0
\$\begingroup\$

I have salvaged a track-point device from an old laptop that ceased working. I couldn't find any information on it in the hardware manual. However, the labels (see below) suggested that I could apply a voltage between VCC and GND, and read out signals from X and Y, so I decided to solder a pin header to it and give it a try:

enter image description here

enter image description here

I tried to apply 5V on the VCC, and found that the X and Y pins would settle at about 2.5V each. However, pushing the track-point in all directions did not have any measurable effect on the voltage on X and Y pins.

Does anybody have any information regarding such device, or ideas of what I could try out to get a response from it?

\$\endgroup\$
  • 1
    \$\begingroup\$ Slide the point, in stead of pushing it. There's wiper pots, that shift in the opposite direction of pressure in the full housing, due to the pivot point created by the laptop's plastic at the base of the head. \$\endgroup\$ – Asmyldof Jun 2 '16 at 21:24
  • 2
    \$\begingroup\$ the output may have been changing but it may have been a really small value, if it was a strain guage (as some have suggested), the changes would only be a few mV \$\endgroup\$ – Sam Jun 3 '16 at 5:04
4
\$\begingroup\$

The Wikipedia page has more information. The TrackPoint needs to be treated as a pair (X-Y) of strain gauges. That Wikipedia page also includes a link to a relevant patents with more info.


Updated - Here is some more info I found from my bookmarks:

The A-to-D conversion (encoding) from the analog TrackPoint signals into either a standard serial or (later) PS/2 mouse interface, was originally done by a separate MCU. Later this functionality was built into the keyboard controller.

The Sprintek SK7100 is one of the devices produced by various vendors, which interfaces a TrackPoint sensor to a serial or PS/2 mouse connection. That datasheet includes this reference schematic - the TrackPoint sensor is connected to J1 (upper right). The PS/2 mouse interface signals are on J2 (lower right):

Sprintek SK7100 reference schematic

I remember that some earlier interface devices were less integrated and used external op-amps between the TrackPoint sensor and the MCU. If you really want to make your own TrackPoint interface, rather than use one of the existing devices, then PCB layout is even more critical as the smallest amount of noise interferes with accurate force-detection. This problem is sometimes seen when people use non-original PSUs for their ThinkPad, and then find that the cursor starts to move on its own, due to the increased EMI (from the lower-quality PSU) affecting the analog TrackPoint signals.

As an aside, this interview on the Microsoft Research site with one of the original TrackPoint inventors, gives some interesting background regarding its development.

\$\endgroup\$
  • \$\begingroup\$ This looks very promising. I will check if I saved the chip connected to the gauges (and whether it's a SK7100), and otherwise will try to increase the resolution of my measurements. \$\endgroup\$ – Dmitry Grigoryev Jun 6 '16 at 9:11
  • \$\begingroup\$ I finally got a signal from it using a precise enough multimeter. The resistance changes by about 0.1% when I apply maximum reasonable force. I guess it's not so useful in hobby projects without that SK7100 chip. \$\endgroup\$ – Dmitry Grigoryev Jun 9 '16 at 7:25
1
\$\begingroup\$

Patent US6509890 shows the following circuit. It looks like a wheatstone bridge strain gauge. I have absolutely no idea if your device is similar.

enter image description here

Try measuring the difference between X and Y. I'd expect it to be small.

\$\endgroup\$
1
\$\begingroup\$

You didn't mention whether you connected the GND pin?

This is a rather common thing to interface to a microcontroller. For example here is how one would connect to an Arduino:

https://www.arduino.cc/en/Tutorial/JoystickMouseControl

enter image description here

\$\endgroup\$
  • \$\begingroup\$ This is very similar to what I've tried, unsuccessfully. X and Y signals were barely changing over time, and with no certain correlation to my actions on the pointing stick. Do you think a 10-bit ADC could've been insufficient? \$\endgroup\$ – Dmitry Grigoryev Jun 6 '16 at 8:52

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.