I'm researching for a project that would need to have some sort of touch detection through a 5 mm thick glass table.

Would it be feasible to use a PCB with appropriate patterns and a mTouch/QTouch/capSense microcontroller with it on one side of the glass and expect it to be able to detect touch on the other side of the glass if there is gap between the PCB and the glass? I'm thinking of having SMD LEDs on the side of the PCB close to glass. Ideas to fill the gap are welcome too.

I did do some research on this and I've seen that some people here have managed to have gapless sensors work through glass, but I'm not sure how this would work with the gap in it.

  • 1
    \$\begingroup\$ What length/width dimensions are required for this glass sensor sheet? Would it be very small (< 4 inch diagonal), small (< 20 inch diagonal), or bigger? Also, how many sense points in this area? \$\endgroup\$ – Anindo Ghosh Dec 3 '12 at 14:55
  • \$\begingroup\$ @Anindo Ghosh At this point, those parameters aren't fixed. We're still researching. Diagonal of 19 inch is the lower boundary of what would be acceptable. Around 30 inches would be OK. As for number of sense points, that too isn't fixed, so the answer is as many as possible. I guess that maybe 1 per 2 cm^2 would be good while 1 per cm^2 would be great. We could make bigger sense points if needed. \$\endgroup\$ – AndrejaKo Dec 3 '12 at 17:54
  • 1
    \$\begingroup\$ For something that big, I can only put forward a totally speculative idea, so I won't make it into an answer: An LCD monitor's panel. Take the backlight out, and you have a big glass plate which is transparent, yet is full of addressable conductors (the liquid crystal cells). By modifying the scanning that takes place for normal display purposes, these conductors can be "activated" in groups, the capacitance between the active (high frequency voltages on them) and inactive groups providing a vast capacitive sensor array. It might actually work... or not. :-( \$\endgroup\$ – Anindo Ghosh Dec 3 '12 at 18:07
  • \$\begingroup\$ Should I get a monitor with unbroken panel, I'll try something like that. It sounds like a good idea. \$\endgroup\$ – AndrejaKo Dec 3 '12 at 18:13
  • \$\begingroup\$ You could actually try it out at small scale and low cost, by taking apart one of the cheap digital photo-frames. Next step: Pick up a non-functioning LCD TV from the local recycler - If the panel is physically unbroken, odds are that the strategy can be tested on it. \$\endgroup\$ – Anindo Ghosh Dec 3 '12 at 18:16

The image below selected gap size of 0.5 mm works well for directing the fringing fields through 10mm of glass overlay. Here button size = glass thickness add 5mm for better response. You will need mechanical support for the glass and certainly need plexiglass or equivalent to couple tracks on button to glass. so overall thickness of 10mm is practical, possibly a bit more.enter image description here

With suitable plexiglass spacers between board and glass to fill the air gap over the buttons, I think it is very feasible. but without that would be very risky.

Gapless with glass and Plexiglass gap fill over each finger track and ground track over 10mm is possible, but with air gap perhaps only ~1mm is practical with ground guard around RF pad due to low capacitive coupling of air to glass.

8mm glass is equivalent to 1mm air gap and you need 5mm air gap plus glass, so Consider silicon pads to fill enxtend the geometry, support the glass and bridge the gap so the finger pad can close the RF capacitor plastic pads below the surface. As long as the gap capacitance is greater than the finger gap of <1pF. BTW 1pF is around 10KΩ at 24MHz

SMD gold wire bond internal joints may get stressed from the glass. Can you use smoked or Mask painted glass with silicon to fill gap over the sensors under the glass? that may work raising the dielectric constant over the pads and give the bonus of structural rigidity and diffused the LED indicator over the touchpad spot or for use as feedback indicators.

Side thought for that matter maybe you can use IR emitter refector pairs for touch pad points for touchless glass just by pointing close to emitter.detector pair set to short proximity threshold.

You can also use plexiglass spacers extending the pad shape to the glass, which both have similar epsilon values of 8 and use a thin adhesive on component side and similar spacers around LEDs to protect and the thin glass from strain.

THis is from a Cypress spreadsheet

Input Parameters Value Units

Overlay Thickness 2 mm Overlay - Dielectric constant 2.8 farad/m Capacitance of trace per inch 2 pF

Minimum Recommended Button Diameter
(based on minimum 0.25pF Finger response)
Noise Conditions - Low (0.05 pF Noise) 7 mm Noise Conditions - Medium (0.075 pF Noise) 9 mm Noise Conditions - High (0.1 pF Noise) 11 mm

Maximum Trace length
Noise Conditions - Low (0.05 pF Noise) 400 mm Noise Conditions - Medium (0.075 pF Noise) 387 mm Noise Conditions - High (0.1 pF Noise) 374 mm

Button to Ground clearance 2 mm


Capacitive touch sensors measure changes in capacitance. The capacitance for a parallel-plate capacitor is approximately C = A/d.

If a cap-sense device works fine with a 1 pf change in capacitance when the top surface of the glass is 1 mm away from the metal pads, it can't tell the difference -- and it will work just as well -- if you can rig things so we get about the same 1 pf change in capacitance when the top surface of the glass is 10 mm away.

One way to do that is with simple scaling. If you want a glass + airgap distance (d) is 64 times as large as on the capsense demo boards, use approximately 64 times the area (A) -- make the metal pads 8 times as wide and 8 times as tall, and instead of using 1 fingertip, use your entire hand.

Because each 1 mm of additional airgap is about as bad as 5 mm of additional glass thickness (the relative permittivity of glass is roughly 5 times as much as air), most people try to press the cap-sense plates directly against the glass without any airgap. One way to do that is to use metal springs to electrically connect metal on the backside of the glass pads on the PCB. The metal of the spring itself might be adequate. Some people attach metal plates to the springs and let the springs press the plates against the glass. Other people glue conductive transparent plates to the glass. (See section 3.5.3 of "AN2869: Guidelines for designing touch sensing applications", section 4.3.2 of "AN0040: Hardware Design for Capacitive Touch", dangerousprototypes: "App note: Using Philipp springs to create touch keys", section 3.3.1 of "QTAN0079: Buttons, Sliders, and Wheels: Sensor Design Guide", etc.)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.