Beginner's luck and madness?

Status
Not open for further replies.

apoikos

Member
Why using a pcb when you have cardboard? Well, actually there are many, many reasons..
This is my first project ever in anything involving electronics, soldering and programming arduino style. I tend to think I was just lucky with the components I bought.

Anyway, it's a 72 signals usbmidi controller working (almost) just fine. Only need to tame some bouncing when some pots are in middle position but I found out the issue is more the quality of soldering than anything else.

I am including a picture of the back which most of you will find mad I suppose :)

Also, I now entirely realise the price of things and I am even more amazed on how can behringer (for example) keep such low prices. Unbelievable.

Also, no caps at all. Analog averaging at 10 (might change as I haven't really played around yet), hysteresis at <> 4 at 10bits resolution, output obviously at 7 bits. Of course I have only just completed it, with some testing I will figure out more. But the fact that it's right now working is incredible at least to me.

IMG_20160927_144144.jpgIMG_20160927_144207.jpg
 
Good for you! There's nothing wrong with trying something inexpensive and learning for yourself what the tradeoffs are.
 
A few comments:

NuCPU_front_x.jpgNuCPU_back_x.jpgTeensy3_6_x.jpg

My first project was built in the 60's when the only tech available cheap was vacuum tubes. Cardboard wasn't up to the task of holding the heavier components, but thin (about 3/16") wall paneling worked for small stuff and plywood worked for larger stuff. Fast forward 50+ years and the same construction techniques still work, but I use "perf board." It is a thin board made of phenolic or FR4 PC board material with no copper, just an array of holes with 0.10 inch spacing, same as DIP IC chips and teensy modules. Buying it through the usual electronics distributors is kinda expensive, but the Chinese web sites and sometimes Amazon have it for reasonable prices.

I have recently retired from a 41 year career designing cell phones and two way radios. I can tell you that those pots that cost us over $1 each cost Behringer less than 10 cents each. They don't have to pay Mouser or Digikey's mark up and warehousing costs and they buy a big bunch at once.

I have built a very similar device, for a slightly different use. Mine doesn't do MIDI, yet. It feeds a 1V/oct synthesizer, software coded in the Teensy. I have 49 pots and 4 rotary encoders wired to a Teensy socket which can accept a T3.2 or a T3.6. The pots are multiplexed, which I assume that you are doing with the little daughter boards amongst the pots. You may want to add a cap (about 0.1uF) from B+ to ground near each multiplexer if there isn't one already on each daughter board. I used the 3 volt regulator on the Teensy module to feed the hot lead of each pot. There is a cap at each end of each row just for general noise suppression. Don't know if they are really needed, but I always put caps on the ends of long DC runs.

For my purpose where pots set the pitch of oscillators, I needed 13 bits of resolution with no averaging or hysteresis. The synth is stable and will hold pitch within 1 or 2 Hz for several hours. There can be a visible (on the scope) pitch change when the house furnace kicks on, but it is generally not audible. I may go to an outboard reference later to see if it fixes this.

My first (much smaller) test board used 100K pots. This caused a slight, but audible interaction between adjacent pots. The large breadboard uses 10K and does not have this issue. The Freescale data sheet for the processor chip warns about using the A/D converter with a high impedance source, so I reduced the pot values. Absolute VCO pitch is set by the rotary encoders which are completely digital and the step size can be small.

My wiring "furball" is similar to yours, except I use thin wire that's nearly the same color as the board for most wires. I use yellow for digital, blue for analog, black for ground. and red for DC voltages. It just makes things a bit easier to follow when you break a wire......hey, it works so why do something different. I pit a wood frame around it so I can move it easily. I threw it in a cardboard box and took it with me on a 2700 mile road trip. I think I wrote some code in each of the 7 states I drove through. The box came through it all with no issues. Not the construction similarities in the pictures. I put the Teensy on a daughter board for easy swapping. At this time I have only one T3.6. The other board is the PJRC audio board. It allows 44/16 high quality output, and frees up the D/A for LFO's.
 
