Extending the soundscape of a hexaphonic guitar pickup

[sorveltaja]

Today I've been mostly scratching my head, about how to draw a 'bus board' for the blocks:


Idea for that came from the Elector vocoder. I'll draw 3d-model of all the boards(blocks), as they are built.
Usually on more complex projects like this, it's the best way for me to make sense, by 'simulating' the objects, that have actual physical dimensions.

[sorveltaja]

Boards are almost ready for transfers:
  [ You are not allowed to view attachments ]  
  [ You are not allowed to view attachments ]  

First is one of the eight band pass filters, and second is one of the four companders. I did some double-checking, and it's worth the effort, as there were some obvious errors.
Now they should have been sorted out.

When taking a break of the drawing/checking, I got seriously distracted, when looking at some other vocoder schematics. They have some nice extra features, such as separate level adjustments, inputs/outputs for each BPF/channel, which could then be cross-linked in any order, using patch cables, to make it behave like ministry of silly walks.

But back to reality. At this stage, that kind of options would make the process a lot more complex. After all, the idea is to first build a bare bone -version(as I have already cropped all the extras out of the layout/schematic) of the vocoder, to get a grasp of how this specific breed(Paia) works.

I haven't yet found a way to draw the 'bus board', or 'motherboard', where all the blocks could then be inserted using headers. It's probably because at the moment I don't have too much of circuit board to spend.

I have a bad habit of drawing the layouts basing of what I currently have on the shelf. Yes, it's unnecessary subconscious limiting factor.

When going back to the hex pickup -project, there were numerous versions of different layouts/schematics/stages, to get closer to the desired results.

So I think, that part of the learning is about remembering, and revisiting the ideas, that once worked. For me, at least, it requires some deep digging, to access that mindset again.

But enough of that jabbering. I'm going to place an order for the components/stuff, that might be needed in the project, plus some more. This time the circuit boards will be 0,8mm ones, as they are easier to cut, than previous 1,6mm ones.

In the end, off-topic, as always. The new guitar, that I ordered, should arrive next monday. Can't wait to get my hands on it. If it's as good as expected(no broken neck due to transport, or other major defects), then it's time to restore the hex pickup-hassle back to the old guitar.

It was very easy to disassemble, but takes plenty of testings(and cigarettes) to get it back in working order. But the sound, that it gives, should make it worth the efforts.

[sorveltaja]

Bus boarding in process:

  [ You are not allowed to view attachments ]  

Really good concentrating exercise, I must say. I started by adding power connection traces, but soon noted, that It's better to leave them last. That way the signal paths have traces with less jumper wires.

Whole layout is about 260mm long, so it is too big for me to make it in one piece. Once the drawing is done, I'll divide it to 3 or 4 parts, which could then be connected together using pins and headers.

[sorveltaja]

Drawing of the bus boards first version is almost finished:

  [ You are not allowed to view attachments ]  

  [ You are not allowed to view attachments ]  

  [ You are not allowed to view attachments ]  

  [ You are not allowed to view attachments ]  

At this stage, I think I'll probably use angled female headers for bus board connections. At least it should make testing of the separate boards easier, by using the same wires as on the breadboarding.

If testing goes ok, they could then be connected together by adding male pin headers between them. For that, I'm going to order longer ones, as the ones I have in shelf, are too short.

But first things first. Next I'm going to build one of the band pass filter boards, to verify that the layout works. Then same for one of the compander boards.
The rest of those 'blocks' use same layouts, but the component values differ.

As the essence of the vocoder forms, lo-fi audio samples are on their way.

[sorveltaja]

Today some cross-checkings:

  [ You are not allowed to view attachments ]  

  [ You are not allowed to view attachments ]  

First are the band pass filter sections, and then the compander ones. It is important to connect them in certain order as 'pairs'.

Bus boards were also checked, but building them comes later, once the numerous 'sub-blocks' are built, and tested.

---------------------------------------------

Few words about current side project. Monday I got the ordered guitar, and oh boy, it needed plenty of adjustments to make its playability better. Not only that, but one part needs no other than machining.

