I was able to modify my Phub-188 for more speed at the expense of torque by changing the termination from wye to delta, small writeup here: Redstar | DIY onewheel
1 Like
Hi Jopj,
thanks for the answer, I have set the madwick at 0.1 and the calibration was done with a buble level.
the angle with roll and pitch is fine but the angle with yaw is impossible to set to 0.
my actual mpu6050 is always drift on the yaw, that why I have buy a mu9250 from sparkfun which will be more accurate with the magnetometer . I am waiting all the stuff because the postmen dont work because coronavirus. I am thinking of buying a tesla gotway board EUC (127 € https://fr.aliexpress.com/item/32879870523.html?spm=a2g0o.productlist.0.0.6f7d44a39PGeAu&algo_pvid=5726bbbe-5f20-48cf-9fca-6c30ae6299d2&algo_expid=5726bbbe-5f20-48cf-9fca-6c30ae6299d2-0&btsid=0b0a050b15847432657745216ea501&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_) and buy some other cells and a new bms to go 67.2 volts ( 16s ). the big difficulty will be desoldering the mpu to change from vertical to horizontal position . more powerful and more reliable maybe.
The mpu9250 should work good, but the 6050 is actually the superior IMU for these applications. It’s more tolerant to vibration, has less noise and we don’t use the magnetometer. Yours must be broken, how much is the yaw in degrees/second without calibration? I tend to buy both of these in some numbers since they cost 2€ or so, and are quite useful little things.
2 Likes
I also started building a onewheel.
Im worried that the rails can still turn a little bit individually because of play in the square hole of the steel adapter. Did you also have that problem and did it cause any trouble later on?
3 Likes
If the axle is the only thing keeping the rails fixed it’ll definitely flex, you’ll put quite some torque while maneuvering around on it. The shaft on the Phub is also very soft metal, I managed to noticeably squish it while tapping it with a hammer to extract the stator out.
Supporting the rails from multiple point worked out good for me
You could alternately use strong enough footpad material to prevent the rails from twisting. I went with both due to paranoia, the result is very rigid but heavy.
3 Likes
man a F1 tire would make a good Funwheel
F1wheel great name
2 Likes
16s, wow! What controller will you use with it?
1 Like
The a200s, it should be good up to 18s but i only have 48 cells, so 16s it will be.
2 Likes
That’s great, you can neatly sidestep the problems of the Phub just by increasing voltage, no need to open things up 
Your frame looks sturdy, no way that’ll flex too much.
1 Like
Thanks Jopj!
Today i added some more stuff to the build and i realzed that im a fucking idiot. Prior to welding i forgot to pump up the tire. Now if i pump it over 7.5psi it rubs on the welded crossbar 
Other than that, i connected a 14s battery and tried to get the balance app running. I watched the video from @NuRxG and tried to configure the balance app. Its is reacting to the inputs from the imu but if i lean in a bit more it always changes to a dead state, no idea why.
2 Likes
Those are some sweet sensors haha. Maybe if you had a integrated screen you could make a rideable MAME arcade.
2 Likes
Dead state is caused by your duty cycle exceeding the Duty Cycle Fault parameter. I’m not 100% sure on how duty cycle is calculated, but it seems to account for both speed and torque, so if youre still at a low speed, you’re likely going over your current setting. IMO you should set the motor current to the max allowed, and use the battery current to limit your power.
However it is worth noting that there are multiple redundant safeties. It looks like you have foot switches, which many people consider to be enough safety to prevent a runaway board. So if you wish you can set your duty cycle cut off to anything over .95 as the vesc never goes above .95 duty cycle, or add a long delay which will give you some time to over power the wheel and regain your balance before a cutoff. Really up to you how you want to manage your safety and liability of a runaway board.
1 Like
Thanks, i put it on 1.00 and could ride around a little bit. There was still some trouble with the external IMU that crashed all the time and i had to cut the power and reconnect. This happend when there where a lot of amps flowing. Probably EMI issues because of the long cables and proximity to phase/battery wires.
I also built the 16s battery and mostly finished the case but the project is now put on hold indefinitly (see last picture)
2 Likes
That sucks man, hope you are ok.
Did you bolt it straight to the frame on the bottom, could’ve shorted the FETS with vibration as there’s no heatsink. Also, where are your two other battery leads? I’m guessing you took them off for packaging reasons but that could have in theory affected things even though it shouldn’t.
Also, it looks like its a V2, some batches of the V2 have misprints on the silkscreen for the halls and the COM port, so that could be an issue.
Check your phase wires as I don’t see any grommets on the cut edges of holes
I have an unused AS200 v2.1 that fixed these issues, also with better caps etc that I may be selling soon since it doesn’t fit my project
Yes it was bolted to the frame bottom but with an aluminium bar and silicone pad in between. One of the high side FETs probably punched trough the pad to the aluminim and created a short. All the a200s screw holes are grounded so automaticly all the metal parts of the onewheel too
1 Like
haha electronic fireworks, at least you know what not to do lol
Yeah it really sounded like firworks haha. Too bad the logic board has taken some damage too.
Already salvaged the good FETs, shunts and caps.
Wow that really did fry! I was gonna say maybe you could try getting a new power board if that’s the thing that fried and the logic was good but that’s gone too!
Heres my 2.1 to compare
1 Like
Ah yes it has the 12V dc-dc “black box” not that stupid tiny TI board.
Edit:
And larger bulk caps too.