laminating a custom, curved PCB?

Forum 2005-2010 (read only) Hardware Development
1

I've been working on an electronic bagpipe project for a while, and am collaborating with a couple other guys to expand the project to incorporate Irish, Scottish, English, Swedish and Spanish bagpipes. Once we get the basic framework re-written, rebuilt and working well, the project should support almost any type of reeded (and some non-reed) wind instrument.
Anyway, the first version, Highland Bagpipe only, with a (mostly complete) project doc is here:

Here's my current issue:
Brass bars as captouch sensors work but have almost 0 sensitivity,
so ......

And the general solution:
A proper capacitive touch sensor design, with ground planes. A standard PCB touch sensor design should work, the touch sensors need to be curved around the inside of the instrument shell.

My thinking right now is:

  • PVC pipe for the shell

Then
EITHER

  • epoxy foil sensor pads to the inside of the pipe
  • laminate epoxy&glass over the sensor pads (a DIY version of G10 or FR)
  • epoxy a foil ground plain on top of the laminate
  • ** As an alternative to foil, perhaps conductive paint and conductive glue would work? Paint the sensors and traces, then glue the wires to the traces?

OR

  • laminate epoxy&glass inside the pipe to make the curved PCB substrate
  • glue foil onto the substrate
  • mask and etch as per a flat PCB

Either of the above should be almost the same (functionally) as glueing a PCB sensor array to a flat overlay ..... correct??.

Any ideas, comments or suggestions?

Tim.

BTW: For anyone who's interested in more info on capacitive touch sensing, TI has a decent app note (SLAA363A), ATmel has tons of design guidelines and App notes on their website, and the QTouch libs are availbale for anyone who can figure out how to use them with the Arduino platform (I haven't put any time into that bit of integration yet, but my initial research indicated it may be possible).

2

i use a capacitance multiplier

for each touch area: (middle lower board)

slightly touched bunch of thin copper wires under thin duct tape:
up to 100 loop cycles more than untouched...

-arne

3

Conductive paint or pcb conductive pens sound like a good option.
You could mask up the surface for good crisp looking traces.

I understand you can even solder to some of the pcb conductive pen paints.

Gordon

4

@Gordon
Thanks for the masking idea. If I can figure out how to get the sensor field projecting through the overlay, masking would be a good way to transfer the sensor & ground plane design from flat boards to the curved pipe :slight_smile:

@Arne
Does your circuit detect proximity from behind a thicker overlay?
I'm currently using code based on this Playground page:
http://www.arduino.cc/playground/Code/CapacitiveSensor
This morning I tried a PCB sensor pad surrounded by a solid ground plane. It required physical touch to work, same as my brass rod sensors ... damn!!

Tim.

5

i dont remember the resistor values, but i was able to detect a finger in 1cm distance (through air), which was too sensitive for my one-hand-keyboard...
but it reacts on glass and it works through duct tape, 2... so it is not necessary to have a low resistance connection (like on old tv sets)...

regarding the playground circuit:
maybe a bigger pullup resistor (like 10MOhm - the arduino has just 20kOhm to 50kOhm...) could help?

i will do some tests tomorrow...
hopefully i find the right values for 1cm again... :slight_smile:

-arne

6

Thanks Arne. I was also thinking a bigger pull-up resistor, but won't be able to experiment until next week.

Tim

7

with this code:

void setup() {
  Serial.begin(115200);
  pinMode(2,OUTPUT);
  digitalWrite(2,HIGH);
  digitalWrite(8,LOW);
}

void loop() {
  pinMode(8,OUTPUT);
  noInterrupts();
  pinMode(8,INPUT);
  uint16_t dt;
  for (dt=1; dt; dt++) if (PINB & (1<<0)) break;
  interrupts();
  Serial.println(dt);
  delay(1000);
}

and the circuit from my first post in this thread (modified: R1=10MOhm, R2=100kOhm, R3=0Ohm) i can detect my finger through 1mm transparet plastic (250 untouched, 260 when my finger is very very close (mechanical cause unlikely))... :slight_smile:

without the transistor and without R2 it doesnt respond so good... but it still does...

so the huge R1 is important here...

PS: i just saw that the pulse pin can be connected permanently to +5V...

PS: i used a 1 sq cm (0.15 sq in) aluminium foil as capacitor plate...

-arne

8

Arne,

I got this circuit working better than my old circuit:

+5V --------|====|------------------||    finger pad
            1-10Mo        |              
                          |
                        uC pin

I'll try higher val resistors, which should make the pad more sensitive.

Also, the above circuit seems less sensitive with a ground plane :shrug:

Thanks for the code sample.

--Tim.

9

cool... :sunglasses:

is there no space for a capacitance multiplier?
u can use it via a MUX (i used a CD4067), but it adds some capacitance itself...

or didnt it improve the signal?

what was the maximum sense distance?

btw: in my design every key behaves differently: some r quick and others r slow...

-arne

10

I haven't tried the capacitance multiplier yet, but will do so.

Timing is VERY critical!! When playing a real bagpipe, note durations can be as short as 5ms, so the entire sensor array needs to be scanned in < 1ms or the latency is too high.

Going forward, the project will diverge into 2 versions - a really simple Highland Bagpipe only version with absolute minimum part counts, that can be built for < $50 US by anyone who can solder .....
then there'll be a mutli 'Duino electronic woodwind framework that will support more sensors, more robust sensors and related circuits, much better audio playback of real instrument samples, will modularize the hardware (chanter, drones, bag, uCbrain) and possibly add wireless.

The simple version has already been built and independently made by a few others - the best one was in done in Brazil (IIRC, Sao Paulo) using PVC pipe and galvanized screws for finger sensors!! All the pieces of the 'framework' version are either scattered accross my desk, or have previously been proven by the 'Duino community, it's just a matter of integrating a lot of bits together into a cohesive project, then improving it over time.

Anyway, this is the sort of commercial product out there today, at 1,500 to 2,000 Euros.

http://redpipes.eu/index.html

I'm fairly confident a few of us can put together a credible, Open Source, alternative for the cost of 3 Arduinos and a handfull of components. So called 'fit-and-finish' would be left up to the individual builder, but we'll probably have a couple of docs on making suitable enclosures, etc..

--Tim.