Diy optical pickup for guitar -- is it possible?

[sorveltaja]

After some testing with current electrical setup on the breadboard, it could be worth of building. The circuits are very simple, but when there are six of them, using perfboard, it is way too easy to make mistakes.

So I'm going to make the actual pcb(s). Bit of an extra work, but the resulting board should be a breeze to populate.

The only thing, that has changed since I made pcb's years ago, is the laser printer. Hopefully its ink has similar properties, as the previous printer.
As always, there is only one way to go where no man has gone before find out.

In the past, I used cheapo transfer method, where the pcb pattern was printed to gloss or semi-gloss magazine/ad paper, then transferred to the copper surface of the pcb.
Results weren't 100% perfect, but in most cases, that was more than enough.

[sorveltaja]

Making the pcb went just fine, but only after etching, I noticed that it was a mirrored version of the circuit

But that aside, the first pcb consists only the amplifier stage for the phototransistor, in hopes of keeping the overall construction 'modular':
  [ You are not allowed to view attachments ]     [ You are not allowed to view attachments ]  

After testing, that the above pcb works as expected, the first candidate for the next 'module' is one, that consist of 4093 ic's, with schmitt trigger configuration.

What I mean by 'modular', is that the amplifier stage isn't permanently tied to any of the following whatever-stages, so if one doesn't work, take it out and replace it with something else.

[sorveltaja]

The first pcb is ready to go for testing. For me, the drilling is always the most thrilling part.
-----------------------------------------------
Babbling:

Thin carbide drills are usually very fragile, and are to be rotated at high speed. Not so in this case, as I used about 2800 rpm(max rpm of my mill) for the 1mm one.

As a precaution, the drill's vertical movement was minimised, so that it just penetrated the pcb.

One might ask: why not use hss drills instead, as they are dirt cheap, and more flexible?
Anyone, that have used hss drills for pcb's, knows, that after 50 holes they are as dull as it gets, and tear instead of cutting.

Although carbide drills cost a bit more, and are not to be used for hand-operated drilling(ask me how I know), they don't wear out easily.
Some kind of (lightweight will do) drill press is required, though.

Reason for this run out? If there are newcomers reading this thread, willing to make their own pcb's, hopefully they get more info to get over the learning curve involved, than I had in the past..
-----------------------------------------------------------------
But enough of that. Drilled pcb as it is:
  [ You are not allowed to view attachments ]  

[sorveltaja]

There was an error in the previous pcb, but it's now fixed:
  [ You are not allowed to view attachments ]  

It's just a bare bone circuit, which could be cropped even more, by replacing upper right 10u capacitor with a jumper, as I'm not sure if it's needed at all.

Although it's working as expected, it seems to need some kind of psu filtering thingamabob to suppress overall noise. I use a simple one like this, which could also be used as a split power supply.
It was used only to power the amplifier circuit, and not the power hungry ir-leds:
  [ You are not allowed to view attachments ]  

So far the testings and adjustments were done by using scope, function generator, and multimeter. But they don't reveal all. The end result is still judged by aural testing.

Next stage after the amplifier could be a schmitt trigger, zero cross detector, or about anything, that is able to clip the output to suit cmos ic's.

[sorveltaja]

Simple zero cross detector(zcd) from datasheet..
  [ You are not allowed to view attachments ]  

..and 4093-based schmitt trigger were compared as signal 'squarers'.

Zcd did work, but its sensitivity(or lack of it) made it to cut the signal from the plucked string too early. Another version, that has 10mV sensitivity, although having more parts involved, could still be an option to test. With that high sensitivity, noise could be a problem, though. 

Zcd with an adjustable/variable input sensitivity would be jolly good to have in this case. But when searching on the net information about it, such things doesn't seem to exist.

Then there was the 4093-based schmitt trigger, which did let the signal from the plucked string to 'squarify' bit longer, making it to look more 'natural' response.

Current circuit has (maybe)enough amplification for the receiving phototransistor. On a second thinnest string(0,33mm/0,013"), when plucked, the output goes to 9 volt(with 9V supply), and fades as the string vibration decreases.

