make than longer lines, or because higher clock speeds were harder to generate back then, or for other reasons.
And by the way, I would certainly not want my weird noise/BBD modeling suggestions to dissuade you from moving on to other projects
ohh it’s a shame that I didn’t find the Doepfer BBD modules before getting the TC pedal. I bought the pedal because one of the reviewers said “as far as I can tell it’s pretty much a part-for-part clone”. The pedal sounds good. But I hope my small improvements in Jorus compensate for the lack of noise.
The Doepfers are fantastic but it can be hard to find them with MN300xs (instead of MN320xs) mounted and factory calibrated. The circuitry is the same and the re-calibration process for switching BBD length or switching from 30s to 32s is pretty simple.
But I hope my small improvements in Jorus compensate for the lack of noise.
I think that for almost all of your users almost all of the time (myself included ) the lack of noise is one of those improvements! Really looking forward to trying it out, Leonardo. Thanks as always for all of the wonders that you’ve brought to Rack.
I mean, I haven’t tried the module yet but from the docs it looks like what @carbon14 said; that’s pretty standard behaviour for HW devices that have M=>S or S=>S modes.
If you want it in mono, just run into L (or R) and take the L (or R) output, ignoring the other side.
If you want it to stereoize a mono signal (which was the core application of the chorus in the original Juno 60), do in L and out L/R.
If you want it in stereo (or dual mono), do in L/R and out L/R.
There’s not usually a use case (that I know of) where you want to plug in L and out L/R but have nothing come out of the R jack. But if you really want that, just deadpatch in R (i.e. plug it into a module that’s not outputting any signal) and it’ll override the normalization.