Lots of progress this weekend, with lots of lessons learned! We manufactured the base and the yaw yoke axis bearing, and mounted the yaw axis on the floor in the basement. We also installed the yaw yoke drive system to test backlash and rigidity.
Yake yoke drive shaft, below:
The drive shaft itself was fairly easy to cut, taking about two hours on the first attempt. The shaft is 1.5″ in diameter at the base and steps down to a 1-3/8″ – 12 thread, then to a 30mm diameter step for the slipring, and finally to a 1-1/8″ diameter for the drive pulley.
The drive shaft attaches on its base to a half inch steel plate, which then bolts on the the frame (the base, in the case of the yaw yoke).
This is the mounted base with the yaw yoke drive shaft attached. The mounting plate is actually welded to the drive shaft. That build process ended up taking a long time, and I think that we’ll need to revisit how to attach the shaft to the mounting plate.
Here is the yaw yoke mounted to the base, with the drive assembly attached. The bearings are almost rigid enough, but not quite. We have some ideas about how to fix that.
The frame itself is also really rigid. There is no noticeable deflection on the base assembly, and deflection on the yaw yoke is minimal.
As always, Hobbes was around to help.
Now on to the biggest problem: the drive chain stretches! A lot. I expected some stretch, of course, but not nearly as much as we are getting. You can grab the yoke arm and push, moving the assembly 5 to 10 degrees with very little effort. Turns out that this is just how chains tend to behave. It’s looking like we are going to have to do a significant redesign of the drive system.
That’s a little disappointing, of course. I at least was feeling fairly confident that we were on the right track with this design. Modifying the drive assembly is going to involve either rebuilding new yoke arms or cutting them apart . Either way, it’s going to take a lot of time… [sigh]
We’ve made a lot of progress on the design for the three degree-of-freedom simulator, and are hoping to order parts to start the build early next week.
This new design allows for continuous rotation on every axis, although it will require a lot of space to do so.
Rigidity might be tricky on this one. After running the numbers on some worst-case scenarios, we think that the frame itself will deflect less than 0.16″ on the inner axis. The axis bearings may cause more than that, though. And it’s difficult to know how the new drive train is going to perform in terms of backlash and rigidity.
But we’ll find out! The build process will start next week.
I get asked this all the time, and it’s an understandable question. Servo motors are the most expensive part of the machine, and you need a one for every axis. I’ve seen forums where people talk about disassembling treadmills to repurpose the high power DC motors. So… let’s look back and talk about the motors we’ve considered and used on this project! I’ll talk a little about gearboxes and other mechanical drive parts.
Clearpath Servo Motors
The first candidate was the CPM-SDSK-3441P-RLN from Teknic. Motor product link here.
Clearpath motors are great, and I can’t recommend them enough! They aren’t the cheapest option out there, but they have a lot going for them:
Their ratings are actually accurate, which is something that can’t be said of motors from many other manufacturers.
Documentation is great.
Price is pretty good compared to equivalents from other brands.
Construction is solid. These are well-built motors.
Great support. It’s easy to get on the phone with an engineer.
Finally, these motors are really easy to use, with built-in drivers that are hardened and ready to go.
Now, in retrospect, I ordered this motor too soon. It was actually underpowered, so I returned it. Within the last year, Teknic actually released a new line of motors that ramps up the maximum power significantly. Their higher power motors are more expensive, but are still a great deal compared to a lot of other options. There are a lot of different models available. If you can find one that fits your budget and power requirements, you can’t go wrong by using a Clearpath motor from Teknic.
We didn’t end up using Teknic on the first-rev. prototype because there were cheaper options, and at the time they didn’t have anything with high enough power output. Now they do, but they still aren’t the lowest cost.
AC Servo Motors from China
I don’t know who makes these motors, but you can get them from a variety of places. We experimented with a 1.8 kW AC servo, model number 110ST-M06030, shown below:
This is a solid motor, and the price is unreal. Seriously, you can get one of these, including the driver and controller, for $322 (USD). We got them from FastToBuy.com. They ship from China, so expect to pay a ridiculous amount for shipping. And FastToBuy is an intermediary, so I don’t know who actually makes them.
