Strain gauge Joystick - Could something like this work?

[PeterFW]

I would like a 2 axis input device, i have a few wonky design restrictions i set myself, one is no "moving" parts and so i  came up with the attachment as a rough idea. 

The middle hole is just for attaching the stick of the joy-stick part.
The deflection gets measured with the PCB strain gauges, the cutout makes it more conforming.

I have seen designs like this, so it should work, but what do you think?

[amyk]

The whole device becomes the (slightly) moving part!

More seriously,

https://commons.wikimedia.org/wiki/File:Pointing_stick_of_a_Lenovo_ThinkPad_keyboard-4488.jpg

https://commons.wikimedia.org/wiki/File:Pointing_stick_of_a_Lenovo_ThinkPad_keyboard-1047.jpg

[george.b]

Aha, I see you've been studying the IBM Trackpoint

That looks like it could work, yes. There has to be a small deflection for it to work, so maybe it's something to keep in mind when it comes to materials/mechanical layout - how much deflection as a result of how much force, and how sensitive a strain gauge you'd have to use.

[PeterFW]

The whole device becomes the (slightly) moving part!
More seriously,
https://commons.wikimedia.org/wiki/File:Pointing_stick_of_a_Lenovo_ThinkPad_keyboard-4488.jpg
https://commons.wikimedia.org/wiki/File:Pointing_stick_of_a_Lenovo_ThinkPad_keyboard-1047.jpg

Thanks for the reply! 
You are correct, that is why i put the moving in quotation marks  

I have read the patent and a few more resources concerning this design, i was just hoping someone maybe has tried it yet and got some practical knowledge. 
Sadly the Trackpoint devices are, as far as i can make out, based on regular FSR. 
I mean, technically a stain gauge is a FSR too but you know what i mean

[PeterFW]

Thank you for the reply!

That looks like it could work, yes. There has to be a small deflection for it to work, so maybe it's something to keep in mind when it comes to materials/mechanical layout - how much deflection as a result of how much force, and how sensitive a strain gauge you'd have to use.

Yea, while i know how strain gauges work and what you have to do to get data from them, i never worked with them. 
On FR4 they should experience a deflection which these usually do not see. 

Maybe i should just design a couple of different arrangements and try them out but i got a problem with resistance. 
Using Copper on FR4 i do not get anywhere near >1Ohm.

[fcb]

Could work.  You might want to put gauges on both sides of the PCB. Zero calibration will be a challenge which everway you do it if you go for bridge arrangements.

[Prehistoricman]

Yea, while i know how strain gauges work and what you have to do to get data from them, i never worked with them. 
On FR4 they should experience a deflection which these usually do not see. 

Maybe i should just design a couple of different arrangements and try them out but i got a problem with resistance. 
Using Copper on FR4 i do not get anywhere near >1Ohm.

Your deflection will be in milliohms anyway so the low total resistance doesn't matter too much.

[PeterFW]

Your deflection will be in milliohms anyway so the low total resistance doesn't matter too much.

 

But in a wheatstone bridge arrangement, i need to match the resistance of all elements to the resistance of the strain gauge. 
If my gauge is below 1 Ohm, i practically create a short between the supply rails. 

Or do i have this part wrong? 

[fcb]

Yeah. It’s going to be super low impedance. Commercial strain gauges are typically 120/350/1Kohms, very thin metal printed/etched onto polyamide.

Principle is the same though - you’ll just have to be creative with your circuitry to cope with the low Z.

[skylar]

Yeah. It’s going to be super low impedance. Commercial strain gauges are typically 120/350/1Kohms, very thin metal printed/etched onto polyamide.

Principle is the same though - you’ll just have to be creative with your circuitry to cope with the low Z.

Not to be annoying, but would not it be easier to glue commercial strain gauges to the PCB cutout originally shown? Then you could use the standard circuit to measure it, no moving parts though there is more assembly than the project seems to aim for.

[richard.cs]

I have tried printed strain gauges on PCBs before, they weren't great. In my case I was comparing tracks on one side with others at 90 degrees on the other side of the PCB and my main problem was that the two identical geometries ended up 30% different in resistance giving large offsets. I suspect this was mostly plating rather than etching. On mine the major killer was the resistance delta meant that the self-heating didn't cancel so it wasn't as simple as removing a fixed offset.

Your sketch suggests you're comparing two areas on the same side of the board which should be bent in opposing directions, they will probably be better matched than mine. You may still find the low impedance problematic though.

[fcb]

Yeah. It’s going to be super low impedance. Commercial strain gauges are typically 120/350/1Kohms, very thin metal printed/etched onto polyamide.

Principle is the same though - you’ll just have to be creative with your circuitry to cope with the low Z.

Not to be annoying, but would not it be easier to glue commercial strain gauges to the PCB cutout originally shown? Then you could use the standard circuit to measure it, no moving parts though there is more assembly than the project seems to aim for.

Sure - but then you have a $20 thing, I think the OP wants to make a low cost thing.  Also, a standard strain gauge setup might still need an instrumentation amp with a gain of 100-1000, ADC etc..

Handling the low-Z is the first headache, compensating for thermal drift will be a big deal also.  But then thats 90% of the fun with electronics.

[Leo Bodnar]

Typical strain gauge has 1-2microstrains limit before deformation becomes plastic and irreversible.
Have you done stress-strain analysis on it?
Self-heating is usually prevented by gating the excitation voltage.
Leo

[PeterFW]

Thank you everybody for your ideas and feedback, i want it to be a cheap solution because i would like at least... 44 or 88 of them. 

