Skiz's 3DOF 90v DC Build

So I've always wanted a motion simulator, and after years of drooling, learning, and finally deciding which way I wanted to go, I finally pulled the trigger on the project. I've spent years with embedded systems I'm a software engineer by trade, and I've had a flight simulator affliction for many years.
I have a pretty decently equipped home garage/shop (mill, lathe, press, MIG welder, etc), and have access to CNC, and a plethora of other tools (and experience) through my company and other friends.

Given a few children who are just as anxious as I am about this project, why not get started?

The platform consists of 1" steel square tube in a custom design that provides enough room to fit a larger captain style seat along with side panels (DCS), and plenty of extra leg room and mounting area up front. I imagine this will be wrapped in aluminum diamond plate once everything starts to come together.

[PIC: PLATFORM FRAME - SOON]

After looking at all the standard consumer options, DIY hacks, and prohibitively expensive professional systems, I decided that a DC powered 3DOF would be the way to go for size, cost, and maintenance.

Starting with the motors, I chose three 1/2HP permanent magnet 90v DC Motors for a few reasons. First, this keeps the motors\ at around 15 amps on a single 110 outlet (generally 20 amp) for home use. Second, they were not as expensive as some other options, and third they are NEMA 56C standard size motors for a reason...
[PIC: MOTORS]

1750 RPMs is a bit fast for a motion platform, but it's a great place to start. After doing countless calculations, some visual experimentation, and some hand wavy assumptions, I ordered three 25:1 worm gear boxes that matched up nicely to the motors due to the NEMA standards, leaving me with a 3/4" output shaft turning at an acceptable speed. If you want to do the math and tell me I'm an idiot, or tell me about something I completely disregarded, I'd greatly appreciate it.

[PIC: MRV040 Worm Gear 25:1 56C]

Given the rating from the manufacturer, these looked like they should be adequate to handle just about anything I would expect to throw at them for a single seater. Unfortunately, 90v DC is a bit of a beast to work with given that I want to control the motors and not just make them spin. To manage this I picked up three KBMG-212D motor controllers. These are made to handle up to 1hp 90v motors, have regenerative & coasting braking, and plenty of features in a small overpriced package. Unfortunately they don't come standard with signal isolation for the bipolar 10v control line (more on this later), so I obviously didn't want to skimp here, so added those in also.

[PIC: MOTOR CONTROL & ISOLATOR]

Since I'm working my way from top to bottom on this thing, it's only proper to buy a bunch of stuff I assume I will need at some point so I'm not waiting for days for it to arrive (Good excuse?). Things like 1/2" threaded stainless rod, 1/2" threaded ball joints & rod ends, nuts, bolts, lots of various gauge wire, heavy duty leveling feet, 1/2" blue PVC (for the rod), Arduino UNO R3, you get the idea. Then I was thinking, where the hell am I going to mount all of this electrical? A box of course.

[PIC: PILE OF PARTS AND BOXES]

Since I needed a control box, I found something I thought would be adequate on Amazon. But of course it needs a way to keep the kids from messing with it, so I found a barrel lock switch to disable the machine completely as needed, or launch a nuke. Of course being safety conscious, we're also going to need an emergency stop, fuse access, a standard power button, and a big warning sticker. Oh, and probably a nice panel mount USB extension, wiring blocks, and other tidbits.

[PIC: BLUE BOX, BARREL LOCK, E-STOP, FUSE HOLDERS, USB EXT, ETC]

The next step is to get the control box wired up. Getting the AC to the DC safely and having everything wired up internally is the first order of business. We want everything properly fused and the emergency stop working right from the start. Any cool lights are just a bonus, but we want to get a motor spinning ASAP, and these come with a potentiometer to do just that (as an initial test for us anyway).

After that I'm going to be designing and welding up the lower platform from 2 1/2" square steel tube based on some hand wavy stability requirements and visual checks, balance the upper platform, and finally mounting the motors and testing the actuation mechanisms (arms).

