Tuesday, February 26, 2019

Building a Low Cost 10V reference for Audio

UPDATE: The Balanced Modular referred to in this post is now done and working!  For Part II please go here.


I've been back at this audioDIY for a few years now and I am still looking for a really good balanced modulator.

A strange sentence right?

This is partially sentimental, my brother and I built a 1492 based PAIA 4710 Balanced modulator when we were very young indeed (12 and 14?).

Long Live PAIA!

I clearly remember how much I liked that module. The damn thing sounded frigging great! Metal klangs! Space gun laser sounds! Weird robot voices! Impress your friends! Baffle the girls!

Looking back, many of the modules in the PAIA 2700-4700 series sounded--well, bad, but not this one. Maybe my mind is playing tricks on me but the 4710 sounded better than anything I have in my rig now.  I have to admit it: I'd like a good, clean B-M.

So just rip off the 4710 schematic? Maybe...but 1492 IC's are hard to come by and hard to design around I think. The power (+18V, +/- 9V) isn't close to what I use in my DIY modular. Whatever.

I've tried recreating the good old days with AD633's as well as a Thomas Henry designed 3080 RM from his 3080 book but neither sound all that good. Also I fabbed up this Korg MS20 style RingMod which is cool but is pretty choosy about the incoming wave forms (ramp works best? Pulse?).  You can see my ring mod webpage here.

Hunting around more I found a AD533 based Balanced Modulator in Electronotes Preferred Circuits and thought this one might sound good!!

A lot of energy for the EN circuit is spent getting rid of "bleed through" which is the stumbling block of the latest B-M's I've tried to create. I'm not sure I use the right audio term for that. Crosstalk?  Here's what happens: The X or Y audio input can be heard, unadulterated, at output.  The digital (VST) BM's I use with Ableton never have that problem.

At output, I want pure, robotic, squeaky clean B-M sound. Looks like this circuit makes it happen.

For better or worse, the EN circuit requires an Analog devices 10V reference IC which is no longer made.  But! wait a minute, I already put together a working +/- 10V reference for a log converter I built several months ago. You can see that here. I will use that.

10V Precision Reference, finished, tests working

The ref circuit is simple.  It uses an LM397 zener at its core. The idea: you can heat up the 397 zener and nevertheless it stays pretty close to spec. So we have e-z super simple temperature compensation here, always a tough part, in my experience, of a good voltage reference design. Plug the Zener into the ADJ of a 317 and 337 v-reg, some glue trimmers and resistors, and the requisite coupling caps, and ahoy, you get a decently, pretty temperature independent steady voltage source.....

Here's a link to the datasheet for the LM329s.  I got some surplus for next to nothing, they are still out there. Can you make a more stable reference, put your circuit under a heat gun and have V++ stay steadier?  I'm sure you can.  But for a 1980's era Balance Modulator, I figure this design is good enough.

More next time: when I actually see if I can hook this to the EN Balanced modulator and make some sound. (update: works! here.) Until then, don't breathe the fumes.

Monday, February 18, 2019

Korg MS20 Filter--the Why in AudioDIWHY.....

OK this one drove me crazy. 

This should have been easy, we know that Rene Schmitz stuff sounds good (I have already built his "FEITW" ADSR) and I thought, let's try his MS20 filter.

So: Did the Eagle design based on what's on Rene's site, put together a board layout, and sent off for fab. 

This is a pretty simple circuit and from a good looking schematic, from a smart synth DIY maven, why wouldn't it sound good?

You know how I sometimes post: worked first time?  This one most definitely didn't work first time. Neither filter worked for beans for at least 3 evenings--and the second one took me more like 5 evenings to get working. This got me to wonder, why not just buy someone else's filter design? Does a kit qualify as DIY (of course it does....but it's not enough torture for me?)

I used a 2x dual concentric pot PCB for filter #2 which I cut down from this PCB.

