I want to replace the mechanical movement and sloppiness of a conventional joystick with a "non moving" or stationary system. I have tinkered around a little with FSR's but the results aren't quite what I am looking for.

The existing system uses 5v pots. Ideally I would like to build a circuit that I could plug in where the x and y pots used to be plugged in. I'm thinking it will need a trim pot to center the system and another trim pot to adjust the amount of force required for full response.
I think it would consist of two half bridge circuits. One for x axis and the other for y axis.

So right off the start I need help to figure out which gauges I should be using. As well as designing the circuit.

I know there will be many questions to answer. Starting with how much force will be applied to the stick.

Any thoughts on how to get started?

Thanks in advance.

SparksnSmoke

 

Am I even close with my expectations?

Half Bridge for each axis. One gauge on top and the other on the bottom working opposite of each other. The Forces applied would depend on how far from the center pivot point I put the sensor I guess. But I think it would be something like 5 to 20 lbs.

What sort of opamp should i be looking at?

Thanks

SparksnSmoke

 

Unless we see what your mechanical arrangement is, where you will put the strain gauges, etc., I think it will be impossible to offer much help.

 

Well to have a look at what I am talking about I can post a CAD drawing with some notes. Should do the trick.

Thanks for showing some interest Ron.

SparksnSmoke

 

It's been done already. Here's one of several vendors:
http://www.ultra-msi.com/miniature-joysticks.html

Yes, it'll cost some $$. But at any reasonable rate for your time, it'll actually save you money. And, um, you'll be sure that it works.

 

Well here is what I am thinking as far as hardware goes.

The disc is 5" dia. x .125" or even .1875" aluminium, which ever works out better. The center post is 2.25" tall and has a handle that adds another 5" to it's height. The 8 bolt holes are what holds it stationary in the base.

The strain gauges are on the x and y axis both top and bottom. Although they could just as easily be put right on the shaft I think.

Hope this helps.

SmokenSparks

 

So what will the output from your design be, for a given force? I hope you can prove it's not zero.

If you absolutely have to make your own, you might as well use just a square post with strain gauges on each side. It would save you a lot of time in the machine shop.

 

Hi John. I don't know what you mean by "prove it's not zero".

I want to upgrade the system I have. I will have to machine the center post to fit the handle that I have. The top of that post has a thread on it. There is a 5 wire ps/2 connection at that joint also. It will pass through and plug into the original board also.

So what I want to do is incorporate a new circuit using the +5, ground and reference voltage that used to go to a 100K pot. The old circuit is looking for a voltage from 0 to +5 volts. The new pcb would simply plug into the existing +5, ground and the two reference voltage connections.

I have been reading anything I can find on the instrument opamps. But it's a lot to take in. The half bridge with a instrument opamp seems to be what I need but I lack the experience to figure out exactly what it will take. This is why I got on the forum here to find some help.

Thanks
SparksnSmoke

 

I'm wondering what shape the disk that carries the strain gauges will assume under load. Probably a kind of S-wave as one side of the central boss goes up and the other down. If that's true, there'd be a point of inflexion (zero curvature) somewhere about halfway from edge to center, just where you might put the gauges, hence zero output. Also, I don't see how that disk can avoid being highly sensitive to stress up and down, if the user pushes along the axis of the column.

A pot can do something that the electronic system can't, namely go all the way to +5 and Gnd. But you could get very close with a rail-to-rail output amplifier, like the AD623. And yes, a half-bridge would be the standard way to build something like this.

 

The flex of the disc is a good point. Sounds like a good reason to put the strain gauges on the tube? I can see this being the biggest problem since the gauges are glued in place you can't exactly just move them around until you get it right.

I can test the pots and see what the real number is with full deflection of the stick.

I just started reading this pdf. It explains in detail. Hope it helps me get a better idea of what I am trying to do.
http://www.cypress.com/?docID=38317

I will have a look at the amp you suggested too.

Thanks again.

SparksnSmoke

 

Here is a link that shows a commercially built stick with the gauges on the shaft. Looks kind of messy.
http://forums.eagle.ru/showthread.php?t=53887

What sort of tolerances do the resistors used to balance the bridge need be? If they are not perfect then the system won't be centered. But at the same time the software will be used to calibrate in the end. Would a .1% be ok? I had no idea that the hi tolerance resistors come with such a large price tag!

Thanks

SparksnSmoke

 

Bump.

I posted here a day ago but I don't see it???

SparksnSmoke

 

Hello,

I have approved the post, as it was "auto moderated" (made invisible) by the software.
This sometimes happens with some sites.

Bertus

 

The "kind of messy" factor depends on how neatly you're able to do the wiring! Strain gauges involve a lot of tiny wires very close together and there's no way to avoid it.

The dummy side of your bridge has to match the active side very accurately or you'll find the output of the amplifier is so skewed that you run out of room between the supply rails. For example, if the amplifier has a gain of 500 and the bridge is unbalanced by 2mV, and you have a 5V supply, your output will sit at 3.5 (or 1.5) V with no load. You could calibrate that out, but you'd end up with more range on one side of zero than the other. Or you might have no range at all--suppose the imbalance were 5mV? A strain gauge amplifier does have an "offset" input that lets you null out the imbalance, but it's a hardware adjustment rather than a calibration procedure.

What you're buying with the high-precision resistors isn't just the accuracy, but stability over time and versus temperature. You really don't want to have to recalibrate the system very often.

 

OK so I need good resistors. I will take your word for it. But the total lack of real world experience leaves me guessing at which resistors to use. The prices are all over the place. 50 cents to 20 bucks.

SparksnSmoke

 

A quick check shows 1%, 50 ppm/C resistors in the 15 cent range.

http://www.mouser.com/Passive-Components/Resistors/_/N-5g9n?P=1yzotm4Z1z0vkozZ1z0wljoZ1z0z819

 

Ya the 0.1% ones are the ones I was looking at. But didn't know if that would be good enough. Thanks #12!

It would be nice to get all this right on the first try but I don't think I'm that lucky!

But all the help here sure does help!

SparksnSmoke

 

I took some time to do a crash coarse with Eagle schematic software. So far so good.

One thing I never thought of though. With the joystick centered it needs to be sending 2.5 volts. The bridge and strain gauges while centered will be balanced, ie no volts! So I need to add the 2.5 volts to the output??? How do I do this?

SparksnSmoke

 

Put 2.5 volts on the ref pin of course.

SparksnSmoke