Wood andMetal 2DFO Rig Version 2.0

I decided to do a new post for people in general watching for new posts here. This is a follow up on this thread:

https://www.xsimulator.net/community/threads/wood-and-metal-2dof-rig.17109/#post-249022

I see it has been more than a year since I posted. I wonder how many people get to a certain point and give up. I suspect a majority or members. For me I started over for version 2.0. First the obligatoy video of the second generation rig in motion. This is AC at Maple Valley which as lots of hills and banks. This time a 1970 Lotus manual:

https://youtu.be/LC-ZBfspqeo

On the moving carriage I have added a steel plate between the U joint and the wood centerpiece. This allows the bolts to be further apart on the center frame. The base is different and relies on steel braces rather than through bolts. This allows some adjustment to make the frame level overall and adds strength.

The main reason I started over was that the motors I selected were pretty much destroyed. I knew from the beginning that they were marginal but I didn’t realize that the strength of the gearbox was going to be the limiting factor. I was concerned about the 8nM of torque and that was barely enough. Their biggest advantage was that they were only $80.

I spent several months investigating lead screws as an alternative to motor levers. I decided that for a lot less time and money I could build a 2D lever rig with better motors and enjoy it. I did build a test leadscrew axis which I intend to use as a traction loss axis in version 2.1.

In my motor search at that time I was intrigued by the RLV40 gearbox. This features a 14mm keyed hollow shaft and a variety of mounting options. The motor itself is removable and can be repurposed for a future 4 axis lead screw rig.

In September of 2023 I bought two of these motors from Amazon:

https://a.co/d/0ekaABoE

Ironically these motors are the same price as my initial motors today. These motors were risky since there is no stated torque or data on the motors and no Amazon reviews. 120W seems low but that was significantly more than my previous motors. The 24 RPM wasn’t ideal but my gut and experience was saying that the motor just doesn’t move that fast. Of course that also implies twice the torque since my other motors were 52 RPM.

I really wanted to know what the torque of this motor gearbox combo was. My guess based on the 120W was that a 52RPM motor would generate 12nM of torque (based on the 80 watt motor generating 8nM of torque). For 24RPM the torque should be doubled to 24nM.

I measured the torque using a luggage scale. The torque was pulled through a 3 inch lever arm and measured 80 lbs at 80% power (PWM = 200). The 100lb scale overflowed at 255. So the torque is 25 foot pounds or 34nM. This is born out by the rig working well for 6 months now.

I have always been concerned that the 2D design would slam the motor shaft via the lever arm. The rotational slamming is bad enough but the perpendicular force on the motor shaft is significant. These forces tend to move the shaft out of alignment with the gears.

After a few fun hours of taking this beating the motor and gears loosened up causing slack in the driving chain causing even more slamming of the gearbox. My fault analysis showed that the shaft bearing in this type pf motor was damaged causing some misalignment. The combination of these factors eventually led to broken gear teeth.

It may be overkill but the new design has sleeve bearings on either side of the lever attachment to the shaft.

Here is the entire drive unit:

On the bench from with both drives:


Top view on the rig:


Hall Effect position sensor in 3D printed fixture:


This fixture holds the sensor exactly in the middle of the shaft. Another fixture is used to exactly center the magnet. If there is misalignment the Hall Effect response is not linear. I modified the code so that the 512 position on the sensor is done in software with a #define offset. Precise rotation of the hall sensor is difficult and imprecise.

NOTE: This hall sensor uses I2C by default. This is a problem because the access time is an order of magnitude longer than accessing an analog input. Removing the resister puts the device in PWM output mode. This PWM output can be read by the analog input after passing through a simple low pass filter (one resister one capacitor).

Early in the process I decided to go with hardware limit switches. By hardware I mean that the software doesn’t know that the motor is at the limit. The diode prevents driving into the limit but allows driving away from the limit. This is a bit tedious to implement and not intellectually satisfying but continues to prevent damage to components and stress on the motor. It is possible to tune aggressively without worrying about overshooting into a hard stop. There is a hard stop 2-3mm beyond the limit to prevent momentum causing damage to the switch. The link below is to good explanation for how this works:

https://arduinodiy.wordpress.com/2019/03/24/using-limitswitches-with-diodes/

NOTE: There is a voltage drop across the diode which means that moving off the limit requires a higher than expected PWM Max. I spent way too much timing trying to fix what turned out to be a non-problem.

The 2x4 bearing blocks are cheap. The 2x4 tend to be what wears so there is some wiggle in the bearing block due to the soft wood. These are 28mm long so the load is spread out.

BOM:

Amazon motors: https://a.co/d/0ekaABoE

Hall Effect Sensors (same as version 1.0): https://a.co/d/0gAJEZBA

Motor Drivers (same as version 1.0): https://a.co/d/07fYOi6o

Long 14mm shafts from McMaster-Carr: https://www.mcmaster.com/catalog/130/1365/1439K36

Sleeve bearings also McMaster-Carr: https://www.mcmaster.com/catalog/130/1426/6658K198

Oversized 6mm keys (6.03) from McMaster-Carr: https://www.mcmaster.com/catalog/98870A351

I started attaching the lever arm with this but it has too much slack: https://www.mcmaster.com/catalog/130/1326/9859T716

For this design I built in exterior sleeve bearings on both sides of the lever arm. In theory these bearings get most of the load torque and the shaft is supported on both ends.

I also decided that to reduce risk I would drive from the back since that was the more common mechanism around this forum. It was an abysmal failure. I knew my rig was not well suited for this as it is much sitffer in the front. I decided to go back to a front drive like the original. This has worked out very well. Now I don’t worry about it because it has been working flawlessly for such a long time.

An inherent problem with the lever drive is the “slack” in the drive. This is due things like the Heim/lever mechanism having flex in order to work and various other sources that might be correctable. The first is the connection between the motor and the shaft. This is a hollow shaft motor but it is keyed with a 6mm key. I noted that there was a tiny bit of play between the key, shaft and the motor. This play led to more “slack” in the drive. The oversize keys from McMaster Carr (6.03mm to fit in a 6mm key) greatly reduced that problem. At first I was concerned about disassembly but things pull apart easily with a pry bar or a slide hammer. OK, maybe not easily but also not a big issue with the right tools.

A bigger concern was the play between the lever arm and the shaft. I used a keyed bushing but even with the oversized keys there was a lot of slack. I added a clevis pin through a hole which helped but was not ideal. The up side was that it was easy to try different lever arm lengths.

So I decided to impose on a friend to mill flats on both sides of the shaft so I could assemble it this way:


This worked fine and there is ZERO play between the shaft and the lever. The problem is that the milling took a long time. In retrospect it would have been less time to weld the lever arm to the shaft. Welding should work if you are sure about the lever length.

So for now am just having a good time using this daily without having to fiddle with it. I have almost finished with the lead screw drift axis. I also have a design for a low cost SFX-150 style actuator which will beused on my next rig. The 2D design obviously has limitations and I think I can make a unique at a much lower cost than the existing options.