I dont even remember, when was the last time, that I milled a tough-ass steel part. Luckily I have some carbide bits for that.

But yeah, that guitar, as it arrived, wasn't really user friendly, although it's on the mid price range. I have a feeling, that there lurks a quite decent instrument, once it has the loving care it needs.

I'm aware, that there are some strict rules, concidering the warranty. But meh, it's mine. Once it entered my apartment, it sure ain't leaving any time soon.

[sorveltaja]

Finally, first ones ready for testing:

  [ You are not allowed to view attachments ]  

I think I'll start with band pass filter(on the right side). With component values that it has, the 'peak' should be at ~4900hz.

Then the compander board, which, to my understanding, acts like a dual channel... something. As it is now configured, one channel's output level controls the other channels output/gain/level.

Still, there is something about it, that I can't get my head around. Maybe it's simpler than that. But we'll see, once I start testings tomorrow.   
-----------------------------------------------

Regular off-topic ramble:

This week, I've been mostly fiddling with the new guitar, to make the best out of it. All the most tedious things are almost done, of which worst was trying to mill the locking nut, that had convex surface at the bottom, when it should have been straight.

Even with carbide milling bit, it was just too much for my small, humble mill. (Obvious) conclusion was, that the part is hardened, or is made of some grade of stainless steel(which I doubt in this price range).

Anyways, quite a lot of material(~1mm) needed to be removed, and the only way I could think of, was to replace the carbide bit with a 20mm diamond wheel, and humiliate that tough part by taking tiny shaves at a time.

Hours later, major pain in the rump -stuff was done. But yeah, now the bottom of the locking nut is straight, and allows a lot easier way to adjust the height of it, by using shims.

It's a part of the Floyd Rose tremolo system, which, when properly adjusted, keeps the instrument in tune, no matter how much string bending, or whammy bar is used.

That's one of the main reasons, why I have wanted a decent guitar with that kind of hardware for years.

[sorveltaja]

Testing of the first boards was rather quick. Band pass filter has its peak at ~5000Hz, which I found out just by dialing the signal generator's frequency, until max.voltage was reached.
I used a scope to see the results, although same might be done using a multimeter.

Compander was tested by feeding sine wave to both inputs, of which one is for microphone, and other is for instrument(keyboard/guitar/synth, you name it).

When trying to clarify/simplify(at least for myself) of how the vocoder actually works, it's actually quite simple principle: microphone signal controls the instruments signal.
Both inputs have identical band pass filter -sections.

In practice, let's take a keyboard, that plays a sustained chord, which is then fed to the instrument input. Then some words/singing in the microphone input.
When the vocalist says 'boo', or 'ess' in the mic, that signal goes through the band pass filters, and tells the instrument side to let through only the very same frequencies of that keyboard chord.

But enough of that. After some etching and drilling, rest of the boards are on the making. Compander boards are already populated, but again, I had to cram some components in, due to making the layouts/boards as small as a stamp :

  [ You are not allowed to view attachments ]  

Band pass filter boards ready for the parts. I managed to salvage all the needed components from the previously built boards:

  [ You are not allowed to view attachments ]  

If and when all the 'blocks' are built, tested and working, then the aural testing -fun should be a bit closer. Some bus boarding, and other supplies need to be ordered before that, though.   

[awemawson]

Surely those electrolytics don’t need to be 250 volt working   Rated at a lower voltage they’d be much smaller and your board less crowded.

[sorveltaja]

That's a good point. I haven't paid too much attention for the voltage ratings.

[sorveltaja]

Band pass filter -boards are now finished, and tested. Regardless of the clumsy looking capacitors, they work as expexted:

  [ You are not allowed to view attachments ]  

Bus board for companders. I'll make the rest for others, once the ordered pc boards arrive:

  [ You are not allowed to view attachments ]  
------------------------------------------------------

Generally, the paper(magazine, or whatever I have at the moment) that I use for the trace transfers, gives acceptable results, but could be better.

Lots of pitting, and tiny fractures in the traces. To this day, it hasn't really been too big of a problem, but I always have to check the continuity of the traces. Once drilling is done, I clean the board thoroughly, and spray a protective 'solder-active' lacquer, to prevent future corrosion of possible weak points.