So the circuit is pretty much same, as it was previously:
  [ You are not allowed to view attachments ]  

[sorveltaja]

After looking at the datasheet of 4093 ic, by lowering its supply voltage, the input threshold lowers also.

At 9V it is ~ 1,8V, and at 5V ~ 0,9V. By adjusting the 50k pot, there is a very narrow spot, where the output appears. Lower supply voltage makes that spot even narrower.
Multiturn trimmer pot could be a better choice for that, as they aren't that costly nowadays.

So far, circuits with minimal parts count are used. As I mentioned earlier, this thread is useless without pictures, but even more so without audio clips.

The mechanical setup isn't up for that yet, and needs a lot of testing as well. By using the current (flimsy)setup, it should be possible to use a least two strings, to make some nice sounds audio output.

The form of output, that I expect, could be fuzz/buzz. If all goes well, there should be some form of cross-modulation, that one gets only by using hex-fuzz.



 

[PekkaNF]

.....By adjusting the 50k pot, there is a very narrow spot, where the output appears. Lower supply voltage makes that spot even narrower.
Multiturn trimmer pot could be a better choice for that, as they aren't that costly nowadays.
.....

Normally trimmers are used only on fine tuning and discrete resistors are used to bring the trimmer "on range". This is because trimmers are not that stable and adjustment is easier if the adjustment is not ruined if a fly sneezes across the room.

This is probably all familiar to you?
https://www.electronics-tutorials.ws/resistor/potentiometer.html



After you find the sweet spot with that trimmer you can find min/max range that is likely to be usable and then calculate to measure the legs resistance and use that to find needed discrete resistors and new trim pot size.

Second thing is that schmitt triggers pretty generally require "reasonable" signal - i.e. they don't have amplification. That means that if your signal level is small you need to boost it before schmitt trigger. Amplifier also gives you possibility to bias signal if it has a DC-component that would drive that amplified signal into saturation. You have used a capacitor, but sometimes that is not without problems and you might need that signal DC-component later to actively control led current to stay in linear region.

 

[sorveltaja]

Pekka, thanks for the link. Yes, using fixed value resistors is in the plan, although on the thinnest strings, which have less sustain than thicker ones, finding closest resistor values could be a bit tricky.

On a second thinnest string, when tested with 100k pot, at the sweet spot, it was removed from the circuit, and its values were measured, being close to 60k/40k, or 40k/60k.

The supply voltage was lowered to 5V, but before testing all the six circuits together, I can't tell, if it's the the final value.

Other than that, I drew some sketches, considering the mechanical setup. The electromagnetic humbucker pickup was remover from the guitar, to make more room.

One thing, that has bothered me about current mech setup, is that you have to drill holes through the guitar to make connections to the pcb. 
 
A small step in hopes to change that:
  [ You are not allowed to view attachments ]  
In that pic, sender ir-leds' common ground legs are brought together with pcb trace. The 2-pin grey headers, where the leds are to be connected, should be identical size in length.
But as I have long ones(40 pin), they should be sawed off of that, and milled to size using depth jig.

There is ~4mm space below the headers, so small 1.6mm pcb should fit there. Not sure though, if there is enough room for all the traces for optical components, when using single-sided pcb.
But as always, I have to test it to find out, how far I can get.

I guess that there is an option to order custom made pcb's, but from what I've read, in this country, they would cost ~200€/piece. Or maybe I'm missing something.

But that aside, to provide more constant signal from the strings, while testing one by one the outputs, could be to use a motorised piano-hammer-like contraption instead of tickling the strings with fingers and toes for the purpose. That may well be the next side-project.

[sorveltaja]

First test with a pcb, that may end up to the pickup housing:
  [ You are not allowed to view attachments ]  

Thinnest traces are 0,5mm wide, and it took quite a few times to get them to transfer without cracks:
  [ You are not allowed to view attachments ]  

Of course, the continuity of every trace has to be tested, before cropping the pcb to its final size.

[sorveltaja]

Using parts of the header(mentioned earlier) to connect the optocomponents seems to be way too complex for this project.

