Interfacing joystick

Discussion in 'General Electronics Discussion' started by SGW, Jul 27, 2015.

1. SGW

8
0
Jul 27, 2015
Hi,

I need to control an electrohydraulic system with a joystick. The hydraulic valves are proportionally controlled by a voltage of 4-12V at about 3A. I bought a couple of joysticks and some LM338's with a view to using the joystick as a variable resistor to control the output of a simple voltage regulator circuit. I have realised that there are a couple of flaws in my plan. The simple circuit I was planning to use (here http://diyaudioprojects.com/Technical/Voltage-Regulator/) increases the voltage as the variable resistance increases, the opposite of the desired effect, and secondly, the pots on the joystick have three terminals, a common and one for each direction. When the joystick is moved in one direction, there is still a resistance between the common and the other direction terminal, ie the other direction is not "off", which might lead to the hydraulic valve fighting itself.
Lastly, I have 3 axes (2 directions each) to control. I was planning to make 6 voltage regulators, one for each bearing in mind more than one axis may be used simultaneously. I know this is a knife and fork solution, any better ideas?

Sorry to ask such a basic question, thanks in advance for any help offered.

Simon

2. Gryd3

4,098
875
Jun 25, 2014
Potentiometers have 3 pins. This is less of a 'common' and 'one for each direction' as you thought.
A 10kΩ potentiometer will have Pins, A, B, and C.
10kΩ will ALWAYS be between A and C. B is the 'wiper' that will slide along the internals of the potentiometer and will vary. All the way to the left, 0Ω (or close to) will be present between pins A and B(The wiper) and 10kΩ between B and C. As you adjust the potentiometer this wiper moves... but the resistance from A-B plus the resistance from B-C will ALWAYS be the same as the total resistance from A-C.
Potentiometers are always used in one of two ways... Simply as an adjustable resistor by using pin B and one of the outer pins. (Which one dictate if the resistance goes up or down as you turn the potentiometer clockwise).
Other method is to use it as a voltage divider.. if you put 5V on pin A, and Ground at pin C ... the Pin B will vary from 0 to 5V as you adjust the potentiometer. This voltage divider method is very common, and is often fed to microcontrollers or other control equipment.

3. SGW

8
0
Jul 27, 2015
Thanks Gryd3. The thing is there is one pot per axis on the joystick, the joystick is designed to control motion in two directions (actually it is a 2 axis joystick, so there are 4 directions of travel, in this case in/out, up/down). What I am hoping for are a few pointers how to get it to control my voltage range at a few amps, ie push the joystick in one direction from the centre, read the resistance between for instance A and B and output an appropriate voltage and current to one side of the proportional electrohydraulic spool valve, push it the other way to read B-C and do the same to the other side. there are two complications with my original idea, the resistance increases in the "wrong" direction, and I need some way to disconnect the "other side" output of the joystick.

4. Gryd3

4,098
875
Jun 25, 2014
Perhaps it will help if you draw some pictures.
If the resistance increases in the wrong direction, simply work with the opposite lead from the potentiometer... (ie, instead of B-C, use A-B)
I do have a question regarding the valves though...
Are they dependant more on current or voltage?
That is... You want 'analogue' control so that these valves can be adjusted anywhere between Off to On correct? At what voltage does the valve begin to open, and is this value consistent? At what voltage is the valve all the way open, and again, is it consistent?
There are all sorts of tricks you can use, but the end result sounds like you want a single joystick to provide you with 4 separate outputs.
JoyLeft
JoyRight
JoyUp
JoyDown
And each output needs to output a variable voltage from 4-12V up to 3A.

Do you need a 'deadzone' .. ie, a portion of movement near the middle that outputs NOTHING to any output.
Is PWM an acceptable output?
Do you have a datasheet or part number for the valves you are working with?

5. SGW

8
0
Jul 27, 2015
Hi Gryd3,
Thanks for sticking with it. I thought of using the other side so to speak, that gives me about 2.7 to 5k. However it still gives me the problem of the output from the other side. The valves are buried in a machine, I was hoping not to have to dig them out, I am not sure exactly what they are. I tested them with a benchtop power supply which is how I arrived at the specs. It seems like most of the action is between 4-5V (start to open) and around 9V (max speed) drawing around 3A. It seems fairly consistent. I do indeed want analogue control, varying the voltage seems to provide this. Feedback is all with the user. You have it with the joystick logic. The joystick has a sprung centre position at which nothing should happen. I had considered PWM, but I don't know if it would be better. As you have probably gathered, I am a bit out of my depth.
The machine is a telescopic handler, a Terex 3013. It has suffered a cab fire, the original parts cost far more than the value of the machine to replace. I have all the non proportional parts working, it is just he parts that require speed control (physical speed of the boom up/down etc) left. I am aware of the safety issues, it is just for occasional private use.

