Simplest possible A(S)R generator

[Consul]

I finally had an epiphany today about an instrument I've been wanting to build. Other members might remember my talking about an electro-mechanical string machine, where stretched strings are pressed against a moving rosined belt by the player pressing the keys. After thinking about a number of ways of accomplishing this and not liking any of them, I finally started to wonder if I could have the strings always pressed against the belt (and thus always excited), and then use some really simple transistor switching to turn the sound on and off via a keyboard. And then I had my epiphany.

The basic idea is to make a keyboard that has 37 audio inputs, each one being gated by the 37 keys. (I picked 37 based upon the idea that three octaves can cover a pretty usable range of notes for studio recording purposes.) So now, I can build any tone generator that creates up to 37 notes (it doesn't have to use all of them, after all) individually, and then I can play it as an instrument. 19 Walkmans playing tape loops? No problem. A big cylindrical 37-stringed hurdy-gurdy with individual pickup coils? Why not?

So now, focusing on the keyboard itself, each key will need some kind of VCA, and some kind of envelope generator. Since I'll need 37 of each, I'd like to make it as simple as possible. I also want all of the envelopes to have common controls so they'll all make the same shape, set from a single panel.

I'm thinking this might be a tall order.

I found the famous one-transistor VCA schematic (I've attached it), which should serve well enough for that part, but once I start looking into even simple A(S)R envelope generators with voltage-controlled A and R settings, and the idea of keeping it simple goes right out the window. I'm wondering if I'm missing a trick.

As for connecting to the audio source, I'm thinking a 40-pin ribbon cable, same as parallel IDE disk drives use. They're ubiquitous and cheap, and would probably do the job over short distances. 40 pins should be able to do 37 signals plus a common ground.

I still have all 88 keys off of the piano I salvaged, and will probably use those for the keys, adapting the old piano keybed as well. I just need to figure out a simple mechanical switching system for gating the A(S)R.

Wish me luck. Any ideas concerning the control of attack and release on 37 VCAs would be much appreciated. Thank you!

 

[JohnRoberts]

If your'e going to make a one transistor gain element please use a JFET.

JR

PS: Not sure about leaving wheels constantly engaged (wear and tear).

 

[Consul]

I've been wondering if I could make something like this work using JFETs. Really, nothing needs to be voltage controlled except for the attack and release times. The VCA doesn't need to be a VCA.

And the wheel would be spinning only when it needs to be making sound. That's a separate problem and build, though.

EDIT: Actually, you might have given me an idea. I'll have to work it out tomorrow to see if it would work.

 

[shabtek]

if you are sending simple logic voltage a piano keyboard may not be worth the effort; an older salvaged electonic organ/keyboard would likely already have some kind of switching you could tap onto.
Now a real piano keyboard may make sense if there is more mechanical action going on than a simple switch--at much greater complexity.

I like the idea of real arco strings.
Moving parts is always an issue, tonewheels and tape loops are low, constant friction mechanisms I guess. The friction of aways on bowed strings could be too involved.

Periodically I consider building something like this, but never got too committed. The thought experiment always comes back around to an acoustic/mechanical generator: the gate is going to be the weak link in expressive possibilities--Though may have some benefit in controlling noise floor if an array of pickups is implemented.

If your platform is a real piano keyboard, have you removed the linkage/hammers?

What about connecting the lever of the key to  a bearing that the moving bow idles against until a key is depressed, whereupon the bow will be allowed to come into contact with string.

...This is an idea that could one could expend much time...multiple bows/motors, reversing rotation, portamento...

or take up violin
or a Midi string module.

cheers

 

[Consul]

Using the piano keys and keybed has a singular advantage: I already have those. Plus, the idea is to make one of these and then use it with a variety of sources, so I should make it good, right?

The piano has been completely taken apart. Even the hammers are separate pieces now. I still have the linkages, too, in a different box. I also have all of the strings but those became a bit of a tangled mess I need to sort through.