I guess it could be done, but cutting 2-pin headers to the exact length would take huge amount of effort.
Reason for using parts of headers was to keep the led/phototransistor pair somehow 'aligned', when adjusting their vertical/height position.

On to the next version, which rely on using two hex-head screws as vertical 'rails'. The first printed version of that was too loose for the led/phototransistor combo to stay in place.

There is a discussion on the 3d-printer thread about snap-fitting printed parts together. That's where I got an idea for this:

  [ You are not allowed to view attachments ]  

At first, the red screws couldn't hold the green part in place, when they were adjusted.

Then came the snap-in thingy;
  [ You are not allowed to view attachments ]  

Screws were pushed into the green sender/receiver housing. It took some force, but once in place, the overall construction is surprisingly stiff. And for me, that was only the first test using snap-in fit.

As the green part(height) needs only a minor amount of movement to find the 'set-and-forget' -position, wearing the printed parts out shouldn't be problem. But time will tell.





 

[sorveltaja]

A new version of the pickup housing. Couldn't find any 'neat' way, that could be easily done, so why not go for quick and dirty. Well, maybe not so quick, but at least it's dirty.

The idea is loosely same as before, but this time, instead of using the drilled holes on the guitar to put the wires through, wires go through the hole on the side of the pickup housing.

The skeleton version of it:
  [ You are not allowed to view attachments ]  

And with the parts:
  [ You are not allowed to view attachments ]  

I printed that out, and although bony, it has some rigidity in it. If that isn't rigid enough in practise, I just have to add some wall thickness.

That kind of 'open' construction means, that the optocomponents should stay tightly in place through all the wild wiring hassle.
With right tolerances, they could be pushed in place seemingly tight, but that isn't enough. The trick is to apply small drop of lacquer or paint to the neck of the led/phototransistor.

When it's then pushed in the place, and let dry overnight, it should be more than enough to keep the bugger in the place.

Quicker way, using super glue might seem tempting, but the fumes of it will ruin the hemispherical 'seeing' area of the component, as they are usually made of some sort of plastic.



   

[sorveltaja]

First tests with a new mechanical setup:

  [ You are not allowed to view attachments ]  

Wires of two pickups in place. But that housing might get too crowded, when all the wires are stuffed there, especially when using a 3mm thick shielded cable for the receivers.
But that's what I have at the moment.

One shortcut could be to mod the housing, so that the wires would come out of the side of it, instead of only at the end, as it is now.

Now, that the two pickups are in place, I tested shortly the 'E' -labeled one, which is for the thickest string.

The screw-adjustment for the height seems to work, as expected.

10Khz square wave was then fed to the sender/ir-led, and to my surprise, there was... well, I don't exactly know what it is, but perhaps a reaction, when adjusting the pickup up or down.

When it was in one position, there were bumps between the square waves:

  [ You are not allowed to view attachments ]  

After some height adjustment, the bumps were gradually gone. So the difference, of what the receiver/phototransistor 'sees', whether it's string shadow, or whatever involved factor, can be visualised using scope. During that, the string stood still, and wasn't plucked, or otherwise excited.

Obviously that effect is greatly reduced, when testing with thinnest strings, but should be visible, though.



 

[sorveltaja]

Now, that the amplifier pcb is almost finished, I got curious of how the optical pickup sounds like. I tested it only on the thickest string. I'm going to order four missing ic's, once the holidays are over:
  [ You are not allowed to view attachments ]  

Output from the 386 amp is certainly very strong, as it's used in x200 gain configuration. The output is distorted in rather ugly way, but that doesn't matter, when the signal is to be conditioned for the cmos devices.

So far, there are no gain adjustment pots. Maybe they could be fitted afterwards, don't know yet. If not, then it's just time for a new pcb.

But to test how the string actually sounds, I tuned the overall output down by lowering the sender/ir-led voltage to the point, where there wasn't distortion.

Yes, it is possible to get a 'clean' output of this setup. Although I wasn't too much interested about that, it might well change.

The sound of the E (thickest string) is very bassy, not in a muddy way, but like a real bass. Didn't even know, that guitar can produce such low frequencies, without octave-down -effects.