Yes. there are 'press-n-peel', and UV-based products, which could be a good choice for verified pcb layouts. But for highly experimental, 'trial and error' -stuff, they tend to be rather costly.

There are lots of different ways to do the cheapo laser printed diy-transfers on the net, but it all depends of the paper and/or the method, that is used. For example, if one sees a video on Youtube about the subject, it doesn't mean that same kind of paper is used on mags/adverts around the world.

So far, I've found, that there are at least three ways to adapt the process for one's currently available paper types: ironing, water-acetone method, or combination of both.
None of the previously mentioned methods produce perfect results; but after some practice, they should get the job done adequately.

After all, the point is to not be too dependent on certain commercial brands/products, as they come and go.

Maybe it's time to end this off-topic-ish ramble for now.

[awemawson]

Years ago there was a thing called a Dalo Pen, that had a thick ink that was resistant to ferric chloride etchant. A bit of a pain to use but a very quick way of making simple PCBs

They were very expensive but I found that Layout Blue was etch resistant so I refilled marker pens with it.

[RussellT]

I've tried a few of the internet methods using photo paper etc, but I decided that buying the press and peel paper wasn't as expensive as wasting copper clad board.

Russell

[vtsteam]

Pretty amazing project sorveltaja, and extremely well detailed.     

[sorveltaja]

Awemawson, yes I use Decon Dalo pens occasionally to patch weak/broken transfer traces. There is a small story, considering the Dalo pens: back in the 90's I went to electronics shop, and wanted one of them, although being expensive. When all the stuff was paid, sales person did put them in the plastic bag.

After some months went by, and there was need to use the pen, I opened the box, and it was full of Dalo pens... dozen or more of them. I thought of returning them, to be fair, but didn't have the receipt anymore.

I still have some of them left, good as brand new. Bit of a history, as that seems to be discontinued product.

Russell, thanks for the suggestion. When thinking of it, Press-and-Peel(or UV-method) could well work better, when the traces are really thin, and high precision is required, like on the layouts, that use smd(surface mount device) ic's/components.

Vtsteam, thanks for the heads up. It is my aim to document the project(s) as well, as I can. That way it makes more sense to me, of what I'm doing. Hopefully it makes projects like this easier to approach, even if one doesn't have a massive experience in electronics(me neither), but is still interested in experimenting.

In the end, the compander bus board is now finished, and connections tested:

  [ You are not allowed to view attachments ]  

The four pin header on the right side is for the op-amp -board(to be done), that sets the output level using a pot. The pot on the upper side is to adjust the 'vocoder threshold'.

At this stage, further testings have to wait, until rest of the bus boards are made. Otherwise the results(and amount of wires needed to connect all together) might end up being a great confusion.

Ordered pc boards should arrive in next monday. All the 'blocks' are finished, except the output op-amp. Next onto that.

 

[sorveltaja]

It is always a positive surprise, when ordered stuff arrives earlier than expected. That's what happened, and now the project is at this stage:

  [ You are not allowed to view attachments ]  

Some wirings and the output op-amp board needs to be made. After that, it's finally time to find out, how all those circuits works together.

[sorveltaja]

Today i've been mostly testing the vocoder, as a separate blocks, and as a whole. Everything should work, but there is no actual vocoder effect. I don't expect to get the sound of full blown 24-channel one, as this has only 8 channels.

Still, there should be clearly audible effect, like on this video of the same Paia vocoder. Demo starts at ~1.10:

Companders work, although they give rather odd waveforms. Not sure why that is. Maybe that's the part of them, that I can't get my head around.

Band pass filters work. To be sure, I did also some aural testing on each of them, using a signal generator's 'random noise'(not really random, more like a loop, which can be heard at a low frequency like 1Hz). They all have different, distinct audio ranges.

One thing that came to mind, was to replace the companders with something simpler, like Led-Ldr(Light dependent resistor) pairs. After a bit of search on the net, conclusion was, that they aren't generally used for a vocoder.
All the diy -designs, or versions that I'm aware of, use IC-based solutions for that.

