What is total cost per module with motors, gears, and controllers. Like, “All in”
From their linked BOM, $482.53
However, looking through said BOM, I didn’t notice any sheet/plate listed. Am I just blind, or is it not on there?
Someone is on the ball!!! $482.53 without any of the waterjet cut Aluminum parts since a sponsor provides them at no charge. Since we did not have to pay for them we did not put them in our test unit BOM. But we did have to account for them on our 2020 robot cost accounting worksheet and added $97 per module for them. Of course, 2 x NEO’s and 2 x SparkMax’s were $230 of the total. FWIW, we are really happy with the NEO motors - ridiculously powerful for their size and weight.
$97 per module for aluminum seems pretty pricey to me. For reference, the last swerve I designed, using 3/16 plate instead of 1/4 (which in my experience is more expensive), used about $15 worth of aluminum plate per module.
Wait so you’re saying each module, motors, wheels, controllers, etc. and everything, would end around $500? That’s super good compared to other similar modules. What makes this one so cheap?
My design is just a ripoff of Nick’s Summer Swerve, but here’s my BOM.
(and by ripoff I mean it’s exactly the same design, but I made all the files from scratch)
BOM.xlsx (18.8 KB)
Two of our goals were ‘low’ cost and simplicity. So instead of an absolute encoder and associated additional gearing, we used a hall effect sensor and magnet on the platten and both encoders in the NEO’s. Also, there is no milling on the wheel axles due to some clever tricks with waterjet cut parts, so we avoided that cost and complexity also. Another thing is that we were able to reduce the weight to about 3.9 lbs each (w/o motors and controllers) - probably one of the lightest of this type. You might even do a little better on cost with alternate sourcing - McMaster Carr is great but not low cost.
Sir, certainly not a ‘rip-off’ since I had never seen or heard of that module until I checked out your link. It’s a slik design and slightly lighter than ours. But there are also significant differences. We spend 8 months carefully optimising and testing our design to meet the goals we had put down. I believe it is unique in several ways and hopefully will suit teams with limited resources. Cheers.
Oh I’m so sorry about that wording. I meant to say that MY design is just a copy, didn’t want to imply that yours was.
How well did using the hall effect sensors/magnet work out in practice?
I’ve been messing around with designing a low cost/simple swerve drive myself, and that method seems like it could be a really nice solution.
When you say hall effect sensor/magnet, can you describe how you implemented it’s use?
The sensor was fixed with two screws on the underside of the swerve mount plate with it’s sensing area facing a lobe on the rotating plate (underside disc), The lobe has a small through hole into which a Neodymium magnet is Loctited. Note that we only take one rising edge trigger at the beginning of the match for zero and then use the NEO encoder for the remainder of the match. During this one zero operation, we rotate about 20% of max speed for consistent results. See our BOM for part numbers.
a_cool_username: no problem. I should perhaps appologise for an unnecessary defense. All’s good!
GP rules! Cheers.
We actually simplified the sensor mounting design on our comp robot. Instead of a small plate, two posts and screws to hold the sensor its now mounted directly to the swerve mount plate and the lobe on the underside disc has a slight relief (done with a woodworking router) to clear the screw heads that hold the sensor. Simpler, stronger and less vulnerable.
How much time does this zeroing operation take?
It’s dependant upon where the wheels are when placed down on the field. But worst case is < 1 sec.
If the drive team set the wheels when placing the bot, it would be < 0.25 sec.
How accurate is the zeroing method using the hall sensors? Very innovative method.
We really didn’t take it as far as we wanted to. We had a serious RobRio issue that NI are currently investigating and so did not change the code from the test base. We have no hard measurements right now but I’d guess +/- one or two degrees (probably similar to the accuracy that a bot would be placed parallel to the wall). We may look at alternative sensors such as optical - this should allow us to rotate as fast as we wish to zero repeatably. FYI, we found that with our gearing, a wheel direction speed of about 50% power was optimal on the test bot. With the extra weight of the comp. bot, that figure may be adjusted. Cheers.