I'm an avid PC combat sim pilot, and I wear out Logitech Extreme 3D Pro joysticks at an alarming rate. The pots in the X and Y axis (more so the Y axis) tend to get dirty and wear out over time and lose centering and become jittery. This causes a loss of accuracy and erratic behavior.

I prefer this particular joystick for a number of reasons. It's cheap. Usually can be purchased for $25-$35 depending on the vendor. The handle fits my hand perfectly and is very comfortable, and the thumb buttons and trigger are very robust. It's the perfect joystick for the money, if only the pots didn't fail so often.

Temporary relief can be had using Deoxit D5 contact cleaner. This stuff works wonders, but it's expensive ($17 for a 5 oz. spray can), and once the pots start acting up, they need to be sprayed repeatedly. Deoxit is the best contact cleaner I've ever used, and it will last a while, but it's a pain having to dismantle the joystick to use. Eventually even the pots degrade even beyond the help of the Deoxit.

I've also owned the Thrustmaster T16000M joystick which uses Hall Sensors rather than pots. The accuracy of this stick is amazing! Absolutely no jitter. But, the stick is more expensive, and it's ball and socket mechanical design wears and gets loose. The buttons are flimsy. The stick handle itself doesn't have that ergonomic feel of the Logitech stick.

I ran across this post in a forum by a guy that claims he replaced the X and Y axis pots with hall sensors, and put cylinder magnets in the axis. He claimed it works, but he didn't give any details about what sensors or magnets he used, or if he had to add other components to the circuit. There's a picture of his mod. Another poster in that thread was also attempting to mod his stick, and he claimed to know that the other guy used SS496A sensors. I can't tell from the picture if that's correct, so I'm not sure how he knew (perhaps the two were collaborating in PM).

http://forums.ubi.com/showthread.php/85431-Successful-repair-of-Logitech-Extreme-3D-PRO/page3

I decided to take a chance that this would work and purchased two SS496A sensors, and two diametrically poled cylinder magnets. I soldered up a wire harness and tested one out, and it didn't work

On the datasheet for the SS496A, it says it requires a min. supply voltage of 4.5V. I measured 4.2 - 4.4V across the V+ and V- terminals of the pots in place, and the output varies from 0 - 4.2V as the pot is rotated through it's entire range. Is the min. voltage for these sensors that critical? Maybe they're not working because it's just not getting enough power? Is there a sensor that would work better?

 

You need resistor across VCC and D. Out.

Vcc is always the same , in your case its 4.2V it doesnt change.Output pin sends out variable voltage between 4.2-0V which changes accordingly to magnets proximity/position, Ouput pin needs to be connected to Microcontrollers ADC pin.

When the stick is in neutral position sensor should output half of the VCC's voltage otherwise joystick is going to be out of calibration.

 

Why do I need to add a resistor?

According to the data sheet for the SS496A, with the flat side up, and the pins down, the left pin is the Variable Output, the middle pin is ground, and the right pin is V+. It's hard to make it out in the photo posted by the guy that did the mod, but it appears that's how he has his wires soldered. The red wire is V+, black is ground, and white is the variable output.

I soldered mine up the same way, and nothing happens when I move the magnet near the sensor. The cursor never moves in the joystick properties window. If I test voltage with my multimeter across the red and black I see a flash of 3 something volts, and then it settles to .5 volts.
Measure from the white pin and black pin, I also just get .5V, regardless of moving the magnet near, rotating, changing orientation, etc.

Not even sure I have the correct type of sensor. It says it's a "ratiometric", so I took that to mean the output voltage varies with the rotation of the north or south pole of the diametrically poled cylinder magnet.

 

Could you post a clear pic of your sensor/wiring? Can't help but wonder if it's backwards.

Also, can you temporarily test the sensor apart from the joystick with a separate 5V supply and a multimeter? I wouldn't expect the behavior you're describing (reading only 0.5V across supply side of sensor) if it were simply a little underpowered, but could definitely see it not working properly at a reduced supply voltage.

I've done a lot of experimenting with the SS495 sensors and I've been quite impressed with what can be accomplished, so don't give up. There should be lots of potential here!

 

My wiring is like in the picture in the link I posted above. I only hooked up one to test. It works with just one pot hooked up, so I'm presuming that only one SS496A should work by itself (for that axis of course).

According to the data sheet for the SS490 series, in their diagram, with the lettering beveled side up, and the pins down, the left pin is +, middle pin is -, and right pin is O.

http://sensing.honeywell.com/honeywell-sensing-ss490-series-solidstate-product-sheet-005843-1-en.pdf

So, like the guy that did the mod in the photo above, I oriented the sensor so the flat side faces the magnet, and the red + wire is soldered to the top or left pin (facing the beveled side), the black - wire is soldered to the middle pin, and the white O wire is soldered to the bottom or right pin. I tried this with both sensors, in case the first one I tried was a dud, and I get the same results with the second one.