I've purposely left out the control loop (positional) and software (arduino, xsim, etc) because 1) I have several hair-brained ideas about dirt cheap positional feedback, and 2) I'll get there when I get there. Although... I did buy a set of linear transducers, but I'm seriously going to experiment with ultrasonic and IR based solutions. If any of you have tried this, I'd love to know your thoughts.

Finally, I expect to have to redo any number of these steps and take different approaches, but you'll never learn anything new if you don't fail right?

Expect the missing pics and more parts porn to be added/edited shortly.

Interesting approach ! I envy your garage and tools that you have and will be closely following your project. I wish there was a better option for feedback over using potentiometer and hall sensor with xsim as well.

It's been a busy month, but after some work in the garage, and a long break I have completed the AC side of the motor control unit.
I'm happy with all the ratings, fusing, wiring, and especially the key and panic button.


I finally got my digital potentiometers in today. I'm anxious to get started with connecting the Arduino up to the motor controllers via the fresh digipots.

I have decided to go with digital potentiometer control route for the motor controllers as I can easily get decent speed resolution (256 steps),
and the motor controllers support 5k pots for controlling direction and speed and these things are easy to control with SPI (MCP4251-502E/P).
Another nice thing is that they also default to 2.5k which should be very close to, if not stopped, along with enable lines and other features.

On a side node I seem to have lost (hid from myself?) my platform frame around here. If you see it laying around, kindly point it out. As for now it's software and circuit time.

I'm still also deciding how I want to close the loop, but I may cheat and use these "cheap" linear 5k pots as a fall back, but I still want to investigate other sensor options.

Okay, I promised more parts porn, so here is what I remembered to take pictures of thus far.


And finally my electronics bench that needs some cleaning (and more surprises in those boxes to come!)

So I finally figured out an easy way to deal with industrial +/-10VDC controllers without building a custom circuit from a handful of op amps and dual voltage power supplies, or trying to deal with digital potentiometers for scaling control, etc.

Here's the kicker - It's as simple as an externally powered h-bridge with low pass filters!

Externally provided 10VDC can easily be flipped on an h-bridge (full reverse is -10VDC), and a proper LPF smooths out the signal to a nice analog voltage to avoid jittering from the PWM. Here is a before & after shot with the LPF on a single axis of my breadboard h-bridge. Check out the scope and you can see the PWM turned in to smooth analog voltage as it sweeps from full forward to full reverse at +/-5VDC.



I confirmed this works as intended across the control spectrum, and went to Fry's and picked up a ~$20 h-bridge shield. It was an Adafruit Motor/Servo Control Board v1 (compatible from OSEPP) that supports up to 4 motors, and provides a barrel jack and jumpers for external bridge power.


Here is the prototype shield with the filters in place (6 of them, 2 for each axis). This makes for a nice compact control unit. I'm saving as much space as I can on the prototype shield as I intend to add a scaling pot and a variable voltage regulator to tune the output voltage.

Thats great guy, but how are you going to control the PID loop?

Well, I rewrote the entirety of Arduino SMC firmware tools last night to suit my purposes!

I dried up all the repetitive code with Axis structs, clarified the comments, variable names, and command set. I also was able to remove all of the PWM & frequency scaling code, as the Adafruit library handles all of this. It now supports speed & directional commands for my h-bridge driver (with or without the LPF).

I'll be posting it as soon as I confirm that it is working as intended in case anyone else is interested in using this as a basis for their own designs. It should be fully compatible with SMC tools and therefore Xsim. I'm super excited to verify this with SMC tools once I get home. Shout out to the original SMC author!

So the updated code is available for Adafruit Motor Shield V2 and clones at https://github.com/SimulatorMotorController/SMC3/tree/clean if anyone is interested. It's the "clean" branch. It now works correctly with SMC3Utils, but still isn't 100% complete or tested (save/restore, instructions/second, etc) but it's less than 1/6th the size of the original SMC3.