6. Gryd3

4,098
875
Jun 25, 2014
I'm not going to give you a safety lecture. This is hardly a project that a 13-yr old kid could get hurt working on. Who has this kind of equipment laying around to play with anyway
We tend to avoid projects that could cause bodily harm such as modifications to cars and the improper use of high power Lasers... they are far too available to be casually talking about them.

In any case. How did the 4 outputs I mentioned above sound to you?
Do each of the valves have a 'pair' of wires, or just one?
If it is a pair... is one of the wires internally connected to ground? (You can check with a multimeter)

7. SGW

8
0
Jul 27, 2015
I am casually building an underground house in my spare time (www.underthehollybush.com), which is why I have kit like this. The handler is to replace an ancient tower crane which really is scary.

The outputs from the joystick are to be as you describe. When it is in the centre position I need to have no output.

Each valve has two sides (eg up/down), each side has two wires, one of which is ground. Everything has its own ground wire taken back to common terminals. I tested them on the machine with my benchtop supply, negative to ground and positive to the other wire, the hydraulics perform correctly and the speed of operation varies mainly in the 4-8V range. The valves seem to take about 3A (from the meter on the supply) so I went with the 338 and 5A to give me a bit of headroom.

8. Gryd3

4,098
875
Jun 25, 2014
Well... I'm not sure what you budget is, but I'd like to point out this component :
http://www.digikey.ca/product-detail/en/OPA2544T/OPA2544T-ND/254708

It's a high output current op-amp pair.
It can put out at 'least' 2A, and can push out 4A at room temperature. It needs a heatsink, as the output capability will fall below your requirement as it heats up past 100 degrees Celsius.

You know the great thing about an Opamp? Well.. they are easy to work with and tutorials are all over the net for them.
These ones are a little pricey, can be be interfaces with the potentiometers to provide the outputs you want. You will need 4 of them... each chip contains a pair, and you need a pair for each axis for each joystick. (2-axis for 2 joysticks.)
That turns out to be a little over \$130 CAD, for these, but the remaining components should be pretty inexpensive.

Although, at this point, I'd like second opinions. @hevans1944 , do you mind if I pick your brain for an ideal way of handling this?
I'd like to suggest using the opa2544, but I have very very limited experience in variable voltage output at high currents.

9. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
I am not familiar with the Texas Instruments power op-amp product. In the past I have used a hybrid power op-amp, the PA-12, made by Apex Microtechnology for various low-frequency applications up to a few kilohertz driving inductive loads. I think the TI product is an improvement because it is not a hybrid: everything is contained on one piece of silicon.

However, I think a power op-amp is way overkill for this application. Firstly, a bi-polar output is NOT required. Second, the feedback is human-in-the-loop rather than electronic, so frequency response is limited to a few hertz. The three-terminal voltage regulators would seem to be a good choice if a means to control the hydraulic valves using a joy-stick can be implemented.

In another thread, started by @pgib8 an LDO three-terminal regulator is used to control the speed of a motor. A small microcontroller drives an external 6-bit DAC resistance ladder network inserted in the feedback path of the regulator. A separate enable pin on the regulator is used to turn the motor on or off. All that is needed for this application is six regulators and six DACs and one microcontroller to read the joy-stick position signals and send digital voltage-command levels to the valves via the regulators. The microcontroller handles the job of sorting out the signals from the joy-stick potentiometers, creating a dead-band about the zero (released) position, and scaling the digital signal to the DACs to implement a 4 or 5 V to 12 V output range for each valve.

Gryd3 likes this.
10. SGW

8
0
Jul 27, 2015
Thanks hevans1944. Your suggestion sounds good to me, and I noticed Gryd3 also approves of it. I read the other thread you referenced, but to be honest I didn't learn much from it as I am way out of my depth (I am a mechanical engineer with virtually no electronics knowledge). I realise you are not there to design my circuit for me, but if you could give me a few more pointers I would be very grateful. Which DACs would you go with, and which microcontroller, bearing in mind ease of programming is high on my list.

11. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
@SGW I did a little web surfing for R-2R ladder networks and DACs and found some inexpensive Bourns thick-film R-2R components readily available. These would require external transistor switches to select which elements are enabled and disabled to form a voltage-divider feedback network to set the output voltage of the regulator. It has been some twenty years since I used these networks (assembled then from discrete resistors) to implement digital-to-analog converters. I will have to do a little research to see if anyone has used them to control three-terminal voltage regulators and how they approached the problem.

I visited your website and looked at the pictures. <sigh> I wish I had your talent. My wife is an artist, but I don't think we would even think about attempting what you have done. Beautiful job. You and your wife should enjoy this for years to come.

You probably don't have time to learn the electronics you need to control your six hydraulic valves. However, I am retired now and do have a little time available to play around with this sort of thing. If you can tolerate the wait (it could take weeks or months to finish) I will prototype the circuit you need and instruct you in how to build it. We need to use the KISS (Keep It Simple, Sir) principle here. I also need to make sure we both are playing on the same cricket field.

Could you describe the joy stick controls you have purchased in a little more detail? What was their original purpose? Pictures would be nice along with an internal wiring diagram. Most joy-sticks have a single potentiometer for each axis as @Gryd3 described. In the released, or centered, position internal springs position the joy-stick straight up and down and the wiper of each potentiometer is approximately centered between the two ends of the potentiometer. From there almost any electrical configuration is possible. Forget about trying to somehow interface the joy sticks directly to a voltage regulator to control the valves. It isn't quite that simple, although it isn't nearly as complicated as I first thought.

Imagine, if you will, that you have two twelve-volt batteries connected in series, so together they produce 24 volts. Call the wire connecting the two batteries together "ground" or common. Then one battery will measure a positive 12 V with respect to ground and the other will measure a negative 12 V with respect to ground (assuming you have the positive terminal on one battery connected to the negative terminal on the other battery). Now imagine connecting the +12 V battery lead to one end of the joy-stick potentiometer and the -12 V battery lead to the other end of the same potentiometer. You now have 24 V applied across the potentiometer. If you connect a voltmeter between the potentiometer wiper terminal and the ground terminal of your two batteries, you will measure a voltage that is approximately zero when the joy-stick is straight up and down. As the joy-stick is moved in one direction away from center, the voltage will either increase from zero towards +12 V or increase from zero towards -12 V depending on which direction the joy-stick is moved.

You could theoretically use the polarity of this joy-stick signal to determine which one of the two valves gets actuated. Diodes could be attached to the wiper of the potentiometer to separate the positive polarity signal from the negative polarity signal. You would then apply the positive polarity to an LM338 or LM350 to control one of the valves. Of course, the negative polarity is the wrong polarity to control a second LM338 or LM350 for other valve. However, it could be inverted easily enough with an operational amplifier inverter to produce a positive polarity, and that signal applied to the second LM338 or LM350 adjustable regulator.

With this approach, all you need is a mechanism to turn the regulators off when the joy-stick is centered. I am in favor of using electro-mechanical relays for that function. The voltage from the wiper arm of the joy-stick potentiometer would have to rise above a certain positive or negative voltage before the associated valve relay would energize and allow power to be applied to the valve solenoid spool. A simple comparator (or even an op-amp functioning as a comparator since it doesn't have to be fast) can be used to actuate the relay coil. You would need six of them, along with three op-amps to invert the negative voltages from the joy-stick potentiometers. And a handful of signal diodes to sort out the positive and negative potentiometer output voltages. Along with a dual positive and negative 12 to 15 V DC power supply we're looking at perhaps fifty to hundred dollars worth of parts and hardware to package everything up in a nice weatherproof enclosure. You should use battery power since the machine is designed to move around. I will assume it has an internal combustion engine to provide prime-mover power for the wheels and hydraulics and alternator to keep its battery charged. The dual power supply should be an isolated DC-to-DC converter powered from the main battery. The voltage regulator for the hydraulic valves will also be powered from the main battery. All the dual power supply operates are the op-amps and relay coils, so it only needs a few watts power capacity.