The idea of pressing a string up against a rosined belt was actually where I started. I just think this idea has more possibilities, plus it modularizes what would otherwise be a much bigger single build.

 

[abbey road d enfer]

if you are sending simple logic voltage a piano keyboard may not be worth the effort; an older salvaged electonic organ/keyboard would likely already have some kind of switching you could tap onto.

Beware that many electronic instruments use a matrix keyboard that needs decoding to produce individual switches.

 

[joaquins]

For the ASR, The key as a switch (not touch sensitive) This switch can from PS could charge a cap really fast, and the cap could be as big as you need for the slow release. A resistor in series with the cap charge it slower, one in parallel discharge it.
But this isn't voltage controled, UNLESS use VCR or fets for charging and discharging...
I think this may work and just 2 fets for key, one for attack and one for release, sustain as long as you keep the key down... Sustain pedal may be done affecting release fets...

For the VCA I'm with JR, another VCR fet and here you have a new advantage, only one part to supply for both pruposes. Opto would be nice but not cheap.

Wish it helps.

JS

 

[Consul]

@joaquins:

That is exactly the approach I'm planning on taking. I still need to do some experiments with the JFETs (I have some in my collection to play with) to see what their properties are, how much I need to think about matching them, and so on. So, I'm looking at five transistors (three JFETs, two bipolars), four in the envelope generator, one for the amp, plus a handful of resistors and caps. I haven't thought much yet if I'll need to buffer the output, but I can use a jellybean op amp for that and still stay pretty inexpensive and easy to build.

Thanks for the help!

 

[joaquins]

I don't know what do you plan to do with BJT, If you prefere a linear decay you can use a simple VC current source with a BJT, but harder to get full discharge I guess without some effort...

I prupose something like this for your schemes, Key/AR not sure about channel type but the idea, Key showed unpressed.

The simple VCA could work, CV thumb at fast attack could be avoided with High Z for both R and cap, Both R same value, RC for lower freq desired.
Input must be low level, the output may be connected to a summing resistor much larger than the input resistor. Noise performance won't be good, low signals, high Z not a good combination.

At the 2nd VCA you should use an opamp that could be discrete but I don't see the point, with U$4 you could buy 10 TL074 and still get 3 other opamps for summing and output buffer or ATT/REL CV buffer. The advantages here are many, low output Z and fet tricks, no thumb with CV at fast attak, higher output level with the gain at the opamp, so better S/N ratio. AC at the gate MUST be 1/2 AC at top of the fet! so choosing both resistor (CV input and the one in series with the cap) must match gain of the opamp.

JS

 

[Consul]

Ah, okay! I see what you mean now. I was going to use the BJTs for switching between attack and release, but you're just using the key itself for that. That complicates the mechanical design somewhat, though, as I'd need to make contact with the release switch before breaking with the attack switch.

EDIT: Actually, maybe I wouldn't. The cap on the CV output would likely stay charged until the other switch makes contact, right? If so, this would be an easy solution.

I could just put an inverter on the input to the release side, so when I release the key, that leg goes high. That's more electronics, but six inverters on a chip is about thirty cents. And all that is still not considering switch debounce...

More things to experiment with. Thanks a lot for your help!

 

[joaquins]

I don't see why make before brake here, trust in the cap...

I don't get you in the inverter thing, what input do you mean?

JS

 

[Consul]

The input to the FET controlling the release time, but it's of no consequence now. Thank you very much for your help!

 

[bruce0]

snip... talking about an electro-mechanical string machine, where stretched strings are pressed against a moving rosined belt by the player pressing the keys...snip

Someone may have beaten you to the punch...
http://antiquitymusic.com/wheelharp/

But for the price you could hire a string section each time you needed them...

 

[Consul]

Yeah, I already knew about the wheelharp. I don't like how it sounds, though. Way to screechy for my tastes. Really, my idea centered around making a string synth where the sound source was actual strings, so there would be chorus and all that as well. But I still think this latest idea is better.