Here's the link to the magnets I got. They're strong little suckers, and were hard to pull apart. Is it possible they're not strong enough to activate the sensors? If anything, they should be too strong...the data sheet for the sensor shows a linear graph of 0-4 volts on the output between - 640 to 640 gauss.

http://www.kjmagnetics.com/proddetail.asp?prod=D36DIA

 

Ok, sounds like the wiring should be right. If I were you, I would try two things:
1) Test one of the sensors with a separate 5V supply to see if perhaps the voltage being below 4.5V is the problem.

2) Disconnect the sensor from the joystick wiring and verify the supply voltage on the red and black wires with nothing connected to them. To me, the lack of voltage there seems far more alarming than any difficulty on the output side of the Hall sensor. If you find the 4.2-4.4V there again (like when you had the original pot in,) then you need to figure out why the Hall sensor is dropping it to almost nothing. If you don't, then the problem would seem to be elsewhere in the joystick.

Also, with one sensor in, does the pot on the other axis still work normally, or has the installation of the Hall sensor knocked out both axis?

 

I've tested the wires without either the pot or sensor connected, and I get about 4.4V across the red and black wire. The joystick works with the pots in place (albeit poorly because of the degraded nature of the pots). Would one tenth of a volt shy of 4.5V be low enough that the sensors don't work? It does say the min. is 4.5V on the sheet.

With the sensor in place, the voltage across both red and black wires at both plugs is 0.5V. Unlike in the picture in the mod above, the wiring harness on my stick daisy chains the red and black wires from the X axis plug to the Y axis plug, but they each have their own white wire (the output).

I'll make a photobucket account and get a picture up.

I don't have a power supply handy. But, I've been meaning to make one out of an old 430 watt PC power supply I have leftover from my last PC upgrade. I fly RC 250 quad rotors and 450 helicopters, and could use one to power a second lipo battery charger anyway. It will have 5 and 12 volt rails to use.

 

I'll make a photobucket account and get a picture up.

It is preferred that you post any pictures pertaining to your post here on AAC. Use the 'Upload a File' button. Use a suitable resolution and picture size to keep file sizes down to around 300K or less to accommodate users with slower connections. The mods will appreciate it as well!

Welcome aboard!

 

The full wiring harness, with just one sensor installed. You can see that the red and black wires are shared between the two plugs, but each has their own output wire, which terminates into a 4 pin plug that connects to the controller board.



And here's a close up of the beveled side of the sensor with the wires soldered on.



Sorry, I know these photos are bigger than they're supposed to be.

I took pictures of one of the pots, and the controller board, and can post those if they'll help.

Mod edit: reduced image sizes

 

Thanks for the pics! Wiring certainly looks right to me.

Probably another dumb question, but I like to cover all the bases: You weren't doing this testing with the sensor sitting on that anti-static bag, were you?

 

No, it was on my desk which is covered in a non conductive laminate.

 

No, it was on my desk which is covered in a non conductive laminate.

Cool. I figured as much, but it never hurts to check.

Well, aside from the 5V power supply test, I'm about out of ideas. One last simple question - did you verify black/red polarity yourself or just going off the pictures? Is it possible black is positive and red is negative, even though this seems wrong? Assuming no polarity problem, it seems like the low supply voltage is the only explanation left, or at least the only one I can think of.

 

Do you mean the polarity of the wires in the joystick? Or the polarity of the pins of the sensor shown in the data sheet for the SS496A?

The data sheet shows + on the left pin, - in the middle, and O on the right pin, with I believe the bevel side up. If it's actually the flat side up, then the - pin would still be in the middle, but the O and + pins would be reversed. I could try swapping them, but if that's wrong, I think it will burn them up and release the magic smoke

I used Digi-key's online support chat, and their rep. spoke to a Honeywell rep. on the phone, and verified that the diagram in the datasheet is correct. So, I think I have it wired right. It may just be that the 4.5V threshold is critical.

Thanks for your help, by the way

You said you have experimented with the SS495 sensor? What's the difference between the 495 and 496? Can you turn your power supply down on one of your SS495's to below 4.5V and see if it stops working? What magnets did you use with them and for what applications?

 

Do you mean the polarity of the wires in the joystick? Or the polarity of the pins of the sensor shown in the data sheet for the SS496A?

I meant the joystick wires in this case. I think we've pretty well covered the sensor pin configuration, and I'm confident you have it right. Just wanted to make sure the joystick side of things is also doing what we expect. Seems unlikely to be wrong, but you'd be surprised how many seemingly unlikely mistakes I run into in my job. I never assume anything is right anymore. On a related note, please don't be offended by my dumb questions; it's not that I doubt you, just that I like to double check everything.

Thanks for your help, by the way

Well, I'm not sure I've actually helped yet, but I'm happy to keep trying!

You said you have experimented with the SS495 sensor? What's the difference between the 495 and 496? Can you turn your power supply down on one of your SS495's to below 4.5V and see if it stops working? What magnets did you use with them and for what applications?

I believe the only difference is their sensitivity. Yours requires a stronger magnetic field to produce the same output voltage.

Unfortunately, I don't have a variable supply, so not quite that straightforward to test, but if I get a chance tonight I could try throwing a diode inline to drop the 5V to 4.3V and see if it works.