So dealing with a few months recovering from a broken knee, I decided to take my prototype filter system to the next level. I based my design around the Adafruit Motor Shield V2, and include my filters and power management on the same PCB instead of stacking proto-shields. I never designed a surface mount PCB before, but with a few days of learning KiCad, I was up and running. The best part is that the original is open source, so recreating the system and adding additional components wasn't too hard, and the PWM chip supports up to 25MHz, so it's silky smooth, and handles the control signals well.

You can find the kicad designs for the shield at http://github.com/skiz/mcu

Here is a preview of the completed board. Everything is surface mount aside from the power jack and the terminals. I had it printed by https://oshpark.com and was very happy with the results.



I now have a lovely control board that will handle up to 4 axes with feedback, and provides very clean +/-10vdc analog output for each motor output.

Alright! Back to the frame build now that I can walk again....

I picked up more square tube steel and a new 14" cut off saw from harbor freight. Cutting those 2" steel tubes with a little 8" chop saw is a real pain in the (~$70). WOW! Having the right tool for the job really makes a difference in the speed and quality you can obtain. This is a must have if you are cutting anything more than 1" square tube. The time saved recutting, grinding, and fitting has paid for itself already.


The completed sub-frame (with adjustable feet)



Here is a mockup with the diamond plate aluminum (YES!), motors, and upper frame. It's starting to look really sexy, and I can't wait to get the frames blasted and coated.



I need to trim the diamond plate... Sure you can use a table saw (and cut yourself in half when it kicks back), or a jig saw (and have it look like a shark took a bite from it), or a grinder (and have the wheel explode like a grenade). I need to do a bit of trimming, and plan on making some mounting brackets and other things so.... I'm doing the only sane thing and buying a plasma cutter. Another toy for the shop? Certainly! However, I have a long list of uses and finally decided to pull the trigger on a mediocre one ($500). Making mounting brackets, fitting the plate to the upper frame, cutting aluminum chunks for the mill, and numerous other things are going to be so much easier with a plasma cutter on hand.

I'm super excited to get it going, and I'm already dreaming of building a CNC table for it after I finish this project.

How about using a 1280 controller? It's a good option for a 6DOF (I use one ) and can drive most motors and servos (AC or DC) , with the daughter card that comes with it - there is even a kickstarter where you can purchase one for a good price. https://www.kickstarter.com/project...-interface-for-motion-simulator-pla?ref=email - Although just looked at the kickstarter - unless something radical happens in the next 10 days, I cannot seeing it reaching its goal.

So quite a bit has been happening since my last post...

I received my plasma cutter, and it was a godsend. I've used it extensively to cut the aluminum diamond plate, cut through a driveshaft like butter, and even blew right through some 1/4" stainless (more on these in a minute).

So I figured out that given each of my 3 axises have a rotation point on each end, and not having a central point simply won't work. Any yaw at all and the platform just flops. Since I'm already going for a 3 axis system here, I've got some fabrication to do. I spent a few days trying to figure out the best way to do this, and decided that I obviously needed a u-joint and a spring of some kind, so I took a trip to the local salvage yard and found a short 2" diameter driveshaft with a u-joint and a suspension spring. My idea was vague, but theres no way I'm using Chinese metal drawer sliders on this thing to prevent the yaw.

My next goal was obviously to figure out the height, load, and range of motion. The heave is only about 3", so nothing major, and this spring should provide just enough suspension to be able to handle it without too much effort or binding.



I found a plate of 1/2" aluminum in my drawer that was the perfect size for making a mounting bracket. Drilled it, and tossed it on the 4 jaw lathe chuck to mill out the area where the driveshaft flange extends beyond the flat surface.


Lingering in my mind was how to prevent the yaw since I had a round driveshaft. I figured that if I got a piece of 2" ID DOM tube, then I could limit yaw by milling a slot through the driveshaft and fixing it with a pin which would only allow the driveshaft it to move vertically. This didn't pan out, as I couldn't buy DOM pieces from my local metal supplied, only full 20 foot lengths. So I picked up a 20" section of 1/4" tube steel, knowing damn well there is a weld seam inside that would prevent me from sliding them together without reaming the tube out.