OK I built two and at first both builds passed audio but both filters sounded--wimpy. Almost like a cheap stereo tone control, not that nice fat MS20 sound with dirt when you crank up the "Peak". I know Rene's designs are high quality, and a skinny sounding filter must be on me.

Building Filter #2....

After probing around a lot with my scope I ended up tracing this to a stupid mistake: I copied the schematic wrong into Eagle!  I drew up a critical voltage divider incorrectly, which was causing a journey to 6db cutoff, wimp ass sounding low pass land. But the 8-28-18 PCB is still useful, although you have to do a trace cut and some kludge wiring to get it sounding MS20-like.

But then again: what would DIY be without trace kludges?

With all that crappy advertising, you can get my PCB design (the one needing a trace cut, 8-28-18, and fixed yet untested version, 1-9-19) here.  My website includes BOM, wiring suggestions, and specifics about the trace fix.

MS20 filter #1 finished, no decals yet, (I think) it sounds Rene-errific

Great! we get that crunchy peak sound, fat lowpass and nice Korg growl!

Next I turned my attention to the second unit (two VCFs so I can HP + LP = band pass? Two on the original MS20 right?) But even with the trace fix, the second MS20 filter had a strange problem--when I cranked the CV up the audio output went completely dead.

Damn! It took me 2 more evenings of comparing the working VCF to the broken one to figure out problem! I suspected a parts placement error, but after checking twice, no, everything is OK.  Bad IC?  No; swapped everything and no change. Could I live with a VCF just not eating any CV above about 4V DC?  No, not really.  And besides, #1 worked, so #2 needs to work the same way.

Arrg!  Too much shotgunning! 

Ended up tracing this to a wiring problem with one of the grounds:

Filter #2, turns out, the normal is the fault--thanks to E-M GRUMBLE for the suggestion....

Goes like this: for an MS20 OTA filter: if you want high pass, stick audio into one end of a critical cap. If you want low pass, ground the same end of the same cap and put your signal into the circuit's main input:

Not having room for a switch in a 1U Frac, I used a normal to ground the cap, and the normal had a cold solder joint.  Yow! So when I put audio into the "LP" input, it would cut out at high CV.  If I put audio into the HP input, it wouldn't work at all. Not at all the way I would think this issue would manifest but there you go.

OK with all that sorted, I did a recorded the quickee demo, here.

Second MS20 filter works! Filter 2 has 2 mod inputs, I was going to use this as the primary filter of the two.

Five evening later, I have to wonder, was all of this worth it?  There are a ton of PCB's for MS20 filters already out there, did I really have to do my own PCB? I doubt it, really. Hell, I could have just bought one of those MS20 repops, and if it went south, bug Guitar Center for help right? Maybe not?

I think I just like to torture myself, building machines that I go crazy trying to fix over the course of 5-6 evenings. Welcome to the world of DIY.

Hey!  They work!

I realized something important: I hate, hate, hate haatteee it when things I build don't work. More than I should. So why DIY at all?  This is like being into sailing but hating water, right?

In my rack--panels redone using FPE

I'll have to ask my girlfriend, who is a mental health professional (really!) why this is, because really I have no idea. Was I dropped on my head at a young age?

Granted--too many fumes. I need a break.

Monday, February 4, 2019

ReneDSR: Fastest Envelope in the West, Finished; Sound file and Build Notes

Back Again!

I needed more ADSR's in my modular so I built some based on a very cool design from Rene Schmitz.....You may want to read part I of this post here., where I create the PCB, discuss what "Tau" means, show you useless bench build photos, and so on.
Left to right: finished modules: ReneDSR, Return of Son of ReneDSR, and Son of ReneDSR

Not just Decay....."Fastest Envelope in the West"--by adding a switchable "YA" cap the entire ADSR gits along, little doggies

Does capacitor size matter?  

That's the "rayson d'tre" for these builds, but?? From last time, the basic design of these sorts of ADSR's sends gate logic to an RC circuit.