I hope I haven't forgotten anything. It looks like you won't need a microprocessor or a DAC if I can figure out how to adjust the output of an LM338 or LM350 with a voltage or current from an op-amp instead of with a variable resistor (that's what the R-2R ladder network was for).

So it appears you were on the right track after all. I hope my explanation above cleared up any confusion you had about how you use the joy-stick potentiometer to select which of two valves will receive power. The joy-stick does need a "dead band" about the centered position where nothing happens for a small movement of the joy-stick in any direction. The comparators and relays take care of this, requiring perhaps one volt, positive or negative, to energize the appropriate relay and send power to just one valve.

Hop

12. SGW

8
0
Jul 27, 2015
Wow, what a helpful and generous offer. Thanks very much. And thanks for your kind words about my house project. It has been a long job with much more to do, but we are getting some benefit now.

Are you sure it is necessary for you to actually make the circuit though, it seems a lot of trouble for you? If you were to design it for me I would be happy to make it and debug it under your supervision.

The joystick I have is as simple as they come. a sprung loaded 2 axis stick with a 5k pot for each axis. The only connections are the three terminals of each pot. You can see it here: http://www.ebay.co.uk/itm/261893495106?_trksid=p2057872.m2749.l2649&ssPageName=STRK:MEBIDX:IT

I think keeping it simple is an excellent idea (it will still look pretty complex to me!).

The machine is a regular telescopic handler as used in the construction industry, it has a diesel engine and 12V electrics. The component costs are no problem (the standard joystick is about USD 1500, replacing all of the burned out electrical items in the cab would be well over USD 10k. As to timescale, the machine is currently on a friend's farm. He wants to use it to erect a barn in exchange for helping me with it, and I promised I would get it working by the end of September. Obviously though, if it is not ready by then it is not ready, that was just my hope.

Thanks again for all your help.

Simon

13. Gryd3

4,098
875
Jun 25, 2014
@hevans1944 :
Single rail power supply, does not need isolation.

Potentiometers from the joystick will be used as voltage dividers.
The wiper from each potentiometer will be used to control the circuit.
2 Opamps per axis will be used, with the wiper of the potentiometer being connected to the non-inverting input of OpampA, and the inverting input of OpampB.
Please correct me if I'm wrong here, but the single supply Opamps can both be adjusted with their remaining input, which will allow for the adjustment of the 'deadspace' for the joy stick. The expected behaviour is that the Opamps will be driven to ground (or close to) when the joystick is at rest. When the joystick moves, one opamp will remain at ground, and the other will begin to output a varying voltage.
Once this portion of the circuit is complete, ideally, you will now have 4 analogue outputs that range from 0V (Or very close) to VCC.
A 555 timer will be used to generate a PWM frequency. Each of the analogue outputs, as well as the 555 timer's output will be run to a comparator. The end result will be a PWM control signal for each of the 4 outputs described above which could be used to drive the valves.

I'm at work at the moment, but that popped in my head. I'm sure there is a better approach, but it would be solid state, and would not require a microcontroller.

14. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
@Gryd3 I don't think this would work. Op-amps require negative feedback to produce linear outputs. You cannot just apply voltage to the inverting or non-inverting inputs of op-amps and expect the output to be anything other than fully at one limit or the other. But, sketch up a schematic and post it because I may have misunderstood what you are suggesting.

Yes, each potentiometer is used as a voltage divider. With a single power supply, the output with the joy-stick centered will be one half the supply voltage. I am suggesting that a bi-polar supply with + and - output voltages applied across the ends of the potentiometers will allow zero output when the joy-stick is centered, and then positive or negative outputs when the joy-stick is moved.

The current requirements are very modest, a few hundred milliamps for all the op-amps and all the potentiometers, so a switching dual-output ±12 V power supply would be adequate. Using two power supply rails really simplifies op-amp applications.

I don't think PWM is necessary for this application because the 3-terminal adjustable regulator provides a simple interface between the potentiometer voltages and the valve solenoid voltages. It is an off-the-shelf solution whereas a PWM solution requires more design effort. However, since I do have the parts on hand, I will see what a 5A PWM driver would involve. The valve solenoids are drawing about 3 A at 12 V (fully open), so significant power is dissipated in the 3-terminal regulator under those conditions. It may require a substantial heat sink even though the operation is intermittent. There must be sufficient time for the heat sink to cool down between valve operations, and for the temperature rise of the heat sink during operation to stay within the published limits for the regulator. I will calculate a thermal analysis to see how much heat we may be expected to remove with the linear approach.