But after a lot more thinking an research i ended up with a FSR matrix from Sensitronics, at least i ordered one. 
I think i can get the desired effect if i glue on suitable rubber feet with considerations of the matrix pitch. 

That should get me where i want.

[jfitter]

Electrically the design is simple whether you use strain gauges, optical rotation sensors, etc. What really needs to be considered is whether the mechanical design is practical.

Real joysticks used for machine control have a mechanical arrangement that provides for a centering force. What this means is that a certain critical force (torque) must be exerted before there is any output from the sensor, after which the output is linear with deflection. Check out radio control joysticks, especially the high end ones, aircraft side-sticks, joysticks for cranes and hoists, coal mining machines, etc.

The problem with simple force sensors on fixed sticks is that there is an output for any force. A tiny force yields a tiny output and the practical reality of this is that it is almost impossible to control anything with it.

This is why non-gymbal joysticks are NOT used for machine control.

 

[Leo Bodnar]

I would not say anything. A lot of critical control systems use force sensitive control as opposed to displacement sensitive controls.
E.g. most modern fighter planes.
Humans control forces much better than position.
Leo

... it is almost impossible to control anything with it.
This is why non-gymbal joysticks are NOT used for machine control.

[MarkF]

I would not say anything. A lot of critical control systems use force sensitive control as opposed to displacement sensitive controls.
E.g. most modern fighter planes.
Humans control forces much better than position.
Leo

... it is almost impossible to control anything with it.
This is why non-gymbal joysticks are NOT used for machine control.

Correct.
The force sensitive stick in the fighter moves less than +/- 1/4" at the top of a 12" long stick.
It works surprisingly well.  Very easy to control.

[jfitter]

I would not say anything. A lot of critical control systems use force sensitive control as opposed to displacement sensitive controls.
E.g. most modern fighter planes.
Humans control forces much better than position.
Leo

... it is almost impossible to control anything with it.
This is why non-gymbal joysticks are NOT used for machine control.

Correct.
The force sensitive stick in the fighter moves less than +/- 1/4" at the top of a 12" long stick.
It works surprisingly well.  Very easy to control.

This is by your definition a displacement sensitive stick.
In a fighter the stick must be controlled when under a G loading of up to 9G. Elastomer sticks are the only practical solution.

The MOST precise joystick operators are high level radio control aerobatic competitors. They require a level of precision which is almost super human. The top radios have precision machined gymbals floating on ceramic bearings with 16 bit hall effect angle sensors on each axis. Force sticks have been tried and found unsuitable.

As a past national champ I believe I am well qualified to speak on this matter. You need to actually get out there and do it to understand the practical issues involved - it's called engineering.

[Leo Bodnar]

The MOST precise joystick operators are high level radio control aerobatic competitors...
As a past national champ I believe I am well qualified...

I am chuffed to learn from the best - thank you for this lesson. 
Leo

[Kilrah]

The problem with simple force sensors on fixed sticks is that there is an output for any force.

That's why you use the difference as your input, proper thresholds etc. I did play with FSR-based sticks in the past and had good results for the intended purpose.

The top radios have precision machined gymbals floating on ceramic bearings with 16 bit hall effect angle sensors on each axis

Nothing in the OP says he requires such precision though. And having a foot in that field, a lot of what you just touted is just clever marketing

[jfitter]

The problem with simple force sensors on fixed sticks is that there is an output for any force.

That's why you use the difference as your input, proper thresholds etc. I did play with FSR-based sticks in the past and had good results for the intended purpose.

The top radios have precision machined gymbals floating on ceramic bearings with 16 bit hall effect angle sensors on each axis

Nothing in the OP says he requires such precision though. And having a foot in that field, a lot of what you just touted is just clever marketing

Garbage - do you actually have one of these. I do. Open it up and have a real look. As an embedded design engineer for 30 years I do know what I am looking at.

The problem with these forums is the answers from people who clearly have no idea what they are talking about. I reply to almost nothing, and only when I see a question that I have experience in and know a great deal about do I answer and offer advice. It is usually met with responses from the peanut gallery, which is why I don't reply very often.

You lot are on your own. Figure it out yourselves.

[MarkF]

The problem with simple force sensors on fixed sticks is that there is an output for any force.

That's why you use the difference as your input, proper thresholds etc. I did play with FSR-based sticks in the past and had good results for the intended purpose.

The top radios have precision machined gymbals floating on ceramic bearings with 16 bit hall effect angle sensors on each axis

Nothing in the OP says he requires such precision though. And having a foot in that field, a lot of what you just touted is just clever marketing

Garbage - do you actually have one of these. I do. Open it up and have a real look. As an embedded design engineer for 30 years I do know what I am looking at.

The problem with these forums is the answers from people who clearly have no idea what they are talking about. I reply to almost nothing, and only when I see a question that I have experience in and know a great deal about do I answer and offer advice. It is usually met with responses from the peanut gallery, which is why I don't reply very often.

You lot are on your own. Figure it out yourselves.

Are you all talking about RC Model airplanes or actual aircraft?
I was referring to the actual aircraft and we just used the un-certified ground version.
Still far too expensive to take apart ($16,000).  I don't know the internal mechanism.  Just that they are force sensitive.

[i_am_fubar]

I question the need for 16 bit hall effect. Assuming 50mm of tip travel, that means your controlling down to 800nm of positional accuracy. That is..... Not going to happen. 50um (10 bit), can't believe you would notice a difference?

[Kilrah]

Garbage - do you actually have one of these. I do. Open it up and have a real look. As an embedded design engineer for 30 years I do know what I am looking at.

I never said it's not what was inside, only that it's a marketing gimmick rather than something actually significantly beneficial. I've seen a few fun blind tests.