The Onewheel GTXS (Half scale GTS)
-
@lia said in The Onewheel GTXS (Half scale GTS):
@NotSure I'd love to sell these or partner with FM to bring these to market.
They would if they know what's good for 'em!
-
Upppppdates!
I haven't updated this in forever oh no... okay sparknotes version here we go!PCBs came and look fab! Like better than I thought. At a glance they look like unpopulated GT controllers till you realise they're tiny!
But I had to make some minor changes so this one has been put in the prototype box. I have a hard time throwing things out and these are too pretty to go in the bin. Maybe a future art project!The PCB use the Cheap Focer schematic but modified so it's appropriate for this build. As such I can do tests on changes with the Cheap Focer I yoinked from a VESC board that I converted back to stock.
I wanted to use this tiny BT module so I wired it up, flashed it and was able to verify this does actually work. I tried a few others and they didn't work so this was a relief since space is a premium and I want to keep it on the mainboard if I can.I'll be using ReFloat with this setup as it has decent LED control baked in. The old float package had an issue where activating LED's triggered an error. The error being FAULT_CODE_DRV .
I traced this back to this line in the float app that set the CPU pin B7 to the LED data pin (servo) but if you check the schematic for most VESCs that pin is fault on the mosfet driver. B6 is servo which it should be assigned too instead.
So you need to reassign this in the code or use ReFloat instead of the float package as it's assigned correctly there with the option to adjust it.This solves me needing to learn to compile a custom package thankfully...
... yeah nah I still need to 💀
I need to make a custom LED profile for the headlights as I plan on adding configurable main beams. This means I need to make changes to the package and recompile then flash it to my VESC. I hate... and I mean haaaaaate software so this pains me to have to do.
After some utter BS figuring out all the nonsense required to do this I got Ubuntu18 LTS to run under WSL on Windows and compiled my own changes to ReFloat. I will get more into this when the time comes.
If you need to compile this yourself you need to be on Ubuntu18 LTS or lower. Some of the dependencies needed are not available on later versions and refuse to install so don't waste 4 hours like I did trying.
Back to something I'm more comfortable with. The real world... not stinky software!
(those that do it for a living are gods idk how you cope 🙏)PCBA is gonna cost heaps so I'll be soldering this up myself. As such I got myself the right gear for it. A big hotplate and digital microscope :3
I accidentally cooked a BT module when testing so I used it to check if my custom footprint for that module was accurate and if I could solder these components. Thankfully it looks like I'll be good!Some amazing peeps on insta have offered to send me some of their dead focers for doing prototyping on so I'm waiting on those to arrive. In the meantime I've been test fitting some components not native to the focers like a USB C port and the side mounted XT30.
The USB port took some doing as the available footprints for them on EasyEDA do not match what USB ports I can order. THankfully I managed to modify one to fit 😅Sadly the motor I ordered wasn't great but I found an even better one and had to strip it down to make custom plates to help mount the hub plates too. Got to crack out my 6ton press to get the bearings off 😊
Modelled loooads of the board over a weekend and got most of the big bits done. Renders as always look utterly fab!
Controller housing printed perfectly and the controller fits!
As a bonus I cut my own phase brackets and soldered them to check the fit and how hard it might be to do.
I the managed to work on the battery box. Originally I spec'd it for a 6s1p 18650 pack but after some deep pondering of my orb (fusion) I realised I could barely... and I mean barely squeeze another cell and make it 7s1p!
It
Is
Tight!
Those are my test cells. Some dead ones recovered from an old ebike battery that I use for validating CAD. I have ordered some Molicel P28A cells as those haveBut with the lid and some rails printed up it all goes together nicely.
And as teased in the pic I managed to replicate the GT Performance Treaded tire 🥰
I'm going to print a master of the tire in 2 halves like on the XS. Glue them together, sand and prime so they look smooth then add some sidewall detailing like the Onewheel logo and tire specs.
22 hours later and I have me a mean looking tire to go with the little guy.
I think that covers everything up until now 🥰
Apologies for not updating this all here as often. I have been teasing bits over on my insta so feel free to follow and get those as I post them ^-^ otherwise the more detailed process shall remain here!
-
@lia said in The Onewheel GTXS (Half scale GTS):
22 hours later and I have me a mean looking tire to go with the little guy.