The specific application is proprietary, but essentially the main project was creating a precision non-contact switch. We were using reed switches, but the trip point varies quite a bit from one to the next, so we had to physically align each assembly differently to get the action we wanted. Hall switches I found had equally wide tolerances on their trip points. Hall effect sensors on the other hand are available with fairly tight tolerances, so by combining a Hall effect sensor with a comparator, I was able to make a reed switch replacement which doesn't require physical adjustment. All the parts are interchangeable, with no calibration needed, making our manufacturing that much simpler.

I've also been experimenting with using their output without a comparator for analog position sensing. The field strength decay over distance is non linear, so it takes a bit of work to get just what I want out of it, but I'm getting pretty close now.

 

Actually found a quick way to test the low voltage question. Used a diode drop to get to 4.1V, and one of our boards (2x SS495A, comparator, and LED) ran just fine on that. Of course two sensors tolerating under-spec voltage is no guarantee that they all will, but it is promising.

 

I don't have a power supply handy. But, I've been meaning to make one out of an old 430 watt PC power supply I have leftover from my last PC upgrade...
It will have 5 and 12 volt rails to use.

Did you get a chance to test the sensors on a separate power supply? I'm curious to see what the problem turns out to be.

 

Did you get a chance to test the sensors on a separate power supply? I'm curious to see what the problem turns out to be.

Not yet. I bought another new Logitech Extreme 3D Pro to hold me over till I can figure out the hall sensor thing...if it can be done.

Sure wish I could track the guy down that did that mod. Why post pics of it and brag about how great it is and not make a nice detailed walk through for others to follow in his foot steps?

I'm not sure if I need magnets that are a very specific gauss strength to make them work properly. Or if I'd get the correct amount of axis range out of them if I did get them to work.

I'd really like to figure this out and thumb my nose at Logitech for not honoring their warranty on the last one...but it's not looking good at the moment.

 

Not yet. I bought another new Logitech Extreme 3D Pro to hold me over till I can figure out the hall sensor thing...if it can be done.

Sure wish I could track the guy down that did that mod. Why post pics of it and brag about how great it is and not make a nice detailed walk through for others to follow in his foot steps?

I'm not sure if I need magnets that are a very specific gauss strength to make them work properly. Or if I'd get the correct amount of axis range out of them if I did get them to work.

I'd really like to figure this out and thumb my nose at Logitech for not honoring their warranty on the last one...but it's not looking good at the moment.

It wouldn't surprise me if you need to experiment with different magnet strengths to get the response you want, or maybe just adjust the distance between the sensor and the magnet.

Either way, the bigger problem is having your supply voltage drop out. There's no way that the sensor sensitivity or magnet strength should cause the supply voltage to drop to 0.5V. Only after that problem is fixed can you work on the sensitivity, strength, distance balance.

 

@Nihilistic

It looks like you have picked the correct sensor for the application given the magnet you are using falls with the gauss range listed in the data sheet. Also, looking at the transfer characteristic charts you will see that nothing below 4.5v is even applicable. Meaning the device likely will not operate below 4.5v.

Likely the device is not conducting below 4.5v which is why you see 0.5v across the sensor. Grab a 9v battery and test the sensor. If your supply voltage is actually less than 4.5v, you will need to find a 4.5v+ power source in that controller.

 

I borrowed the 9V battery out of my food scale and hooked up the + terminal to the left pin of the SS496A (with the beveled side up), the - terminal to the middle pin, and then measured across the + and - pins and got about 7.5V (battery is a little pooped out). I then measured across the middle pin and the output pin with the magnet near the flat side and I got 2V. I then rotated the magnet a bit and the output voltage changes with the rotation of the magnet, just as the pot would.

So, apparently the sensors are good, and as you've suggested, they're simply not getting enough supply voltage to operate properly.

I can't tell if the variation of voltage in proportion to the degrees of rotation is close to the same as the pot, because I lack the proper testing equipment like breadboards and alligator clip leads, so I was moving the magnet while holding the test probes to the pins at the same time...kind of awkward. Maybe the joystick calibration routine would take care of that anyway.

I guess I either need to get sensors that operate on 3.5 - 4 V or step up the voltage there.

With the pot removed, I measured 4.42V with the red probe on the red wire of the plug where the Y axis pot plugs into, and the black probe to the black wire. With the black probe to the black wire of the plug, and the red probe to the white wire of the plug, I get about 2V.

With the pot plugged back in, I measure about 4.1V across the red and black wires. I measure about 2V with the axis of the pot centered, and it varies between 0 and 4 V through the range of rotation of the pot axis. The actual operating rotational range is much narrower than the pot axis is capable of moving. When installed, the flat side of the pot axis is facing the 3 O'clock position. When the stick is fully pushed forward, the flat side of the pot axis is approximately facing the 2 O'clock position, and at the 4 O'clock position when the stick is pulled fully back. So, we're looking at a range of rotation of about 66 degrees.

I tried plugging the SS496A back into the Y axis plug, but left the X axis pot in place. Naturally the axis with the sensor plugged into it doesn't work, but neither does the one with the pot plugged into it. If I leave only one pot plugged in with nothing attached to the other plug, the axis with the pot plugged in works normally.