The Led-Ldr pairs are used on some diy compressors, phase shifters and so on. Of course the linearity isn't at the same level with the IC-based 'level adjustment' -devices.

Despite of all that, I think it's worth of testing, how they could possibly work for this purpose. The form will probably be like on an envelope follower from Craig Anderton's "Electronic projects for musicians":

  [ You are not allowed to view attachments ]  

Basically, the above Led-Ldr envelope follower works as follows: an audio signal is fed to the input, and goes through the op-amp(IC1), which drives the Led portions of the optoisolators(OI1 and OI2), so that the leds light up, roughly following the input signal's amplitude. Amount of that light tells the Ldr, how much it can 'open', by lowering its resistance.

So, plan is to start by cropping the above schematic's parts count to bare minimum.

[sorveltaja]

To see, if the compander bus board could be used for led-ldr pairs, I made adapters for the pin headers, so that it's easier to connect the breadboard to the rest of the circuitry:

  [ You are not allowed to view attachments ]  

I did roll some temporary optoisolators to test the concept. They aren't lightproof, and need to be covered with something like a cardboard box, when testing. I had to order more ldr's, as it is probably better to have the same model for all the eight pairs:

  [ You are not allowed to view attachments ]  

First I connected some leds to compander bus boards' header pins(which are connected to band pass filter's(bpf) outputs), and it was like a christmas tree. Frequency sweep from 0 to 6000Hz showed, how each channel responded to certain frequencies. Adjustable gain, first stages(for mic and instrument) had expected results; when turning the gain down, leds dimmed correspondingly.

So far, it looks like the bpf's outputs itself might have enough drive to make the leds light up at the usable levels.

One part of me tells that the signals should be buffered, before they hit the leds, but not yet; only if there are several, noticeable impacts to the rest of the circuitry, then it should be done.

Version 0.1 layout of the possible compander-replacing led-ldr portion:

  [ You are not allowed to view attachments ]  

Mic input signals drive the leds, and should ideally be in a dc form. Diodes do some rectifying by allowing only positive peaks of the ac-form audio waves to pass through, and then the 47uF caps smooth out rest of the signals.

Instrument input signals go through the ldr's, that act like variable potentiometers. Again, microphone's signal controls the intrument's signal output level.

As can be seen in the above layout, 'outputs' of the ldr's are fed to the final stage, being an op-amp. They are simply tied together, which shouldn't work, as all of them will fight against each other, to get a place in the sun. Buffering seems to be the key, and is added later on, if definitely needed.
 

 

[sorveltaja]

First aural test, using previously mentioned circuit. Currently 5 of 8 channels are connected:

  [ You are not allowed to view attachments ]  

Resulting audio, rather muted and distorted, with little of an eq added to clarify:
  [ You are not allowed to view attachments ]  

Audio files used to drive the vocoder: a sample from an audiobook "Gods of Mars"(freely available from archive.org) for mic input. For instrument input, an audio file, using the hex pickup, recorded back when I made final adjustments for it.

It's very tricky to get the adjustments to the point, where the effect is most clearly audible. Changes in input signal levels affect the result also.

I'm thinking of testing the inputs by using compressed audio signals, to get more stable action. First it could be done by compressing the audio file's signals, and if it helps the overall performance, then build a compressor for each input. But we'll see.

After all, I am very much surprised, of how that rudely simple concept, where I just threw some components to breadboard, gives signs of life. Absolutely worth further testings.

[vtsteam]

Amazing. And fascinating. Beautifully presented.

 

[sorveltaja]

Vtsteam, thanks.

The ordered ldr's arrived, and now all the 8 channels are connected. First sample was made using the same hex pickup track, to feed the instrument input. Control input (Mic) was fed with 2Hz(120bpm) square wave. Raw output, so there was no post audio processing(who needs a drum machine, sequencer, or a keyboard anymore):

  [ You are not allowed to view attachments ]  

This one was made by feeding the same clip from previously mentioned audio book(speech) for mic input, and 50Hz square wave for the instrument input(or was it vice versa). But anyways, that required some heavy eq'ing afterwards, but still lacks the definition:

  [ You are not allowed to view attachments ]  

So far, as this vocoder is a bare-bone version, it has already exceeded my expectations, by opening a whole new audio landscape for experimenting.

