Big holes for big wires 
1 Like
Dug deeper into this for anyone who stumbles on this post. As @Wetmelon says, the power vias are really designed for direct wire contact. I did find that you could fit a #6 machine screw through them and with the addition of a captive nut and a ring terminal connector < 0.3" (typically red sleeve) you could definitely make a screw terminal. However, tough finding ring terminal contacts under 0.3" wide for anything higher than 14-16 AWG, meaning you are limited to 20-30 amps depending on your wire distance.
SO, we’re moving toward a direct wire approach as we need 60 amp and 10-12 AWG wire. @madcowswe not sure where to put issues/upvotes on hardware, but a slightly larger distance between the power terminals that would allow a 0.34" ~ 0.38" ring terminal for high gauge wire would make assembly and repairability of production units more convenient.
1 Like
Thanks for the input. These holes were originally meant for M3 ring terminals, but I didn’t actually validate which ones were readily available at the time.
ODrive v4 is designed to use up to 10AWG wires all around.
2 Likes
I have 8AWG fitted to mine, they are reduced down to around 10AWG at the hole and soldered directly on.
You could always manufacture your own from a copper rod the diameter of the pad and die the rod to bolt it directly to the board (using a copper earth nut) and bolt the cable onto the rod via a tapped hole in the top.
@Alexander_Jones - would be awesome if you could share your work on the enclosure once you’ve made some progress! I have a similar project on my roadmap and it would be cool to keep it all open and in the commons. We may have some slightly more stringent criteria for our enclosures (IP67 / waterproof likely) but intend to kick it back to the odrive community when we have a solution worth sharing. Warm regards - Matthew, Farm-ng
1 Like
@mabitter definitely. I just posted our load testing setup here for benchmarking our integrated passive heatsink design. Still generating the baseline data but once we have some confirmation on our heatsink we’ll start working on the enclosure. And I’ll definitely post on the forums and look forward to other folks creations.
As an aside, I’ve done some research on connectors etc and we’re moving in the direction of the e-sk8 community like Stormcore’s Enclosure Design as apposed to more traditional robotics packages. Stormcore has a similar form factor to ODrive and from this post we’ve identified that we really need to solder wires directly to the board and port outside of the enclosure. The Stormcore model is a great start and is getting towards IP65/66 depending on the sealing around your connectors.
At this point we’ve selected rubber grommets for porting cables, MR60 connectors for motors, XT60 connectors for power and 90 degree JST-XH-A1 connectors on the ODrive Board itself. That’s really what the ODrive board layout lends itself best to, but if you have a larger enclosure footprint you could certainly bring all that into a bunch on panel mount connectors. We have a small space that is enclosed within a rover chassis so we have double protection from the elements thus the IP64/5 target.
I’ll post a link here to the project post on our enclosure when I get it started, hopefully this week…
I am really hoping to get a ‘farm rated’ enclosure with anderson powerpoles ( https://powerwerx.com/anderson-printed-circuit-board-pcb-contacts )
If I am going to manufacture anything, it’s going to be an entirely new PCB with anderson powerpoles… the problem then being I’d have to go back to ODrive v3.4 or so that had schematics and PCB layout.
How big of an order would we have to collectively put in to get a custom version with support from Odriverobotics that can drive 2 motors and is easily adaptable to industrial applications?
It would be quite a task to add any industrial capability to the board. You can use the 3.5 docs since there isn’t any major changes in 3.6. At least in small quantities it would be quite expensive. I looked into this once, the increased copper thickness made it almost a hundred per bare pcb in small batches.
I selected ODrive as the main motor drive controller because the source code was available and it started as an open source project. This is going to be part of a $15,000 tractor, and it could cost $500 per board in small quantities and still be the most cost-effective solution I can find.
I have two choices… design my own board with no feedback from anyone else, or leverage the ODrive community and hopefully end up with something that we can collectively order in 10,000 unit quantities from reliably suppliers, and bring the cost down and the quality and reliability up at the same time. When that happens I can sell kits for the tractor so it’s actually affordable to build it yourself if you don’t want to pay the price I’m charging for a fully assembled field-ready machine.
I quickly moved away from odrive since, while its a good dev and hobby board, would need a lot of additional work for “real” industrial applications. Also at high powers the low voltage dc approach really makes for an inefficient design since the current defines the heat generation. Hence lower voltage = more current = more heat. The cable thickness was just too much.
Also having two drives on one board meant you had twice the heating on one board and I wanted separate drives. But that would’ve been too much extra work. The future v4 seems like a lot better solution, single drive that can be mounted on the motor itself.
And realistically, you’d still be looking at least a few hundred for a single “industrial” drive. If you want features and quality then that costs. There’s a reason even chinese industrial servo drives are a few hundred. The firmware also lacks the safety features you’d want.
@tmagik we’re developing an electric tractor with similar form factor as well and have met others on the forum with the same ideas. From our assessment the ODrive is definitely capable and depending on your application certainly worth it. Our experience has shown so far that derating hobby motors to 30% of their rated max power still provides plenty of torque with gearing and hub motors are also an option. I’m sure there are some application demands that make it more challenging as @Roiki1 has said, but for agricultural applications we still think the open source benefits and performance/cost ratio of ODrive are well worth it.
From my research and testing 12AWG wires soldered directly to the board mated to XT60 and MR60 AMASS connectors is the most robust option unless you want to go strait to 4mm banana plugs. You can find all these connectors on Amazon for reasonable prices. If you have power needs beyond 12AWG and 60A connectors then indeed ODrive may be a limitation.
Noted above, checkout the Stormcore’s Enclosure Design for inspiration, it’s a good starting point for outdoor use.
The reason I’m building my own tractor instead of buying one is the ‘commercial’ machine that cost over $100k new lacks basic safety features such power-on self test and on-farm diagnostic capability. I once had this machine start backing up while I was behind it due to a malfunctioning sensor, and now it’s got a leak in the electronically controlled hydraulic CVT that’s going to cost more to fix than what I’ve spent so far on building a robot.
So even if I have to spend thousands of dollars per board for farm-rated ODrive derivatives, it’s still worth it as I have full ability to run a diagnostic test and do root-cause failure analysis.
Also the first thing you notice about any farm equipment that’s been in the field for more than a few years is any safety shield that slows down repairs or diagnosis of problems ends up in the “I’ll put that back on later” pile.
The critical safety feature I need is the ability to understand exactly what my equipment is doing, which means source code for any software, and ability to have a replacement part fabricated 20 years from now. Generally the only stuff on the market I see that meet those requirements right now are mechanical relays, switches, and hydraulics.
Wire gauge ampacity is pretty dependent on length of wire. 12awg isn’t thick enough if we’re talking about several meters or more at 60A. You’d want 6 or 4 awg for that.
Also MR60 is rated at 30A. Did you even read the specs? Anderson has proper connectors that are actually rated that high.
No one’s saying the firmware isn’t capable. But it’s still an evolving project and lacks safeties to prevent runaway motors. Ive had that happen a couple of times. If you want to tale the firmware and build a new, more capable hardware then that’s perfectly doable. You just need to think things like signal isolation, IP rating, heat dissipation, component lifetimes, cables and connectors etc. It adds quite a bit complexity and cost.
I referred more to the drives than the entire tractor. Sometimes it’s just worth more in time to buy off the shelf parts than to try to make everything yourself. Or you can always hire people to design stuff for you for your specifications. I’m sure the odrive guys would do it for a fee.
I would very much like to get a quote from the odrive guys so I can build that into my production cost and fundraising plans. I do think there need to be several levels of E-stop functionality, and the best way I can imagine to get the required broad level of review I’d want to see for safety critical systems is have the entire mechanical design in Freecad, the PCB in Kicad, and the firmware available in Git.
I’m not going to get into the potential silicon bugs (yet), although my long-term plan is fabricate a RiscV processor at Skywater Technology foundry in Bloomington, MN. Then I can sponsor cross-functional teams of students at Iowa State or University of Minnesota to analyze the history of agricultural farm accidents and propose improvements to whatever level of design is most appropriate.
To your point about buying off the shelf parts… as soon as I saw @tlalexander’s brush motor patches I ordered 3 odrives… the only blocking point I see with moving forward with this hardware is whether or not Odriverobotics is going to be willing to design and deliver physical hardware for my tractor that includes the KiCad design files, or if it will be more cost effective to start from scratch and design my own controller with anderson connectors and conduction cooling options.
Let’s set up a meeting, we’re happy to quote. I suspect, however, that v4 with an industrial enclosure is going to go a long way to solving your issues.
Wtf. Name and shame. My day job is in the industry, and “unintended movement” is insanely unacceptable.
There’s really no need, Europe already knows how to make a safe machine - see ISO 25119. I can’t comment on whether or not any specific company is compliant, as I don’t know, but as a general rule, OEMs are less interested in functional safety and norms in the US, as the laws are less strict. There are companies that make Functionally Safe (high ISO 25119 AgPL or ISO 13849 PL) with “Category 2” or better electronics. That means “Test Channels” on inputs and outputs, lock-step dual core redundancy, etc, and they will guarantee replacement stock out 20 years, no problem.
1 Like
This was a Fendt transmission machine, with electronic park lock. I’d be happy to give the manufacturer or whatever European/US standards agencies an analysis at no charge if I can publicly talk about the failure mode. I rather doubt I’d be given access to the source code and schematics so I can identify the root cause, and I don’t think there’s much interest in diagnosing something on a 15 year old machine.
If I don’t have public documentation and a tool I can use to exercise the test channels 30 years later on the transmission and park interlocks, it’s not safe. We are still going to use it anyway until I can replace that equipment with machines with ODrive and other open-source components in the safety-critical path.
Does ISO 25119 have an operator safety override for conditions like this? https://www.youtube.com/watch?v=0ZjCOAnmUaQ
Sometimes safety means making the machine move because the consequence of not moving is far worse than moving in a potentially unsafe way.
I also pretty sure ANSI is not going to be happy with me if I include the PDF in odrive/docs, so we probably need some other source of truth for safety discussions. Even if the paywall at https://webstore.ansi.org/Standards/ISO/ISO251192010-1355953 was not an issue, is that URL going to be accessible 45 years from now? A lot of the equipment that we use for farming is older than I am, and is missing safety shields that get in the way of repairs.
Nothing really preventing the machine from operating in those conditions usually, but putting something in like an override switch that lets you drive without the arm rest down or without a driver in the seat would probably be considered “cycle beating” or something along those lines.
The performance levels / categories are really more about making sure your arm rest switch is redundant so you can detect a failure of the switch or a short circuit on any of the input pins, for example.
If I don’t have public documentation and a tool I can use to exercise the test channels 30 years later on the transmission and park interlocks, it’s not safe
I agree with you that open source is the way to go in general.
Well, the test channels are more an architecture thing, not like “click this button to test”. Category 2 is constantly monitoring itself: Command 1200mA but only getting 600mA? Throw a fault. Is there voltage at the pin when the output is off? Throw a fault. Etc.