Good news: 

OK we've already determined that the size of the resistor and cap (the resistor being a pot and/or a resistor network) determine attack, decay, and release speed. So, should we build these with large caps and small pots?  Small pots and large caps? I think the jury is still out, because to me it all feels and sounds good. You can mix and match caps and pots here and have fun; the sun will still come up the next day.  And, set up right, either combo can provide mix-cutting, DJ boogeying, MDA popping sweaty-bodies-dancing-w/-body-odor bodacious snappy synth decays.

I tried to create a very simple demo for these modules, and you can hear sound samples here.

About the sound sample--other than the WAV loops this is all my modular with ReneDSR:

00:00: multitracking 6 mono notes on a DAW (Ableton Live 10). Attack time is set to minimum; release is set to max; it's a very fast attack (I didn't measure it on a scope, but you can hear, IMO anyway, it's a reasonably fast attack) with 23 second decay for VCA, and same for VCF.  I put a lot of rez on the filter to make the sweep more noticeable. You're hearing a 10uF cap and 2.2M audio taper decay pot creating the ever import tau (see the last post).

From this I get an MiniMoog's long release (>20sec) or maybe a Prophet 5-ish sound. I am happy.

00:33: Dual Osc bass line, maybe a bit of a Stevie Wonder? I always like that; Stevie and the late great George Duke are my all time favorite synth bass players. I remain happy.

00:55: Sequencer patterns. This is 2 sequences overlapped, making heavy use of ReneDSR's fast decay, without which the moving patterns sound mushy. I am not a big Gorgio Moroder sequencer guy but I know a lot of synth freaks out there are and for this application the ReneDSR performs well.

01:44: Single Oscillator to VCF to VCA, with SonOfRene controlling the VCF and Return of Son of controlling the VCF cutoff.  In many mixes I find that a single OSC bass sounds sit better than dual or 3 osc basses. Maybe a bit Roland SH series sounding. Why not?

The Roland SH-5 has one of my all time favorite ADSRs.  From the shop manual, it's a pretty simple discreet design from those crazy and creative 70's Roland dudes, so a PCB maybe for a later build?

The Return of Son Of uses a 2.2uF cap and 3x dual gang 1M audio taper pots. Each switch selects very good control of a 1M audio taper pot; or, tap off the wiper of a 2M pot. This allows for a good fast attack and decay with a longer release as needed.
Return of Son Of RenayDSR Switch Wiring.

Another build idea: In any Rene 2.2MB pot build, you might want to put the 2.2uF cap on a DPDT switch with say 10uF cap for longer decays and releases.  Or for 1MB pots, a 4.7uF and a 22uF.  Experiment and have fun! You get the idea.

Front Panel didn't come out very well for ROSO, lumpy and bumpy, but it's good enough.

The other 2 units--ReneDSR and Son Of ReneDSR--are pretty much a straight builds of Rene's two design variations on his site.  The 1U device has 1MB audio taper pots for ADR and a 4.7 and 10uF cap on a DPDT switch with 100ohm resistors in series. The second unit was featured in this post, and uses 2.2MB Audio taper pots, which can be a bit hard to find.

One more tidbit: Rene's design calls for CMOS 555 timer chips, I bought about 10, and couldn't get a single one to work in the 3 circuits you see here.  A standard 555 worked every time. So they may be something wrong with my PCB, or maybe I got a bad run of CMOS chips. Whatever--gotta move on.

Overall, a fun build, and more good sounding ADSRs were very much needed in my rig.

So if you're looking around for an EZ ADSR that performs well, and is fun to build, this may be it. FDITW is highly recommended.

And as always, don't breathe the fumes!

ProMicro HID Keyboard Emulator

Quick one this time. The Arduino ProMicro (examples here and here ) is based on an Atmel 16u4 MCU and has HID keyboard emulation ready to go...