Brava! Looks awesome. How soon till I can buy one on Amazon?
XR's got what plants crave!
-
@notsure If only 😭
-
@lia Amazing work!
-
Looks so good!!
Onewheel needs a super brain like yours. -
Thanks @Lemur and @Wheelwizard 🥰
If they ever open up a department for "toys" I'm their gal 🙏 -
Nearly forgot to make the maghandle. That'd have been tragic.
It's surprisingly fun to play with and snaps back nicely with 2 tiny 3mm magnets tucked away in the axle block and handle 🥰
Probably gonna leave this as plastic because it looks like hell to machine which means expensive.All the current bits laid bare and it's very familiar looking. Like you'd just gutted an actual GT.
A lot less physically going into this than the XS but don't be fooled. The lack of components doesn't mean this has been any easier. That custom controller still needs populating and testing which has taken the majority of my R&D time.Speaking of testing. 42km/h freewheel speed ain't bad especially with just plastic test pieces and an incomplete axle 😅
With 35 amps max current available and just shy of 30v this little monster is not a toy. Still cute, but a right ankle biter if it ever ghosted.
Finally... I can safely say I have nearly completed the CAD.
I wanna go with recurve rails but I have no dimensions for the angles and where they are in the rail so I might ask (beg) FM for a rail. Also wanna give that rear pad the high kick version for extra flair. I am after all trying to make a tiny GTS so why not give it some of the cooler accessories.
I was considering FST but that's a lot more faff and I'd have to get even more creative with a sensor 😩 It's huge maybe especially since it's not like your hands could take advantage that well. -
Hiiiii guys.
Guess who forgor to update the thread again 🥺 (It's me, hi, I'm the problem, it's me)
I'm really bad at keeping this updated. At least you all get a condensed version of events this way.Let's hop in the time machine and zip back to December last year~
@cheppy44 was over here and we decided to take a crack at the custom VESC. We prepped another cheap foccer 2 as the transplant patient provided by Quinn 🙏 (thank you again)
Taking the parts off was made easier by the big hot plate I got but by the end of it this PCB was cooked.
We got everything moved over and just needed a few special components. Compared to a can of Pepsi it's absolutely tiny 🥰
A broken pint was used as a donor for the mosfets. When designing the board I thought it'd be funny to use actual FM parts in the build where I could.
Other than a bodge wire on the IMU which took way too many attempts than I'm willing to admit the PCB was perfect.
But no like seriously we spent days trying to get the IMU to work 😭My resistor sample book somehow managed to package 51k resistors where 1k ones should have been so that's what the resistor is off to the side. I didn't have any others spare at the time to swap >.> But once it was in the LEDs worked fine too.
Head and tail lights needed testing so during Christmas we got it to demo itself being festive 🥰
The white LEDs are not lit in these with how I set it up. The first LED in each headlight are sent to a ws2811 chip that on the blue pin goes to all the white LEDs. With this I can control the high beam by setting that LED in the chain to anything including a blue signal and it will light up. This way when front and rear swap so will the main beam without any additional code since red has no blue but white does. In code I can then pull the main beam LEDs to 0 if I want to toggle them off which I will do with a custom version of refloat I'll get to once this is mostly together.At some point in adding the daughterboard a single lil solderball managed to kill the main STM chip. It booted up and then upon trying to hold the motor sent like 29v to the chip down one of the sense pins. Oops. Cooked it :( After some tears I did manage to pull off the old one and put a fresh chip there then used an STLink to flash it.
With a bit more faffing we had it running. There was another mistake on the daughterboard where the pins were mapped backwards so we had to run it using more bodge wires while waiting for the new ones to show up.
Nick had to leave mid January and things took a brief hiatus while I readjusted to being here on my bill. But eventually I got back into the swing with the arrival of the new parts.
It was however time to stop faffing and get on with the battery. We tried it before Nick left but my 2x3s LiPos were out of commission and some 6s ones I got were too much for the KWeld. Instead with a new Turnigy battery I got the welder working an was ready to get the 7s1p pack made.
Balance leads were a bit thick but I didn't want to re-crimp and pin the header so used the one that came with the BMS and used a little bit of hotsnot to hold it in place before putting the blue goodness around the pack.