Reaming the tube was a failure after about 4". A piece of HSS and a hammer was working well to peel the weld flush, but without creating some kind of jig, there was no way I could continue. Off to Harbor Freight...

After getting some new files, rotary burs, and a few other odds and ends thinking about how I was going to pull this off, I walked by the trailer hitches... *WHAT? IDEA!* Trailer hitches use a 2" insert, and after looking at a few receivers, I found an 18" receiver that was barely just over 2" internally and the seam was towards the corner. BINGO!

I brought the hitch home and my driveshaft fit, but a bit looser that I would have liked, and I decided to avoid milling a slot in the shaft as it would weaken it quite a bit, but I did find a 3/16(?) piece of square stock that fit perfectly in the corner along with the shaft inside the receiver. Yes! This will work to keep the shaft from turning!

I cut the receiver hitch with the plasma cutter (LOVE THIS THING!) down to size along with the driveshaft, and tack welded the square stock on to the driveshaft tube. It tightened up the fit a bit better, and it slides quite smoothly. I'm going to pick up a few more pieces, and tack them on to tighten it up even a bit more. It has about 1-2 degrees of yaw which I should be able to minimize with a few extra pieces. A 2" square tube may have worked better, but oh well.


Now to mount this contraption I need a base for it. I picked up a 16" square piece of 1/4" cold rolled steel while I was at the metal dealer for this purpose. A quick bit of measuring and welding, and I have a fully working central spring on a platform that I can bolt to my frame.

I welded up a few cross-members to hold the plate, and drilled all the holes. I need to pick up a pile of 3/8" hardware tomorrow, and I can get this thing bolted together and finish up the wiring for a test run.

Still planning on grinding, degreasing, sand blasting, and then finally powder coating it once it's functional.

So, what's been happening? I like your build.

I've got lots to share! I'll update tonight. I'm actually in the middle of racking my brain on the center post. It's binding and giving about 5 degrees of rotation. Even with more shims it looks like it's going to grind itself to a halt eventually, so I'm off to the hardware store and steel yard tomorrow. It needs bearings of some kind, and I'm too impatient to have something custom shipped out.

So as promised, here is an update..

It's been a while, and I finally got more than a few evenings to work on this project. I've made some great progress, and wanted to also share some findings. The current state isn't "complete" by any means, but I'm feeling very close to completing some major hurdles.



Motor Power
After trying to decide how to make my system somewhat modular for upgrades and fixes, I wanted to add some kind of plugs on my motors so they aren't hardwired to the control box. I picked up some of these, and wired them up. It was great. I am happy. I can now easily pull a motor unit from the system or unhook the control box at whim.



Motor Feedback
I decided that I really don't like using potentiometers due to the wear, inconsistencies, and general quality concerns. Not to mention over rotation, breaking, mounting, etc. Previously I was investigating other distance sensors (and I still have those huge pots!), but I'm going to save those for perhaps auto-zeroing or such if I ever have time.

I did originally decide to just accept cheesy rotary pots for the time being, and I designed and 3D printed some mounts for them. The mounts worked great, and it was a start, but I was really unhappy with the whole "just connect them to the shaft" part as I also needed to drill and tap the shafts to have something to hold on to, sync them, and hope I don't rip past their stops. Gears and chain reductions are a bit too complicated IMO for something that should just be a rotary encoder, and 10 turn pots are a pita to get back to the proper center. But whatever, they worked, and got the thing moving. Progress is progress...



After looking at really expensive encoders and trying to find something that would be acceptable budget wise, I ran across these magnetic rotary sensor breakout boards. After looking at the spec sheet, I immediately ordered 3 of them, and began to check them out. I2C isn't really made to go over distances more than a circuit board, but after some research and testing I found that with proper capacitance and terminating resistors it may be possible to go 6ft at a reasonable rate. I decided that these control signals also should run on CAT6 (for capacitance and shielding) and wired with ground pairs to avoid interference. Guess what? It worked like a charm, and I'm happy again.