Although more complex in terms of circuit design, a PWM solution does not require heavy heat sinks for the switching transistor because it dissipates zero power when off and very little power when on. It would be nice to know what the inductance of the valve solenoid is, but this is probably information that is not available. I must assume that a PWM frequency on the order of a kilohertz would be adequate, well within the capabilities of the NE555 timer.

@SGW I prefer to breadboard one-off designs before making a commitment to a circuit board. This allows "what if" experiments to select an optimum design. Others here would do a software simulation before constructing anything. I don't really trust simulations because in the end you have to use real hardware to prove a design works. A simulation can tell you if you have wandered off into left field and perhaps point to a way to get back on track, but I prefer to wire up a breadboard and go from there. We'll see if a working model, ready to package into a weatherproof box, can be produced by the end of September. It only takes a few weeks to get a circuit board manufactured after the final circuit is tested and cast in stone.

I have a question about the joy-stick: does it have a third potentiometer connected to the joy-stick that allows you to rotate the stick while moving it side-ways, thus providing a third axis? If so, are you planning on using this third axis for anything? This would also be a good time to think about how you are going to mount the joy-sticks in the cab, convenient for the operator to use. Plan on using a waterproof circular connector to make connections between the joy-stick box and control electronics, and one or more waterproof connectors between the control electronics and the valve solenoids, and a waterproof power connector on the electronics box to connect to battery power. Is there room to mount the control electronics under the seat?

Hop

Last edited: Jul 29, 2015
15. Gryd3

4,098
875
Jun 25, 2014
Yeah.. I forgot to include that in my description. Feedback would be a requirement, and would be adjusted to tune the 'full-on' position of the joystick.
I'll play around with some drawings later tonight and see what I can provide for you.

Thank you for popping in and helping with this.

Edit : The term I was looking for was difference amplifier...
The wiper of the potentiometers would be connected to difference amplifiers that will output 0V to Vcc/2 with a gain of 1.

Last edited: Jul 29, 2015
16. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
Well, I am kinda stuck on the Flashlight Project because I haven't heard anything from Jeff since last May. I did finally get my PIC10F206 programmed to sip less than 10 μA current while still recognizing button presses, and turning off the flashlight LED for 1 ms every 135 ms if the LED is on. The trick was to use a 10 kΩ pull-up resistor for the switch, but connect it to an unused output port instead of Vdd. Each time the PIC wakes up, the program sets the port output to 1 to activate the pull-up resistor. It then reads the switch state and stores it in a user register. Then it clears the port output to 0 and finishes its other business before going back to sleep.

The program spends most of its time asleep, waking up to exam the push-button switch every 135 ms and staying awake during the 1 ms re-charge interval where it is busy running a software timer to time that interval. The only time the resistor is drawing current is during a 15 μs interval each time the PIC wakes up and then only if the switch is pressed. It works beautifully, and a 1 μF energy-storage capacitor only drops about a volt every 135 ms, most of that draw attributed to the watchdog timer.

I have some time available for this project because I am not going to continue and finish the programming for the Flashlight Project until I am sure those 1 ms interruptions every 135 ms don't wreak havoc with the LED controller on the other end of the flashlight. IMHO it would be a waste of time to build this and find out the LED controller responds to the 1 ms interruptions by re-programming itself! The only reason Jeff needed this circuit was so he could obtain a high-current push-button switch that was capable of some rudimentary programming. Well, I have a program that does that now, but if the LED controller on the other end of the battery responds to the 1 ms interruptions in current through it... well Game Over! I have no solution to that problem, unless Jeff can machine a slot in the battery well to embed a wire from the front end of the flashlight to the rear, so as to pick up the positive battery terminal and power the PIC even when the flashlight LED is on.

So, anyway, I will dig out something to simulate a pair of solenoid valves and see what it takes to drive them with a joy-stick. I happen to have a large joy-stick on hand that was intended for video gaming, but I have never used it for that or any other purpose. It just sits on a shelf collecting dust. This seems like an ideal opportunity to open it up to see what's inside!