I don't know why but the pack just feels unsafe until the blue stuff is wrapped around it. Before I'd tentatively pick it up but now I'll throw and twirl it about without a care. Might say more about me personally. Out of sight out of mind... definitely out of my mind!
I still needed some stuff to make the harness so I instead moved to cutting the motor cable to length. Risky step and I don't like it. Too much cut off and it won't reach. Too much and you have to do it again >.> But I managed to nail it so we good 🥰
However when I went to test it I got an error I wasn't expecting. Apparently phase 3 was drawing current no matter what the board was doing.
Long story short Phase 3 doesn't get monitored by the DRV chip but instead via U3. This appeared to have cooked itself so if you have a cheap foccer 2 and get HIGH_OFFSET_CURRENT_SENSOR_3 remove U3 and replace it. Funnily enough Nick also had this happen to a VESC he was using and the same fix was needed so seems to be a common enough fault.
No more fault light and a happy lil VESC. Didn't bother recording it spinning as it's the same as previous and I needed my phone to control it and record so wasn't gonna happen. Just have to trust me it works 😊
-
Next challenge was making the controller lid. I wasn't sure if I wanted to resin print it, resin cast it or pay to have it CNC'd out of polycarbonate. Since the former is easiest and the latter the hardest and more expensive I decided to give it a shot. My curing station looks like I'm up to some evil shenanigans so enjoy a picture of that.
I won't bore you with the early prototypes and failures with me snapping one in half because I overcooked it with the print settings. Instead here's pretty happy final versions
Bottom one is the new one. Seems counter intuitive at first but printing with all the nasty support on the flat nice side allows you to easily sand down the surface and get a suuuuuper smooth finish. 1000 grit final and a light buff with some clear coat made it pop.I still need to work on the gaskets and I had a dilema. Do the stock black one that's kinda boring and keep it vanilla ooooor use a much more thematically coloured TFL gasket. I put a poll out on Insta and the results spoke for themselves.
So guess I'll be making go faster blue gaskets when I get the material and dye.
Meanwhile a little care package came in the post. Thanks to the legend himself Radim Klaska for sending me a GT controller, BMS and a pint controller for some future devious plans. Using them side by side with the GTXS you can kinda see how similar I managed to get them.
I'll be pulling it apart to reverse engineer a custom controller and get a schematic out of it for some experiments I have planned. However here it serves well to demonstrate scale 😊Back to the main goal again. The board won't do much if I don't get a charger made for it so with callipers at the ready I managed to replicate a brick and cram some magic smoke inside. I'm using a USB PD module to take in power and request 20v. Then a CC-CV boost converter to take the 20v and send a set current into the board until it gets near 29.4v where it'll hold that voltage to top off the cells. Basically how any normal lithium charger works that's not complete trash.
For the charge port itself I'm using some magnetic pogo pin connectors. They work really well and despite being 6mm in diameter this one can do 2amps. Added bonus is they're super satisfying to snap in. It's like magsafe for Onewheels (for ants)
Will come back to this when I resin print the final parts for the plug and charge brick. In the meantime I needed to figure out the harness. I wanted it to be orange like the GTS but finding multicore cable with 2x14awg + 3x24awg wire inside less than 8mm in diameter was a bit of a stretch so I came up with a solution.
Use a mould to encase the wires taught and in position then pour a flexible silicone over. Once cured demould and use heat shrink to get a uniformly round and flexible cable. Heat shrink alone will be weird and kinky. as it conforms to the wires inside and be less durable. I did a test piece first with some offcuts and despite being very messy it worked!
You can also see I made my own glands. The existing ones on the market just didn't scale right and the fitment was poor so a bit of CAD resulted in some functional yet scale looking glands.In an older prototype the cable fits nice and snug. Minor kinks where some silicone hadn't set right or the wiring wasn't taught but overall looks great!
With that being a success it was time to crack on with the final one. I did have to crimp some insanely small connectors for the headlight and battery harness wires. Did these with some tweezers. They are tiny! Here's one under the scope.
I changed the mould a little this time and taught the cables before putting in which helped a lot although made slotting them in a faff. But it all went in well and cast really nicely.
I ran out of time last night to put the headers and connectors on controller side but I'll hopefully do that tonight and share the results of the board running off it's own power 🥰Until then enjoy this render and little pic I took for my application to this years OpenSauce.