Some aspects, that need to be worked on:

Compressing the signals in audio files by using an audio editing software(like Audacity), especially speech, is more complicated, than it seems.
I'm not an audio engineer, and just fiddled with some settings, but it's way too easy to add unwanted distortion to the original signal, by 'ballparking'.

Another way to approach the problem, could be to use a soft-clipping circuit, to squeeze the speech signal between certain values, and adding needed harmonic content to it at the same time.
To realize that, something like 4049 cmos ic(which is known for its properties to clip in tube-like manner).

I should have some of those ic's in the shelf to boot, so it's going to be one of the subjects of forthcoming testings.

Then there is one thing, that I noticed earlier: the band pass filters don't have equal output levels, even at their fundamental, or 'peak' frequency ranges. I have no idea, if that is what they are supposed to be, but I'm going to find out, if they could be 'leveled' in a simple way, like using trimpots, that I have.

[sorveltaja]

Not so much of an aural testing today. Instead I've been drawing a breadboard version of a compressor, that uses led-ldr pair for gain control. If it makes noticeable improvement for speech processing, more about it later.

I also measured outputs of the band pass filters, one by one. 0.1 volt sine wave was used, at each one's peak frequency. Power supply at -15/+15 volts. Results:

  [ You are not allowed to view attachments ]  

Quite a lot of deviation, to say at least. I started modding with the 560p one, to add more gain for it. Schematic of the mic-instrument pair:
  [ You are not allowed to view attachments ]  

First I removed resistors Ra and Rb(both 4.7k), and replaced them with 10k multiturn trimpots. Still not enough, so I detached one end of the Rc and Rd(2.2M each), and added 2.2M resistors in series with them. Then I was able to get the max. output to 12.0 volts, which might be enough for now.

I just don't understand, why there is so much deviation between bpf's outputs. 560pf one had 072 dual op-amp, and I replaced it with 082, to make sure, that the ic is working, but still same results.
Those bpf's have identical components, besides frequency setting capacitors.

Modifications are quite simple to make. On the right side is modded one:

  [ You are not allowed to view attachments ]  



 

[sorveltaja]

Each of the band pass filters were adjusted to have ~12 volts(p-p) outputs. It alone made audible improvement to the overall output.

Just today I read from some forum, that this(Paia) vocoder, has an inherent property in form of deviation of the bpf outputs. When I was measuring the compander's output levels, it wasn't so dramatic, so that could the reason, why the designer of this device(Craig Anderton) did let them be, as they are.

But now it's been taken care of, and the previously mentioned compressor for speech signals may not be necessary at this point.

Optocoupler pair, as a module, and built one(1 of 4):
  [ You are not allowed to view attachments ]   [ You are not allowed to view attachments ]  

There were two odd things, that I stumbled upon. First one was, that I had to swap mic/instrument connections for the led-ldr pairs 2-4-6-8. Otherwise they wouldn't give any output. Probably some logic error in the bus boards. I didn't want to spend too much time to find them(errors), instead it was easy to fix the problem, when they were on the breadboard.

Second one was utterly strange. When I built the above led-ldr pair as a module, there was usual testings, to see if it works, as expected. Procedure was this: feed a low frequency square wave(1Hz) to the led inputs, and higher one(audio range) freq. signal to ldr's. That way I expected to see, how the slowly flashing led turns on and off the signal, that goes through the ldr.

No such thing. The signal went through the ldr, and the flashing led did affect it only barely visible amount, when looking at the scope. I used signal generator for the signals, and dialed the levels up and down, but no visible reaction.

There I was, thinking: "bugger! this doesn't work!". Then I did the same tests with 'previously aurally tested and working' -setup, that was on the breadboard. Exactly similar results.

My conclusion was, based on level measurements, that it shouldn't work, but when connected to the rest of the circuitry, and listening to the output, it works, for some reason.