I immediately started working on an enclosure and mounting system for these, and came up with what I consider a good solution. Simply epoxy the magnet to the shaft, and screw on the box. There is ZERO WEAR when things don't touch, so 1 less thing to worry about. Want to take it apart? No problem. The 14 bit position will easily find it's origin even if you spin the motors, remove the gearbox, etc. I'm happy with this. I also updated my custom SMC3 firmware to support these chips, and things are reading as expected. It's digital, it's air gapped, and they aren't going to break regardless of how the motor spins.



If you aren't familiar with the I2C bus, it allows you to interact with these chips by their address, and do things such as burn settings, set the zero position, set/retrieve registers (including position, etc). This really gives full control of the feedback system (set the system zero in EEPROM for example). I'm really enjoying the features, and for $15/ea and some 3D printed enclosures, they are even cheaper than a half decent potentiometer.

Center Shaft
So I found that my shaft really needs some work. It worked okay for a while when testing, but it now has too much play to be successful, and no amount of shims is going to be ideal with the twisting force being applied to it over time. It is causing binding, and grinding inside the tube, and I'm debating on how to replace it. Shelf slider bearings? Spline shaft? Custom linear bearings made from skateboard bearings? Slick tape? I guess we'll both have to wait for the next update to find out.

Next Steps
Tomorrow I'm heading to the hardware store and the steel yard to pick up what I hope is the rest of my required materials for the shaft and the upper platform. I intend on finding a solution to the shaft issues by tomorrow, and I'll update as soon as I come up with a working solution.

I need to replace the phono jacks with CAT6 plugs on the back of the control panel and connect the motor position sensors to the Arduino shileld. They are just breadboarded for the moment as you can see, but they will likely just plug in to a prototyping shield with molex connectors for now. This may actually be worth revising the control board and putting the arduino hardware all on the same custom board instead of shields, but all in good time. I will most likely CNC a new rear panel for it with engraved labels once I'm completely satisfied with it.

My seat needs a riser to sit on, so I will be whipping that up tomorrow also if I have time. I'm still waiting on finishing the diamond plates until I have all of the mechanical issues worked out. I also cut some motor mounts from 3/4" plywood for a temporary solution, but these will also be replaced by nice 1-1/2" steel riser mounts.

Cheers!

I built a spring assist for the heave on my rig as well, I used a set of 4 of these:

https://www.aliexpress.com/item/2-P...-d9e8-4837-bd79-d930131635cf&rmStoreLevelAB=3

https://www.aliexpress.com/item/LMK...-a242-4e05-9307-6f065ff8aed8&rmStoreLevelAB=3

They've worked great so far

I followed up on your build, and I really like your design. I was thinking about something very similar tonight, albeit slightly more simple currently, and will need to continue to work it out, or perhaps even base it on some of your awesome work. I guess I'm going to be warming up the mill soon, or waiting until after Xmas for my CNC to build a proper spring assist.




So, given I'm tearing this spring assist apart... what if I removed the upper section of the existing slip joint completely, and just attached the remaining assembly directly to the spring? Would I end up with an assisted version of one of these?


Welding/clamping the top plate + u joint directly to the spring, then welding a plate to the bottom of the spring with a square hole that fits the square lower section snugly seems like it would be an interim solution as I build a better trap. Resolves the binding issue and should control excessive yaw. It's a stiff car spring, so it should handle the load. Of course I'll have to tune the height of the spring and adjust the arm length to avoid excessive heave, but it may just work well enough to keep moving forward...

Leaving the spring free-floating might work, but I'd still worry about twisting, as it's not an inherently stable configuration. The CAD mock-up that you posted would definitely work, as long as you have proper bearing/bushing surfaces for the surfaces to slide over. I got around needing to tune the height of the spring by using a nut/bolt mechanism to move the rest height up or down.