Not only that, but when I increased the voltage of sender/ir-led, the overall output was again very high, I noticed, that plucking the other strings(whose pickup wasn't connected),
the E string ringed, responding to some of their plucked note harmonics.

[sorveltaja]

This one might be boring one, but as it is essential for the project(or at least with current mechanical setup), I'll give it a go anyways.

It's about height adjustment of the pickups. Today I tested, how to get the pickup(sender ir-led/phototransistor pair) to the position, where the string is at the center of the ir-beam.
That could of course be done by listening the output, and guessing the sweet spot by ear.

But guessing games aside. To see, if I can find anything useful for the matter, I fed all kinds of signals to the ir-led, and looked the results on the scope, while moving thin, string-like wire across the ir-beam.

The pickup used for this test is separate one, but similar to the ones, that are attached to the guitar.

Method, which I mentioned earlier, that feeding square wave to the ir-led, and judging the result by the looks of receiver output, when moving an object across the ir-beam, doesn't seem to be reliable.

There are simply too many things to be adjusted, before one can see the results(if any).

While frustrated about turning all the knobs of scope and function generator, there was one thing, that I hadn't noticed before.

When I slowly moved thin(and thicker ones also) string across the ir-beam, the current readout of the psu went temporarily from 23 mA to a few mA lower, as the string was moved up and down.
  [ You are not allowed to view attachments ]  

After repeating that procedure few times, there was constant reaction of the current readings of the psu.

The circuit, that was used to amplify the phototransistor output, is same as in previous post.

The thing seems to be to feed that circuit through the ir-led, with a signal, that makes the receiver/phototransistor amplifier 'busy', while 'receiving'.

Even a small amount of deviation in the ir-beam seems to give the amp time to take a breath, so the current consumption decreases shortly.

In the end, I think that the deviations of current consumed by receivers amp could be more reliable way to measure the best position of the string, to get maximum output.



 

[sorveltaja]

While the current mechanical setup is in the guitar, and the amplifier pcb is fastened to the guitar also, I decided to have a look of another kind of mechanical/optical setup, which I tried very shortly in the beginning.

It was left aside, as I didn't have an amplifier circuit to provide enough boost back then. But now that I have, I thought that re-visiting that design could be interesting.

Don't mind the bigger setup, as the one, that I'm babbling about, is the small one on the upper side of the picture. It's glued in place:
  [ You are not allowed to view attachments ]  

The sender(ir-led), and phototransistor have 50 degrees angle between them. No any particular reason for that angle, other than to see, does the concept work:
  [ You are not allowed to view attachments ]  
  [ You are not allowed to view attachments ]  

That test housing has a "backrest" to keep the sender/receiver combo from rotating around, when adjusting the height. Seems to work quite well.

As can be seen on the first picture, that thing was tested on the thinnest string(~0,2mm). If that works, any string thicker than that should give even more usable output.

To my surprise, there was stronger output, than was expected, after adjusting the pickup closer to the string.

So far, I have only used the scope to see the results, but it's the resulting sound that counts after all. I haven't got that far yet, as projects like this tend to expand(explode) in form of countless possibilities involved.

But yes, I think the mentioned sort of setup could very well be more appealing to the most of experimenters, that are willing to get their feet wet.

Quick summary of the two lately tested concepts so far, but take it with pint of salt:

One with sender/receiver pair in opposite(horisontally) sides of the string:

+ immune to string bending(which is also a horisontal movement)
+ the "sweet spot"(max. output) could perhaps be found by measuring the current consumption of the amplifier circuit 
+ overall output level

- not a very practical to use, as the sender/receiver pairs (and their housings) protrude above the strings, so it's not possible to dampen the strings in natural palm-assisted way.
 
Then the one, that relies on reflecting the ir-beam from the string surface from the below:

+ remains under the strings
+ allows free playing, and string dampening at will

- rather sensitive to string bending(maybe not so, if multiple pairs of sender/receivers were used)
- can't think of reliable way to judge the "sweet spot" yet

But as always, all that needs a lot of testing.

To expand the subject even more, I'm thinking of ordering some hall-sensors, to see, if they could be abused used for this purpose. But that's beyond of this thread, though.