Downsides? Documentation is poor and inaccessible – it’s just a bad Chinese-English translation. And expect to know something about PI gain tuning to get the servo operating like it should in your system. If you don’t know how to do that, don’t let it scare you – you can figure it out. But it isn’t going to be as simple as getting a Teknic motor set up.
Also, the driver needs 240V single phase, or 3-phase power (3-phase is better if it’s available. Less ripple on the internal rectified DC voltage used to drive the system). So you can’t just plug your system into a normal single-phase wall outlet.
But it’s hard to argue about the amount of power these offer at the price! These are actually significantly overpowered, but that’s OK. I can’t say a lot about longevity, since we haven’t run them for a long time. So far they are doing well.
The linear actuators on the Version One 2-DOF prototype actually use an even larger 2.3 kW servo motor from the same manufacturer and supplier: the hefty 130ST-M15015. It’s only $395 off the shelf, some $70 more expensive than its 1.8 kW counterpart. We thought, “Hey, at that price, we may as well upgrade!” These servos don’t even warm up on the 2-DOF prototype. They run cold. We haven’t even begun to make them work hard.
Caveat: We haven’t run these on a dynamometer, and can’t say that they are really 1.8 kW or 2.3 kW. If I had to guess, I’d say “no, they aren’t that powerful.” They are clearly more than enough for this application, though.
On the current linear actuators, we are using the 130ST-M15015 servo motor with a 10:1 NMRV063 worm gearbox, also from FastToBuy. Had to get the speed down and the torque up! The gearbox goes to a ~3″ diameter timing belt pulley, which moves the assembly. It isn’t clear on FastToBuy’s website, but there are actually a lot of different gearbox sizes and ratios available, and the factory (whoever/wherever it is) can do custom shaft sizes, input and output flanges, etc. FastToBuy sells smaller and larger gearboxes of the same type as well. They can work with you to get the one you need for your application.
Interestingly, this exact gearbox model, possibly from the same manufacturer, is also available from Automation Direct. I know it’s the same because I just bought one from them. 🙂 You’ll pay more from Automation Direct, but it ships from the US, and it actually has specs available on the website. You know, output torque, max speed, input torque, etc., etc. Important stuff like that. It’s a lot easier to work with, but there are fewer options when it comes to input and output shafts, flanges, and so on.
Pulleys and Timing Belts
We are using 1.5″ H-series timing belts and pulleys. You can get them all over the place (McMaster, for example). But you can get them for cheap from shop.polybelt.com. They have lots of pulley and belt sizes, belt types, and so on. It’ll be way cheaper there than from McMaster.
Motors for the Next Revision
We’ll most likely be switching motors for the next revision. Why?
In spite of the low cost, working with companies in China can sometimes a pain. :-/ Communication is difficult, supply-chain isn’t transparent, shipping is super expensive and takes forever, specs and documentation leave a lot to be desired, quality control is opaque, bla bla bla.
We want to be able to run the 2-DOF simulator from single phase 120V AC. It would be nice to just plug it into a wall, you know?
We think that with the motors we are considering, we can actually reduce the total cost per axis. Possibly. It’s hard to say for sure until we have the solution completely figured out.
Here’s what we are looking at for the next version:
This is a 3/4 HP, 90V brushed DC motor from Automation Direct (ironically… made in China, but at least Automation Direct handles the overseas issues). The math says that these motors should be fine. The gearbox is a NMRV model basically just like the ones we currently use, but the ratio is much higher, and it’s available from Automation Direct, sized specifically for the DC motor.
The motor doesn’t have an encoder, and of course it doesn’t include a driver or controller. That will drive up the price, so it might end up being a little more expensive than what we are currently using. We’ll find out! Encoder and servo drive are on the way, and I already know how we are going to get the encoder set up.