diy touch sensors for knobs

[jrp]

Hi

I´m thinking about doing my own touch sensors instead of buttons for my yet to build seq3.

I guess it should work here, but i´m not sure so i´m asking.

The type i´m thinking of is like this: The Din has a pullup R. The "button" is nothing but a smaal pice of copperboard with a very tiny cut between it´s upper and lower half. One half is connected to the input, the other is connected to ground. A finger touching this sensor will close the switch.

Anyone tried this with the din module?

The pullup resistors should be larger than the 10k on the board. Maybe 500k. Is that a problem to the module?

Will it start to get unstable?

do i need cabs to protect the switch against multiple trigger? i noticed that the layout for normal buttons doesn´t have any cabs.

If worked nicely this could be a cool, futuristic way of playing the sequencer, and it costs almost nothing.

[SLP]

A touch sensor function is already available:

http://www.ucapps.de/mbhp/mbhp_din_touchsensors.pdf

[jrp]

cool!

So it works with bodycapacity? Just a simple metal plate?

Do you have to change anything in the code for that?

[SLP]

So it works with bodycapacity? Just a simple metal plate?

Yes!

Do you have to change anything in the code for that?

Don't ask me anything about coding.

The 4 stars below my name do just mean that I've posted a lot of stuff in the forum.

They do not mean that I know how MB works  ;D

[stryd_one]

The 4 stars below my name do just mean that I've posted a lot of stuff in the forum.

LOL I think I've said before, that big postcount of mine is from asking questions, not answering them hehehe

I am trying though :)

Do you have to change anything in the code for that?

Check out the comments in the main.asm, they may help... I must admit that a very quick search didn't find any info that I could paste to help you out (it was all in relation to using the touchsensors for motorfaders), so if you do find out how exactly to do this, it would be very kind of you to document it for future midiboxers :)

Good luck!

[/tilted/]

You would need to be sure that you keep a good distance between sensors, and that the sensors themselves are well isolated from the front panel (or use a plastic front panel). It occurs to me that you could use small rubber grommets to isolate the plate from the panel, and just glue the wire from the back to make it all stay put. (don't use solvent glue, or you'll melt the grommets). The usual caveats apply to the use of touch sensors, ie the better ground sink you represent, the better they will work (I think). You may find they work better if you have wet or sweaty fingers (fine for live use, then), or if you have bare feet, and stand on a wet, metallic plate (probably during a thunderstorm*) (*not recommended)...

[Wilba]

I did a lot of experimenting with the touch sensor feature of MIOS, with mixed results. One single touch sensor worked fine, but when I tried to expand that into an array of touch sensors, I got lots of false triggering, and spent a long time trying to work out what was going wrong, and eventually put the whole idea on hold.

I'm not trying to discourage you, but warn you before you spend a lot on a front panels or PCBs to experiment fully... don't just think that because one sensor works that ten or twenty identical ones will work fine.

There's also the issue of touching two or more sensors at once, which you will want to do with MB-SEQ I think... it may not work that well. I think what happens is J14 outputs a high pulse just before the DIN registers are sampled, and so untouched sensors will stay high (through some effect of the 47k resistors) but a touched sensor will be low because your finger just absorbed the current (that's the capacitance bit). So if you're touching two sensors, then your finger might not absorb enough for both sensors to go low.

I wouldn't even begin to start experimenting again until I borrow another oscilloscope, or I'd go nuts trying to work out why it doesn't work  >:(

[jrp]

Yes, i can imagine this. First thing i would try is making the resistors a lot bigger than 47k. Don´t know..

As about the code, i´m sorry, but if not in this forum i will never find out what changes have to be done.

The only programming i ever did was with gw-basic about 15-20 years ago...,

[Wilba]

The rest of the application will work whether the switches are touch sensors (with the 47k resistor connected to J14) or normal switches (with 10k pull up resistors). Increasing the 47k will reduce the current getting to the shift register pin and it might not be enough to be sensed as a logic high when untouched.

The code to turn on the touch sensor function is trivial.

This is what usually appears in the MIOS_USER_Init routine:

	movlw	0x00			; disable touch sensor
	call	MIOS_SRIO_TS_SensitivitySet
[/code]
Change 0x00 to some bigger number. AFAIK this controls the duration of the pulse on J14, which controls how sensitive the touch sensor is... if it's a long pulse, I think you need to touch the sensor harder so your finger conducts more and stores the charge.
You might also want to call MIOS_SRIO_DebounceSet in a similar way.
Have fun experimenting, but don't get too disappointed if it doesn't work well for you... i.e. there's probably going to be a lot of bounce on those triggers so 

[Screaming_Rabbit]

... I remember, there was a guy doing some touch pads by etching some copper areas on a PCB and I think his project was working out.

After a short search I couldn't find his posting... you should try to find it.

Greets, Roger

[stryd_one]

Polyphonic Touchsensors ? That one? Unfortunately he didn't document it :(

[Screaming_Rabbit]

Polyphonic Touchsensors ? That one? Unfortunately he didn't document it :(

exactly!

[Wilba]

Well he said he had it working when pressing two at the same time... so maybe it will work fine... in my case I had a custom DINx4 PCB with all 32 inputs being used as a touch sensor... imagine a lot of thick copper wires being used to simulate four strings on a guitar, i.e. they were as long as a fret spacing, so four rows of eight sensors. I recall I was getting heaps of false triggering... now it could be that the load was too much, so that sensors other than the one I touched weren't being held high... I don't really know why it wasn't working (otherwise I would have fixed it!)

[henrygr]

Well, it just goes to show that actually reading the forum is of great help. I have been researching touch sensors for the last 4 months, totally unaware that it was already programmed in to the MidiBox. (Pause while I slap my forehead)

The QT series of chips from Quantum have been bandied about before, but since their last discussions, the technology has come on in leaps and bounds. Have a look here

http://www.qprox.com/products/qt1081.php

I don't see why you couldn't use this and interface it with a DIN. OK. You will need to look at SOIC ti DIP IC for soldering purposes. But in effect it will provide you with a mechanical over a programmed solution.

I will be looking into it more at some point, but I have other fish to fry.....

......now where is that mackerel?

MP

[audiocommander]

yes indeed: the qt chips are great, they even can be calibrated very easily, just apply a signal to a pin (high or low, don't remember) and touch!

should be no problem to use more than one.

I used the QT301 (capacitance to analog converter),

the datasheet also contains some useful tips about mounting;

here's a (probably most useless) video:

http://www.audiocommander.de/blog/?p=66

I think I also saw some ICs with multiple inputs in one case and (of course) some with digital output as well.

Best,

Michael

[stryd_one]

Ahh that 'secrets' PDF really opens up the possibilities, thanks AC!! Before, it was basically an IC and a big mystery about how to use it ;)

As for the SOIC, I wouldn't worry about that - you put all that junk on the CS panel PCB assembly as per that PDF. I don't know if you've seen the scanning matrix chips they have... Pretty tasteful stuff....

[DrBunsen]

scanning matrix chips they have...

;D