0

I am looking for ideas to implement fail-safe function in joystick control box. At present, the 5V joystick HF11S10 is connected to STM32 ADC inputs via simple 5.1k:9.1k divider. We are revising the schematics to conform to all requirements for powered medical devices, so we would like to be able to detect the disconnected or damaged wiring. With current schematics disconnected joystick results in valid 0V signal. In terms of application logic it is identical to "full reverse speed" input, which is extremely dangerous.

Note that we already solved this problem in the future projects by ordering joysticks with limited output range 0.25V-4.75V, which makes 0V signal invalid. Another possibility is to use joystick variants with digital interface available from manufacturer. However we do have a stock of old devices, and considering their price we'd very much like to find a simple solution that allows us to use them safely.

Important characteristics from datasheet:

  • Ratiometric output voltage 0V-5V
  • Output current 8 mA
  • Output load min 1 k
  • Output impedance 2 Ohm
  • Supply voltage range 5VDC ±0.01VDC

Also, just in case, there are couple digital inputs available and by moving things around I might be able to free one analog input.

Maple
  • 11,755
  • 2
  • 20
  • 56
  • Are these potentiometer type joysticks, or hall-effect? – Ron Beyer May 22 '20 at 18:35
  • they are hall-effect – Maple May 22 '20 at 18:37
  • Not that it helps with your "old stock", the way I do this in industrial systems is a multi-safety approach. First we use 4-20mA output ones so we can detect shorts/opens, then we use a "deadman switch" and third we utilize the microswitches in the joystick to determine when they are off-center. All 3 together must be in a certain state for it to move the equipment. – Ron Beyer May 22 '20 at 18:42
  • @RonBeyer You are correct, it does not help much. Even without old joysticks we have to accommodate people with limited dexterity and some specialized input devices have output options limited to 0-5V signal only. Would be nice to cover those too, if possible. We do have emergency shutdown switch, but we cannot use typical enabling switches mounted on industrial joysticks because a lot of our clients are unable to activate them. – Maple May 22 '20 at 19:03

1 Answers1

0

Have a circuit that sits between the joystick and the controller.

Now you can handle this a few ways. 1. Have a window comparitor on the joystick input. If it falls outside the expected voltage range. Force 0V to the controller on all motion inputs. While it may be slightly annoying for the user. You can cut the last 5% of travel for this safety if you want. When the elderly tend to have a problem they push to the extremes of motion on a controller.

  1. Measure the supply current through the joystick. If something goes open. It will be outside the range of that current and again halt all motion controls.

Or if you have the option. Just define in your software that 0V = stop. And anything above it is valid (with some head room for your adc noise. Just remeber to also have vcc as stop. As if say just the ground wire failed on the joystick it would also go full forward speed.

Reroute
  • 4,377
  • 6
  • 14
  • Sorry, I did not get how this works. The joystick output is 0V-5V with 2.5V in the center. So, anything coming out of it is valid. When the wire disconnected or joystick unplugged the incoming voltage will be floating (before the divider) or 0V (after divider) which falls into valid range. Or do you mean physically restrict stick movement? Measuring current supply is interesting idea but it does not help if only feedback line fails. – Maple May 22 '20 at 23:21
  • To rework it with that new info, say you had a diode drop between joystick ground and circuit ground, if the voltage is below about 0.5V, there is an issue, if the joystick can run off a reduced supply, you could do the same on the top end, to know if there is a fault with either supply lead, if the sense lead is broken you divider should pull it to ground, so a way to catch all 3 cases. – Reroute May 22 '20 at 23:34
  • Moving ground reference is a great idea! And simple too, exactly as I wanted. Too bad I can't use it. Just checked datasheet - the supply voltage is 5V±0.01V. Very weird requirement for a device with ratiometric output, if you ask me. – Maple May 23 '20 at 15:08