My first project was built in the 60's when the only tech available cheap was vacuum tubes. Cardboard wasn't up to the task of holding the heavier components, but thin (about 3/16") wall paneling worked for small stuff and plywood worked for larger stuff. Fast forward 50+ years and the same construction techniques still work, but I use "perf board.
And we get the term breadboard from people putting nails in a wooden board (i.e. a board often times used to slice bread) and stringing wires between the nails: https://en.wikipedia.org/wiki/Breadboard.
 
Except, yours is much much better, much more polished than mine!

It's great to hear that a "pro" isn't laughing at me.
I'd love to use those boards but they are nonetheless quite expensive, and you find them in standard sizes. Plus, for the difference in price having a custom pub would make this project be completed in less than an hour!
But yes, knowing myself a little I will definitively go through the process of learning Eagle, and trying to make one, although I am 100% sure I will make some mistake in the gerber files to send, I have 0 experience and no idea how the manufacturing really works.
Also I have no idea if the noise and capacitance on a pcb is actually lower or higher than using wires, and I don't know what sizes the tracks should have for different purposes (3.3v vs digital signal vs ground). Too bad I am not an engineer.

You really get 13bits resolution with no hysteresis and averaging and with multiplexes??? How much is the voltage actually jittery in percentage? Well, you must be good, but considering it's an audio application I guess you can cope with continuous data. In my case I can't really keep sending tons of micro changes through the usb port as software would go nuts, hence the need to stabilise.

I did have a color scheme that made sense but was getting short of some colours so I mixed them up.

I paid 0.25€ for the small rotary pots, which was the cheapest I could get. I wonder how much less those could cost at volume to big companies such as Behringer, you really think something like 1/10th?
 
And we get the term breadboard from people putting nails in a wooden board

I started out in the early 60's with brass upholstery tacks and a pine board with a couple of tube sockets screwed to it. Then I learned to cut the terminal strips out of a discarded TV set and screw them onto the board.

It's great to hear that a "pro" isn't laughing at me.

Everybody starts somewhere. The fact that you made something that works on the first try is an incredible accomplishment. I have seen senior students in engineering school do far worse. "Pro" is a relative term. I have designed, and laid out the PC boards for complex things like cell phones (it was my day job before the layoff), but I am mostly a hardware guy. In the software world, I'm still a rookie.

I will definitively go through the process of learning Eagle

I use Eagle for most of my personal projects. For a project where you are only going to make one unit, especially something large like this, a PC board may not make sense. It would cost too much, that's why I used the blue perf board. There are actually two pieces glued together, because that's the biggest ones I had. It is possible to make your own PC board directly from Eagle without even making Gerber files. Google the "toner transfer method" of making a board. I use that a lot, especially if I'm not sure its going to work.

You really get 13bits resolution with no hysteresis and averaging and with multiplexes???

I have not actually made any measurements on exactly how much resolution I'm actually getting. I needed to be able to sweep a VCO without actually hearing the discrete frequency steps, since I am simulating a 70's vintage analog music synthesizer. I just kept cranking up the resolution until I got a smooth sweep. It needed " analogReadResolution(13); " to sound like an analog VCO. Keep in mind that the chip in all the Teensy 3.X's have two only A/D converters. The A/D pins are already multiplexed inside the chip, so we are actually multiplexing twice. I do not externally multiplex the pins that are directly controlling the emulated VCO's since they respond to the musician's controller playing. I multiplex the inputs from the pots since they are only changing synthesis parameters, and a small delay or stutter is not a big deal. I got all that working with a Teensy 3.2 which was pretty much maxed out, so I did not use averaging or hysteresis. I have switched to the Teensy 3.6, so I may try them in the future, but not until I get all my other code and hardware working.

How much is the voltage actually jittery in percentage?

I send some of the data out the serial port and watch it in a terminal window for debugging purposes. Any A/D converter will exhibit random changes in the LSB, so I pushed that bit out beyond the limits of audibility by going to 13 bits of resolution, I can see the LSB, and sometimes the adjacent bit bouncing around in the serial data, but I can't hear a perceptible pitch change, so I don't care.

but considering it's an audio application I guess you can cope with continuous data.

