Author Topic: A bit of advise for mTouch  (Read 2812 times)

0 Members and 1 Guest are viewing this topic.

Offline ivan747Topic starter

  • Super Contributor
  • ***
  • Posts: 2046
  • Country: us
A bit of advise for mTouch
« on: August 14, 2012, 08:28:22 pm »
If you are testing Microchip's mTouch library, and you are using a breadboard, do not twist the wires connecting to the sensor pads like if they were differential pairs or something, you will mess up all of the optimization functions!

This was more appropriate for the chat but that was removed. Any comments on mTouch guys?
 

Offline Burke

  • Newbie
  • Posts: 2
Re: A bit of advise for mTouch
« Reply #1 on: August 20, 2012, 08:02:16 pm »
The reason is because you want to always minimize the capacitance of the sensor.

The microcontroller is only able to measure the total capacitance it sees on the pin. If the base capacitance of the sensor is high, the addition of a finger's capacitance to the circuit will be small. Alternatively, if the base capacitance of the sensor is small, then small changes due to a finger can be more easily detected.

When sensors are not being scanned, they are typically held to the board's ground potential. So by twisting two sensor traces together, you're increasing the base capacitance of both sensors. (You've essentially wrapped a tightly coupled ground plane around both.) This results in a decrease in the amount of "percentage" change in capacitance caused by the finger.

It's worth noting: this coupling effect to ground will also apply to VDD and is important for all designs, not just breadboard prototypes. FR-4 is the typical PCB material used. It has a relative permitivitty of ~5 which is about the same as glass and twice as large as most plastics. So anything near the sensor on the PCB will be able to couple through the PCB's material in the same way that having the sensor traces twisted will increase the coupling.
 

Offline johnnyfp

  • Frequent Contributor
  • **
  • Posts: 261
  • Country: nz
Re: A bit of advise for mTouch
« Reply #2 on: August 21, 2012, 07:45:47 am »
I've seen many different designs including a pair of triangles. Some are filled, while others are honeycomed. Some are set of arrows. Is there a online tool or something that can help with designs of touch switches, and dimmer/pot type circuits?
 

Offline Burke

  • Newbie
  • Posts: 2
Re: A bit of advise for mTouch
« Reply #3 on: August 21, 2012, 11:10:43 pm »
The only tools I've heard of are specific to the PCB layout software.

Altium has a set of plugins for generating Atmel-specific sensors. Microchip will have a similar plugin released with the next Altium version. Capacitive sensors are just areas of metal. For normal buttons, after you have a reasonable amount of metal area (10x10mm or 15x15mm -ish), it's much more important to control the other components around the sensor than to be concerned about the design of the sensor itself. (See the rant on ground planes near sensors in my above post.)

Slider design is different than a typical 'button' design, however. Now the 'significant information' will come from the ratio between how well pressed one sensor is in relation to another. If you want there to be a linear shift in that ratio as you slide your finger from left to right, the sensor design itself must be linear. (Said another way: the transition from one sensor to the other should be as linear as possible.) This most easily lends itself to the "two triangles facing each other" design. (Think: the 'diver' flag.)

The multiple sets of arrows (aka chevrons) is one step up in complexity, but can be used to increase sensitivity by having smaller sensors, thus less base capacitance. You'll still want each arrow/chevron to be sized based on the sensing target. For fingers, this means ~12x12mm. You'll also want to make sure the depth of each arrow is significant enough to allow two sensors to always be pressed at once on the slider when someone is pressing. If the user can slide their finger from one location to another and hit areas where they're only over a single sensor, the signal will have a stair-stepping behavior. This is because there'll be a small period of time where moving the finger from left to right doesn't change the ratio between the amount the current sensor is pressed vs a neighboring sensor - so the signal's value will remain steady despite the finger changing location.

I know one of the tools that will be provided with the next Altium version is a chevron slider designer to make this easy... and a dual-triangle slider designer. I would suggest using the dual-triangle design for prototypes since they're easier to make in the PCB design programs and have more linear outputs.

(Three disclaimers:
I work for Microchip.

Non-linearities in the slider design can be removed through software look-up tables and other techniques, but it's not as nice as having it linear-by-hardware-design given the low-cost nature of most applications.

There could be many other tools available that I simply haven't worked with. Cool capacitive sensor shapes rarely matter that much... but there ARE some designs that may only be used with specific companies' methods. If using a tool like Altium, make sure to read the fine print on the plugin you've chosen to use.)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf