During the lap that I had setup, there was one spot where it was possible to full throttle from low speed. I reach battery current limits within maybe half a second of actually full throttling and then current starts dropping. Then same for braking. But lets say I stay above this magic threshold for 2 sec per lap then. Rubber has a specific heat capacity of 1800 J/(K x kg) and we have 1.2kg of it, so we need 2200J to heat up one Celsius. 13kw goes into the motors, lets say 11kW comes out which would be fairly optimistic still probably. That can support 0.9G but we are getting 0.5, so 0.4 is waste. 11kW x 0.4 / 0.9 = 4800W, which would be 1200W per tire. For 2 sec that’s 2400J energy per tire.
2400J / 2200, there would be 1.1 degree temperature increase per lap due to chatgpt’s “saturation”. Given that maybe every 6-8 laps at most I need to take a short break, and I could build decent temps (even though not quite up to ideal yet) it doesn’t sound implausible. Then there’s the wind cooling the tire and the alu rim taking out heat too. I don’t think this would cause overheating for the track setup and riding I did on the last session.
That said, this explanation is still super weird and I need to look into it more to know if I should believe any of it or not.
I think I’ll throw the 22T on the rear motors again for now, wire up the rear sensors which I still didn’t do since last september, and then do a bit of drag racing. Which I’ll repeat on a lower teeth next week in Prague to have data for reliable direct comparison.
This may be right too. Since the Stooge setups run 12T for high RPM motors, there’s less torque on them. And they get the final torque via a smaller wheel. Whereas my 12T mod1.5 should see more torque on the teeth for the same amount of total acceleration due to the bigger wheel.
Short term I’ll try quality gears, but then long term maybe the move to mod2 would be warranted.
I think you have it there. Those soft tire compounds are raced on almost polished tracks. On regular asphalt with way more narrow eskate weight distribution the tires will definitely be deforming and absorbing up the Gs.
Stooge pinion gears are already pushing their limits as they are made for rc cars not eskate and definitely not go carts.
My guess is the majority of damage on that 12t happened on braking. Belt drives might be a less expensive consumable option. Also lower motor kv might increase efficiency.
Awesome work, i might be able to get to Prague and see it!
KZ and OK are raced on smooth outdoor asphalt tracks. My parking lot is not that smooth but it’s definitely fairly smooth asphalt too. I also have less weight on the tires than a kart does. And by high level karting standards these are actually considered medium compound.
Yeah, but what gives me hope that mod1.5 can last with good quality gears is that the BN mod1 gears handled 5.2 ratio on mod1 on a roughly 8.5" tire, on a similarly small gear. Well, for a long while. Then I broke half a teeth off a pinion, but I was racing that for a long time.
It’s from acceleration, I barely run any braking current on the rear and my rear ones are chewed up way more than the fronts. The fronts also look worn, but those could take a session or two more before they fail.
I don’t think it can handle the ratios and the torque I need. Plus, you want to be heard on the track in a positional race, makes it safer. Most of European racing is time attack, but I do hope to take this board eventually outside Europe too. And we do have a positional race at the end of the time attack too.
Yes, but reduce total power, I’d be worse off overall. It’s not inside the motor where I’m experiencing crazy losses unfortunately.
Nice! Hopefully see you there! I plan to be running this setup for the practice sessions before race day, at this point I am considering switching back to the Supernovas for the actual race though if my issues don’t get resolved by then. On the old setup I know I have good chances and I can swap back in about 2-3 hours with ratio change. About an hour if I don’t bother with gearing change, as I think 12T will still give me just about enough speed for the Pitland track.
Clearances worked out kinda weird on this chassis (as this is basically me doing unplanned stuff). So mounting is a little weird, only held on by two bolts for now. Anyways it will work. Just hoping the kingpin can actually pass through as clearances are gonna be a little tight there.
An M10 threaded rod screws into this and is tightened with a nut. On the other end I currently have a female M10 to male 1/4-20 adapter which seems to be made from aluminium. I am a little worried about the strength of this part. I am considering welding a 1/4 screw to the threaded rod. But then the weakest link will be the threaded insert inside the camera, which is not the best failure point…
Just checked and the angles look pretty good, I am excited to finally be able to record footage from the board! Definitely expect some videos from Prague!
So one of the motors also moved which contributed to the faster wear. I lost one motor screw and the other weren’t as tight as they should’ve been. That said, the other side (that didn’t move) still looks like it had a hell of a time
I’m putting on 22T on the rear right now to do some drag testing, and will bump down front amps to essentially zero for the drag test. Then quality 12T will go on in Prague to have comparable drag time data.
Also installing crossbars right now. Not like there’s any flex in the mounts, but doesn’t hurt to have them on anyways.
That is a thick plate, but I did some quick math because my theory is that the problem is with something in the gear drive, and I think there is around 800 N of force between the two gears, which is a whole lot. My math is that the whole setup has an inertia of about 130 kg, and at 0.5 g, that means 637 N of force at the ground. Assuming mostly the back wheels, that is 318.5 N per wheel at the ground. The wheel radius is ~0.127 m, so 40.4 N*m on the wheel. The wheel gear radius is ~0.05 m, so that would mean 808 N between the gears. That is the equivalent of a fairly heavy person standing on the end of your motor shaft.
Also, it’s probably easier to do from the motor torque. Just using my numbers there, though, that is about 8 N*m of motor torque, which seems reasonable, and the force on the gear will be F = 8 N*m/Rp, where Rp is the pinion radius. If you measured/knew the current, you could get the motor torque from that and the Kv, though.
It makes me think boards at this power level (especially with a future SSS upgrade) might need a 2 stage reduction, just so you can use larger diameter gears and reduce tooth forces. Or to at least find a source for 20mm wide gears
Hmm. So you think the issue is the backlash getting messed up? The motor mount plates are 8mm, by hand it was pretty hard to get them to move side to side pretty much at all. But now with the crossbar it’s pretty much fully solid.
To increase the rigidity of the motor shaft, I’d need to increase the 12mm slot that I put in there so that I can slide the motors off with the gear still attached to easily fit the outer diameter of the base of the gear - that’s about 16mm, so has to be maybe 18mm, 3mm material removed per side. Then I can move the mounts closer to the wheels and the pinions further down the motor shafts. And then I need to also cut off the part of the shaft that sticks out of the gears. It’s a bunch of work, too much to do before Prague. And I’d also need to change the mounting hardware towards the clamps if I take the plate off, because they are starting to round out, but I don’t have the right size on hand.
In the rear motors I’m running 200A each and they are (on paper) 175kv. Opening a random log and dividing ERPM by pole pairs and voltage and duty cycle I get between 155-165, lets say it’s 160. That should give 11.9Nm on each of my rear motor shafts on paper. Then I have 60A to each of my front motors to, on paper same KV. Also, back in the 2WD days of a couple raceboard iterations earlier, I did confirm that these motors do not saturate at 200A. I dragged 180A and 200A and 200A repeatedly had a 10% faster drag time.
Part of the problem I think is the width of the gears, but unfortunately laser cutting can’t get me much wider wheel spurs, and getting my mounting pattern into hobbed gears sounds like an expensive thing at the quantity I need it. I do have a contact who can make it happen though.
I did back in the days do logs where I was drawing the full 200A out of each motor. I don’t have any recent metr logs, I’ll make some after I wire up my sensors and put on the 22T set of gears that I have. I plan to confirm both that 200A is flowing, and do drag race acceleration test to 20mph / 32km/h to confirm the roughly half G reported on track. Then in Prague once I get a second set of good quality 12T from Tilo I’ll be repeating the same tests on those. That’ll make sure the data we are comparing is reasonably accurate.
I’m definitely considering doing multi stage for the SSS, or at the very least mod2. Multistage is a ton of work and money though so won’t happen for reachers.
I’m not sure exactly what is flexing; I just get the impression that something is flexing. Also, the worn gears seem pretty asymmetric. I’ve only ever used helical gear drives, so maybe that is normal for spur gears, but it seems to me like that could be because something is deflecting and causing the gears to not be exactly on the same plane. It could be the mounts, and that you’ll just have better luck with the new gears and with the crossbars installed. The point I was mainly trying to make was just that there is a lot of force, and even thick things might start flexing and giving you issues as you reduce the pinion size.
The walls aren’t straight on lasercut parts like it would be on a machined part. It’s usable, but it’s definitely not perfect. Lasercutting the gear teeth was one of my experiments actually. It reduces cost from maybe 60-70€ per gear to something like 30€. And Stooge’s spurs were out of stock at the time. So my thought process was that if I get half the life of a stooge gear out of it I’d still break even. It didn’t cause issues with the (actually) hardened 22T pinion. There, the wheel spur started to wear into the shape that it always should’ve been. This 12T gear definitely wasn’t properly hardened though. The wheel spur that was run on 12T has pretty much zero wear on the teeth.
The wiring inside honestly doesn’t look too bad either. But the space is pretty well filled up.
Need to reconfigure rear vesc for sensors and front vesc for 0 phase amps. Then I’ll do a drag time. The board is ready for Prague otherwise. In Prague I’ll start with putting on quality 12T in the rear and doing another drag run. Then I’ll up my front motor amps back to 60 and then it’s ready for track.
Happened right after I redid detection with sensors wired up and went for a ride. Don’t think it’s a coincidence. I guess the rumors about reacher sensors before V8 being iffy are true…
On the upside 47.5/17.5 truck angles feel nice at 120cm wheelbase. But if I am racing this board (technically I can get a replacement and put it in at Prague), then I’ll be swapping back to the supernovas. I realllly don’t wanna blow up another D100S. The chances of it happening yet again on the kart wheels would be higher.
It’s the combination of sensors plus 200A phase current. Even though it was sensored, the readings must have been shite because I still had some cogging on startup, just severly reduced compared to unsensored.
This is how they started with the new detection (empty uarts) and sensors at large throttle, set to RWD only. Low throttle there was some slight cogging still but was much much smoother.
When pulling throttle above sensor speeds, they were working perfectly like they should even for full throttle.
At the end you can also see zero reaction to my throttle, it was cooked by that time…
In fairness, I shouldn’t have kept pushing so hard after it was struggling that bad at the end. But on the other hand, decent hardware and software shouldn’t allow shoot throughs to happen in the first place no matter how much it is abused…
I did at least manage to get some acceleration data. I did a 0-20 under 3.30, the actual drawn phase current wasn’t confirmed though as they ESC decided to blow up before I could did anything with my metr module.
I reckon with zero stuttering that could’ve been about 2.7-3.0, so about 0.3G. Add front motors then it would be 0.4G. So the close to 0.5G dots on the previous log were a little overestimated perhaps, but reasonable accurate still for 22T.
So I do think something was messed on 12T. Maybe gearing maybe the tires just couldn’t take more and were heating from deformation like chatgpt thinks. I don’t know and won’t know for a while. One thing for sure though, today’s incident makes me not want to drag until everything is replaced with G300s and I configure HFI or perhaps run AS5047 encoders. Definitely not running sensors again with these motors.