There must be some 'plan 9 from outer phase' -audiochemical processes going on.

But seriously, the way they work, isn't obvious(at least for me). I would most probably have skipped the whole concept, if I was counting on my skills to find out, what's happening, just by looking at the numbers or curves.

In the end, no off-topic ramble this time.

Makes one wonder, why the optoisolators aren't used on diy-vocoders(as far, as I'm aware of).

[sorveltaja]

All the optocoupler boards are finished, and the whole device works, as expected. Now it's time to look, what needs to be added/modded, to get the most out of it.

First one is simple; 100k potentiometers(left from the arbitrary waveform generator -project) between bpf- and optocoupler boards, to adjust each channels levels individually:
  [ You are not allowed to view attachments ]  

It works like an equaliser, but has rather subtle effect. That is probably because the leds on the optocouplers could use more 'juice', to make the ldr's more reactive/responsive. The outputs of the band pass filters, although their gains are cranked close to the max, are still lacking the grunt. But then again, they weren't originally designed to drive leds...

I discovered a possible trick to increase their current levels, while testing earlier with a single potentiometer. So far, I can't tell, what the cost will be(like roasted op-amps, and such).

As always, there is only one way to find out.
   

[sorveltaja]

The previously mentioned trick to increase the current levels for the leds didn't work well at all. Leds flashed a lot brighter, but the audio output was not usable, being heavily distorted and muddy.

There I was, reminded of the thing, that everything doesn't have to go to eleven(like Spinal Tap). I encountered the very same thing, when building the electronics for the hex pickup.

In fact, the band pass filter's gain levels are so high, that the output op-amp clips/distorts way too easily. Maybe just replacing it with a pot(and perhaps with a buffer(which has a unity gain)) could be enough, as the output seems to be already 'hot' enough, to need any amplification.
 
But the idea of using the 100k pots between bpf's and optocouplers seems to work quite well, when inspected and listened closer.

It is true, what I've been reading about analog vocoders: it takes a lot of effort to get a grasp, how to operate the bugger, to suit one's taste(or get whatever usable effects).
Not an object for me, as I tend to go to extremes, when testing/abusing such diy-built devices.
----------------------------------------

Partly off-topic philosophising:

When I got my first stereo system, some time went by, and I read that an equaliser would improve the listening experience. So I bought one(that I still have).

At first, when turning the knobs up an down, I wondered: "where is the beef?". Again, after some time, I started to notice slight differences in sound quality, when playing with the knobs.

Eventually that eq became an essential part of the stereo system. The key is to develop the hearing,

Same seems to apply to this device also: Hearing/ear training. Instead of turning the knobs to eleven, there are subtle things, of which the secrets are made of.

In the end, I have ideas about extending this vocoders range of usability further, by adding something like 4046-phase locked loops for each channel. We'll see about that.

[sorveltaja]

I can now see, why there is a fuzz option for the instrument input, in the original schematic. It adds, when used very sparingly, a bit more harmonic content. So I breadboarded the 'tube sound fuzz':

  [ You are not allowed to view attachments ]  

It uses a 4049 cmos hex Inverting Buffer -ic:

  [ You are not allowed to view attachments ]  

I left the R6(10M) out of the circuit, as only very mild portion of fuzz is needed. Actually finding a sweet spot could be easier to dial in, if a multiturn pot was used insted for R5(1M).
They seem to cost at least five times more, than ordinary pots, so I rather look for other options. One could be to use gears to increase the ratio of rotation.

I have some plastic gears from printers etc. and also guitar tuning machines, that use brass/steel worm gears. Printing the gears with small teeth, like mod 0.5, and expecting them to mesh smoothly, isn't too good idea, as I found out on one of the past projects. In future it could be done, once the entry level printers develop enough.

Certain level of precision is preferable, so perhaps the guitar tuning machines(having something like 8-12 ratios) would be more than plenty. They also take less room, than similar ratios done with spur gears. I have a feeling, that it will be the next 'sub-project'.

In the end, an audio sample, where the fuzz was used for instrument signal:

  [ You are not allowed to view attachments ]