[sorveltaja]

Continuation for the previous post; I tested that configuration, where the sender/receiver pair was perpendicular to the string.

Results were quite disappointing, as there was a lot of noise, although the ir-led was fed with dc. Also the pickup should have been very(too) close to the string, to get usable output.
So far, that configuration is done and dusted.

Variation of that is kind of similar, but this time the sender/receiver pair was set to be parallel with string:
  [ You are not allowed to view attachments ]  

Output was bit better, with less noise. Also in that configuration, if the pickup was too close to the string, output decreased. I guess, that the reflection of the ir-beam from the string could be better also.

But the form of the output signal seems strange, as the waves are formed from tiny spikes:
  [ You are not allowed to view attachments ]  
  [ You are not allowed to view attachments ]  

Maybe the schmitt trigger could translate those spikes to more solid, rectangular shape. Haven't made a pcb for that yet, though.

So far, I have found one major drawback in that concept; the output signal changes, when the ir-beam is reflected from players hand, near the strings.

One commercial manufacturer uses the latter concept, but there the optical components have a lot more distance, apertures, and fancy ir-filters between them.
I'm not going to copy that thing, as it was never advertised to have properties, that I'm after. Perhaps slightly adopt the ideas, if needed.

But the general principals are there for testing purposes, after all.


 

[sorveltaja]

One basic testing version of the 6 x schmitt trigger board:

  [ You are not allowed to view attachments ]  

For some reason, I insisted on making as small as possible board. Then I had to cram the capacitors in:

  [ You are not allowed to view attachments ]  

It lacks only two caps, which I'm going to order, besides other components. But when testing it with function generator and scope, it worked as expected so far.

All those clumsy jumper wires, I wish I could replace them by using two-sided pcb. That's a subject, which I haven't explored at all, as aligning the two pcb transfers with each other seems overwhelming, to get usable pcb. Maybe that's one tough side project for the future... 

Another version could be a bit simpler, with similar layout, using 40106 hex inverter gate, which has the needed six schmitt trigger inputs/outputs in one ic:
  [ You are not allowed to view attachments ]  

I'm not sure if that 40106 works as 4093's, that I have used, but maybe I'll order some, to find it out.
 
 

[PekkaNF]

.....
All those clumsy jumper wires, I wish I could replace them by using two-sided pcb. That's a subject, which I haven't explored at all, as aligning the two pcb transfers with each other seems overwhelming, to get usable pcb. Maybe that's one tough side project for the future... 
......

Not sure how you make the boards, but could you use the same method with transerfer sheets than is used with UV-method? That is to tape the transfer sheets facing each others into an "envelope" I.E. index them firs to each others and then slide the PCB blank in and then tape it into correct place.

Or you could use something to allingn the PCB into a "pocket" like this:
https://justaddelectrons.com/blogi/double-sided-pcbs-aligning-the-sides-for-exposure/

Or just drill holes on opposie corners of the PCB that has a mating parts on masks and then tape the mask (or in your case transfer) which has the same indexing holes.  Hell....you might even 3D print an alligment fixture for mass production:)

[sorveltaja]

Pekka, thanks for the info. I was thinking of that envelope method, but the paper, where the transfers are printed, should be plain, instead of the magazine paper that has either text or images in it.

Glossy photopaper for laser printer could be a better candidate for that. Or some pcb-transfer specific product.

By using the uv-method, as the link provided tells, making two-sided pcb's could be a lot easier. Haven't got that far yet, when making pcb's. So indeed it would make a lengthy sideproject.

To get back to the project, yet another variation of the recent setup appeared. Previous one used 50 degrees angle between sender/receiver, to reflect the ir-beam from the string.
Output was there, but it was somewhat rather lacking.

Then I got a plan 9 from outer phase, for increasing the angle, and distance between optical components. That way, the sender(ir-led) could shine its light to a longer amount of string, meaning more string surface to reflect, and hopefully more output.