Hop

Gryd3 likes this.
17. SGW

8
0
Jul 27, 2015
Hi Guys, thanks for all your efforts. I was sort of laying low as I have little (nothing!) to add to your solutions.

Hop, I didn't mean to tell you how to do it, I am just concerned about the amount of time you will be putting in for me. If you prefer to build a working prototype that is great. There are three things that require speed control, boom up/down, boom in/out, and what they call crowd, which is tilting the forks. I did actually buy one of those 3 axis twisting joysticks you described, but I think I would prefer to use the two simple two axis joysticks I have (one using only one axis). I think this will reduce the chance of accidentally operating the wrong axis.

I have fabricated a dashboard from aluminium with a fair amount of room beneath it. I already have quite a lot of controls on it, ignition switch, emergency stop, jacks, tilt, hi/low gear, steering mode (front wheel, four wheel, crab) etc. All of this except the ignition and stop controls electrohydraulic valves in a simple on off way. If there is not enough space, there are plenty of other places in the cab including under the seat. I have reserved the most convenient position for the joysticks. I understand the need for it all to be waterproof.

It is very difficult to get to the valves themselves, so I don't know what model they are to find a spec sheet. If it is imperative to do so I could, but I would have to disconnect the hydraulics and unbolt them to get at the model number.

Simon

18. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
@SGW Don't bother trying to get to the valves to read whatever grime-encrusted information might be there. I don't think it is necessary to actually identify the solenoid valves by manufacturer and model number or part number. The fact that when 12 VDC is applied about 3A of current flows says a lot... about 4 ohms DC resistance... and any more information, such as inductance, is not necessary for this type of operation.

The only reason I want to build a prototype is to make sure the direction signals (boom up/down, boom in/out, fork tilt) are properly derived from the joy-stick position potentiometers and the signals are separated for application to the appropriate valve. The prototype would not be used to actuate real valves. It's only purpose is to verify the proper operation of the control circuits.

Gryd3 likes this.
19. SGW

8
0
Jul 27, 2015
OK, it is in your hands. Thanks.

20. hevans1944Hop - AC8NS

4,543
2,116
Jun 21, 2012
I am "bumping" this thread for further comment. I offered to design the circuit for @SGW back in July 2015 but have dropped the ball on getting a parts list together. This was supposed to finished by last September! We have been in communication privately, both here and via e-mail, but now it is time to throw this back to the forum for further comments to see if it can be finished. The OP has created a schematic from our discussions, but I haven't been able to check it yet. If someone would care to separate the op-amps from their block diagram representation, and re-draw them as four individual symbols, with labels for all the components (R1, R2 ... U1, U2... etc.) it would make it much easier to discuss.

The attached schematic basically shows one of three joystick axis controls. Bi-polar voltage applied to the joystick potentiometer allows direction determination, with voltage comparators providing plus and minus dead-bands around center where no output is desired. Between the mechanical limit of joystick movement and the end of each dead band there will be a linear voltage applied to the selected control valve by means of an LM338 3-terminal voltage regulator capable of supplying 5 A at outputs between 4 to 12 V. The minimum output should be adjustable but that isn't shown on the schematic yet. The output should be linear with joystick movement.

The design approach is this:

First, buffer the joystick potentiometer outputs with a unity-gain op-amp buffer so subsequent signal processing does not load the potentiometers. Next, apply the joystick signal to two comparators and an absolute-value circuit. The two comparators establish the two dead bands around the joystick centered position and control application of output power through relay contacts to either one of the two solenoid control valves, based on the direction the joystick is moved from its centered position. This arrangement guarantees the two solenoid valves are not energized simultaneously, and that both valves are unenergized when the joystick is centered. The absolute value circuit provides the output control signal for the LM338 providing a variable drive voltage to either valve. The absolute-value circuit does not have to be very accurate around zero input (which is where many absolute value circuits have difficulty) because power is not applied to the valves until after the joystick is moved outside the dead band. When that happens, a substantial positive or negative signal is presented to the absolute-value circuit which then produces a positive output. The absolute-value circuit output is irrelevant and unused until the joystick displacement exceeds the dead band.

So, we need some help here (since I currently do not seem to have the time or bench facilities to complete a prototype) coming up with a list of parts. The OP has already selected an acceptable DC-DC bi-polar output power supply for the joystick pots and the op-amps. The LM338s will receive power directly from the battery used to start the loader.