I don't plan on ever sending any knob data out the USB port, so I don't have that constraint to worry about. My box is actually on the other end of the wire. Eventually I will need to plug a controller, MIDI keyboard, or computer into my synth, and play it over MIDI. Got to get the sounds all working before I even think about USB or MIDI.

I did have a color scheme that made sense

Mine went out the window when I put that multi colored ribbon wire in. I originally built the box for a Teensy 3.2, but ran short of processing power and I/O, so I ripped out the original processor board and made the new one that's in there now. Since it fit's either a T3.2 or a T3.6 (and maybe a T3.5) I needed more pots and connections, hence the 60 conductor ribbon cable. I still have room for a few more pots too!

cost at volume to big companies such as Behringer, you really think something like 1/10th?

I used to work at Motorola designing cell phones and two way radios. When you are making a million phones a month, yes. Many common parts cost us about 10% of the single piece price from a distributor, and sometimes less. We had purchasing people whose full time job is negotiating cost, delivery and scheduling a particular "commodity" years in advance. I don't know how big Behringer is, or exactly how they buy parts, but I'm sure they get pots far cheaper than we do. The pots I used cost me about USD $1.20 each when I bought 50 of them from Mouser. I could believe that a large company gets them for $0.12 each.
 
The main reason why I would like to try out pcbs would be designing a cool case (making it somewhat special and I have an idea) and trying making a kickstarter to fund all the bureaucracy to have the product certified for selling. I have in mind a 300 boards goal. of course I know it's totally premature etc, but I guess the sooner or later if you want reliability you have to get to pcbs?
Or do you think it is possible, both production wise as well as certification wise as well as reliability wise, to use perfboards? Would you buy any product made with a perfboard?

By the way your project sound really, really cool.
And there are no limits with what you can do with the tools we have a disposal today!
 
bureaucracy to have the product certified for selling.

I don't know where you are in the world (my guess would be Greece) but the way to get around this BS in the USA is to NOT sell a finished product, and not have whatever it is be powered directly with line / mains power. Here a box of parts, a kit, or even a "prototype" need not go through all the certifications needed for a "commercial product". A MIDI controller, or in my case a DIY synthesizer kit can be powered entirely with USB bus power or a cell phone charger that has a USB connector on it. Let the end user supply that part. It should already be certified for connections to mains power. Of course you can still be held liable if someone decides to take your product into the bathtub with them, and gets fried, but having no power cord and only 5 volts at limited current removes most of the risk.

I have in mind a 300 boards goal

Would you want to hand wire 300 perf boards? I know I don't have the patience for that. I make a PC board even for my own projects if I am going to make more than 2 or 3 units, or if really delicate wiring is involved. Making a PC board ensures that every unit is the same. The reliability, and troubleshooting abilities are also far better. The only issue I had while dragging my synth box along on a 2800 mile car trip was a broken wire. That would not have happened if the unit was PC board based. If you are making a box to be used by musicians it has to be roadie proof...and idiot proof.......and beer resistant!

I'm not sure I would buy a perf board product, but I may be a bit biased.....part of my job at Motorola was laying out PC boards, some with tiny SMD parts and 10 or more layers, so it comes easy to me. In fact the only "products" that I currently sell are PC boards and parts kits for vacuum tube amplifiers. No certifications are needed to sell these items as long as they are ROHS compliant, and that technically shouldn't matter for a small volume device.

My synth is currently a learning platform, but I believe there will be one or more "kit" products developed from it. There will be a front panel that goes over the blue perf board once I quit adding more pots, switches, or encoders. Then it will be time to decide on a final form factor (one complete mini synth, of several Eurorack modules) and make PC boards to replace the blue stuff.

And there are no limits with what you can do with the tools we have a disposal today!

And when you do reach the limits, someone will make better tools. I have a killer synth that secretly has a PC motherboard inside it and runs several commercial software synths from Arturia and others. I can also plug my guitar into it and run Overloud's TH3.

I discovered the Teensy earlier this year quite by accident. A friend of mine has an awesome modular synth and he was demonstrating a new module. I liked it enough to order a kit, and found that the "main brain" inside it is a Teensy 3.2. That started me down this long road.
 
Status
Not open for further replies.
Back
Top