Now the angle between sender/receiver pair is 90 degrees:
  [ You are not allowed to view attachments ]  

I printed a new housing, which should be a bit more 'universal', if the distance/angle between optical components needed to be increased even further:
  [ You are not allowed to view attachments ]  

Result is more responsive, and more 'solid' looking output, than the one with 50 degree setup. The sender is again fed with dc:
  [ You are not allowed to view attachments ]  

Although I'm looking for hard-clipping, maximum output, there is also a great potential for experimenters, who are possibly looking for a clean output from optical pickup.
Just tune things down enough to get to that territory. In that case, using photodiode instead of phototransistor could be a better one, as it has lower output.

After all, these things always require testing, to find the options, that suit.

[JohnHaine]

Interesting thread, haven't had a chance to read it all, but do you not have one problem with the configurations you are using?  Essentially the string when it is stationary obstructs the beam, so there is no output from the photodetector.  When the string moves, the output increases whether the string moves one way of the other.  So it will double the frequency of the vibration, and be veru distorted.  When plucked, the string moves essentially side to side doesn't it?  So to pick that up you need an LED shining down on the string and two phototransistors below it, equally illuminated when the string is centralised, while being differentially illuminated when it is moving.  Actually then you could use just a single LED and a "light guide" to distribute its light to each string, or one of those multi-LED strips, possibly with a diffuser above it.

Also in some of the early posts you seem to have a strange hybrid of constant current and constant voltage drive to each LED - really you want constant current so that each LED gets enough forward volts to work.  You could use a significantly higher voltage and a bigger resistance, or better use an inductor as a "dropper" using your chopper, with a buck diode to allow the current to flywheel when you switch the voltage off - this would be much more efficient, and the light output would be constant rather than "sampling" the string.

I hope these comments are useful, and apologies if I am covering old ground.

[sorveltaja]

John, thanks for the reply. Yes, there has been plenty of problems with the previous setups. Current mechanical/electrical setups aren't hassle-free either, and are more like testbeds for different concepts.

I guess, that when the string stands still in the ir-beam, only its shadow is detected, but as it has no motion, the receiver has nothing to generate. Perhaps a bit like a motion detector.

Movement of the string, at the very time, when it's plucked, could be seen as a very slow movement, as the plectrum/finger 'drags' it, and releases.
After that, when looking the string at the end, it moves like a jump rope(in exaggerated way).

I have thought of using a 'common' or single light source(consisting multiple ir-leds) for all the strings, but there is one thing, that I have noticed. The string thickness seems to have an effect, when adjusting the sender/receiver pair for the maximum output.

Also, if the receiver gets too much light from the sender, it goes 'blind'(maybe saturated), and doesn't respond anymore. That's what I have observed, while testing different mechanical setups so far.
Meaning, that if one setting works for one string, it needs to be altered for the other to fit.

That considers only the setups, that I've tested so far, which use single receiver/sender -pairs.

I wish I could explain all that better, but many things appear, when using components for the purposes, which they aren't designed for.

And, no need for apologies. Constructive input is always welcome.

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

What comes to the project, again another mechanical setup, where the sender/receiver pair distance is ~15mm, and the angle between them is 120 degrees:
  [ You are not allowed to view attachments ]  

I already tested it on the thinnest string, and the output was there, not necessarily better than on the previous, 90-degree configuration, but I just had to choose one of the setups, to move on with the project.

[sorveltaja]

Three pickups in place for the first audio test. Another three 386 amp ic's are still missing:

  [ You are not allowed to view attachments ]  

Again, that kind of mechanical setup's pickup outputs get easily disturbed, when moving hand over them.
They probably have their uses, but the strings should be plucked near the neck to avoid that.

One commercial maker uses 'strings-through-the-bridge' configuration for their guitars and basses, which doesn't have that problem.

Earlier I tested somewhat similar concept(where the sender/receiver "stare each other in the eyes"), although it had a bit different approach. I might get back to that, after all.

Results of the current setup: what comes to the 386-amplifier circuit, I wanted all the gain and boost it could ever offer. Now, that I have it, when listening to the output, it's overly raw, and very tricky to control in any way, as I left the gain pot out of it. It produces all kinds of farty sounds on its own, and isn't too responsive anymore.

In this case that amp clips very hard at x200 gain, which could be helped to some extent by increasing its supply voltage to get more 'headroom', but I'm not looking forward for that option.

So, as the mechanical setup will change, also the amplifier circuit has to be modified.

For the audio test, I didn't even bother to use schmitt triggers, as the output is beyond erratic already.

But all that aside, there is a hint of the effect, that I'm after, in the recorded masterpiece clip, in form of harmonies, which was made using the current setup:
(zipped mp3 attached)

   

 

[sorveltaja]

I tested if the gain pots could be added to the current amp circuit, to make it more usable. It wasn't that hard, only one trace had to be cut and three holes drilled per amp. 0,8 mm solid wire was used for that, to make it stand straight, when adjusting:

  [ You are not allowed to view attachments ]  

At first, I tried it with only one amp on the left side, and it makes a huge difference, and gives possibility to tame it down. But even after that, the output isn't exactly 'crystal clean', as it has some 'grit' in it.

The overall sound, when tested on the thickest string, is quite dark(as it was before, when testing with thinner strings).
Meaning, that the lower frequencies are certainly there, but the 'definition' is missing.

Not sure, if it's just a nature of the optical components, or the amp circuit itself. 

But here we go, instead of trying to get the maximum output, I'm getting interested of how the thing actually sounds.


   

[AdeV]

Pekka, thanks for the info. I was thinking of that envelope method, but the paper, where the transfers are printed, should be plain, instead of the magazine paper that has either text or images in it.

Glossy photopaper for laser printer could be a better candidate for that. Or some pcb-transfer specific product.

By using the uv-method, as the link provided tells, making two-sided pcb's could be a lot easier. Haven't got that far yet, when making pcb's. So indeed it would make a lengthy sideproject.

I did quite a bit of messing about trying to make my own PCBs last year.... I can summarise my results here (I'll do a proper write up someday):

1) UV exposure method is a right pain in the backside in a home shop. Getting the timings *just so*, and the exposure chemical strength *just so*, and then etching it just right - there are a LOT of variables. I reckon I must have used about 5 sheets of 100x160mm board, just to get one working 10x25mm board. And even that wasn't perfect.

2) Laser printer method is WAY easier. Forget the magazine paper, the laser printer will crinkle it, and it's hard to see the PCB print over the magazine print, especially if you're trying to go double-sided. Instead, use Vinyl (the sign-maker's type, that you cut in a machine). Works beautifully. I successfully transferred prints using a hot clothes iron, and latterly a slightly modded cheap laminator.

The only downside to laser transfer is pinholes - the laser doesn't cover 100% perfectly, so you can get pinholes in the traces. Not a big issue unless you've got very narrow traces. The upside is you just dunk the board straight into the acid after it's come out of the laminator. Have some acetone handy to wash away the toner once you're done.

Drilling PCBs is a whole different bag of fish. I bought one of those X-Y engravers, which works fine IF you get the boad perfectly lined up. Still working on that one... (well, will be, once I've got some spare time again).

Anyway - enough on that. I know it goes against the spirit of MadModder, but I think if I ever want a PCB that I *know* is going to work properly, I'll probably outsource it next time. PCBWay seems to get good reviews, and is cheap enough.

[PekkaNF]

UV-method does needs some "dial" in. At the school we had repro camereras/films/pen plotters and all the works. Worked out fine. Secret was to use only fresh factory coated PCB:s and use two layers of film/paper that had darkest lines you could get and greatest UV pass trough you could get. Also the "envelope" had to be absolutely flat and UV-tubes have serious power.

I tried direct transfer method and I had best success with high coated high gloss paper, but applying just Goldilocks pressure/time/temperature turned out pretty difficult.

I have had some hit and miss success with UV-method, but my last try ended up to desperation, mainly because I had trouble with transparent material that would allow UV to pass. I bought the material that was confirmed to pass near 97% of that wave length....but I got the "improved batch" that was filtering about 99% of UV. Found out that when I made the "gradient" to find out, took like three minutes instead of usual 15 sec.

I have been thinking of routing method, but there is much CAM, indexing and all sorts of intricacies in that too.....

And then there is chemical copper and vias....

Probably best to buy board, but there is a definate apeal on "roll your own" .