Thursday, 20 December 2018

The Sinclair C5 - Part 3

It's taken me some time to get round to writing up the Sinclair C5 at EMF Camp. Perhaps that ought to be a one of my new years resolutions? Any way on with the writeup!
One major concern I had was how was I going to transport the Sinclair C5, my tent & camping gear as well as all the extra widgets I wanted to bring? I also offered to give a friend a lift so needed to fit in all of her luggage too!

As I live in Cambridge, UK I have no need for a personal car. The public transport is dire compare to London but works well enough and everywhere is very accessible by bike (or Sinclair C5!) However I wasn't planning on building a trailer and somehow cycling from Cambridge to Eastnor (~135 miles). Instead I hired a van and hoped the Sinclair C5 would fit inside!


Luckily the one I ended up getting did, I had to remove the boot but it just fit inside and didn't have any room to move around much! I had a few last minute jobs to try and make the Sinclair C5 Hacky Racers legal which involved adding a fuse, power kill switch and a spot for the transponder. I used Fusion 360's very useful canvas feature to take a picture of the rear of the Sinclair C5 and then sketch on attachment points for these parts. I then laser cut this out of 3mm laser plywood on one of Makespace's laser cutters.


Everything fit brilliantly first time and I was thrilled with how it turned out. The self tapping screws went straight into the original holes in the Sinclair C5 chassis. The fuse holder and wires fitted really well into the holes I'd placed, the killswitch fit perfectly and the "OFF" sign was pointing in the correct direction and clearly marked what to do. The velcro tie points for the transponder worked really well and I was all set for EMF Camp.



Once I arrived at EMF Camp and had set up my camp as well as providing a little bit of assistance with the Hacky Racer course. Here's a few action photos:



Unfortunately the Sinclair C5 is not designed for racing. The wheelbase is too long and it just can't corner quickly at speed. The poor design of the original chassis makes it really hard to race in and my modifications today haven't helped much with the handling. This made driving round the courses rather challenging. The VESC speed controller also would often go into thermal overload which would stop the motor entirely until it cooled down.

Regretfully I pulled out of the Hacky Racers but continued to assist and roam around the campsite on the Sinclair C5. Which went really well right up until the gearbox mysteriously locked up around lunchtime on the last day. Saddened I had to pack it away in the van for the return trip. 

I did manage to catch one of my favorite moments of EMF Camp though, the silly vehicle parking at the closing ceremony: 


It was the wide verity of silly vehicles all collected together in an organised parking lot.


This is a good picture of how packed the van was for the trip home. 

Eventually once I got home I got round to taking apart the locked up gearbox to see what was wrong with it. Hmmm that doesn't look good, those pieces of planet gear in the planetary gearbox.


Oh dear, that's not what a planet gear is meant to look like. It's literally exploded!



Once I cleaned up the pieces I found another problem!


So in addition to one of the planet gears shattering it appears that the pins on the carrier plate have worn though most of the planets which is probably why one of them cracked.

I'm hoping that this is because these planet gears are the original gears that came with the planetary gearbox and they were not particularly well made. Either poorly machined or poorly hardened. I'm hoping that I'll be able to replace them with higher quality planet gears. If that doesn't work then I'm going to have to come up with another scheme to reduce the power. I may have to return to using spur gears. But until I get this fixed, the Sinclair C5 is off the road.



Wednesday, 31 October 2018

Upgrading Bourbon's Drive


Previously I used the standard 25mm gearbox and 1806 brushless gearmotor combination to drive Bourbon around. However one of the biggest problems with these gearboxes is that the gears jam and/or strip which makes using them for combat robotics a difficult prospect.

The main positive is that they're much lighter and more powerful than the original 25mm gearmotors. In addition because the 1806 brushless motors are so much lighter than the original DC motor they're less likely to pull apart between the gearbox and motor. Once I found a seller on AliExpress selling the raw gearboxes without the DC motor it was an easy switch for me.


In fact Bourbon is designed around these motors. The much longer length DC motors would not fit inside Bourbon's chassis. It did originally present some difficulty by having the rotating brushless motor can abrading wires but once I 3D printed a special electronics shield/caddy it all went together very nicely.


Finding new Gearboxes

However now I am starting work on the next version of Bourbon and I wanted to increase the drive power whilst increasing the drive reliability. The best way to do this I felt was to ditch the 25mm spur gearboxes for something better, capable of handling more power.


Weirdly while looking for more robust gearboxes I came across these planetary gearboxes which are actually smaller than my original 25mm gearboxes. And yet despite the diameter reduction they are full metal planetary gearboxes which means they are a lot more robust than the original spur gearboxes. The spur gearboxes would bind and lock up if the 2 internal axles were to go slightly out of perpendicularity. Also as spur gearboxes each gear only had ~1mm of tooth width to engage with the next gear which made it easy for teeth to strip.


These new planetary gearboxes have 3 gears at each stage which leads to almost 3mm of tooth width engaged. However the eBay listing for these new gearboxes does not list an exact ratio so I set about calculating it.

The Pinion gear has an Outer Diameter (OD) of 7.09mm and a tooth root diameter of 5.38mm. It has 16 teeth and so to calculate the pinion tooth Module: 7.09mm / (12 teeth + 2) which results in a tooth module of 0.50642MOD. That is probably not quite a real tooth module so I've rounded it up to 0.5MOD.

The gear ring has 44 teeth, the planet gears have 16 teeth and the carrier plates have 11 teeth so

12 teeth on motor pinion
12 teeth on 1st stage carrier

motor 🠢 12 tooth pinion 🠢 3x 16 tooth planets 🠢 44 tooth ring 🠢 12 tooth sun 🠢 3x 16 tooth planets 🠢 44 tooth ring 🠢 output shaft

The drive motor to the first carrier plate = 1:0.214
The first carrier to the output shaft = 1:0.214
The drive motor to the output shaft = 1:0.046 or 21.84:1

I also took the opportunity to upgrade the motors from the QUANUM MT1806 to the QUANUM BE1806 Race Edition. This boosted the drive power from 85 Watts to 230 Watts (if HobbyKing ratings can be believed...). Another advange of these new motors is the motor shaft has been increased in diameter to 3mm from 2mm. This makes it easier to mount the new pinion as well as improving the robustness of the motor. In addition it seems like the MT1806 is now discontinued and unavailable.

Fitting them in new Bourbon

One downside to these new gearboxes is that they're longer and heavier than the original 25mm gearboxes and they also have a completely different mounting pattern. I also wanted to fix Bourbon's main design weakness. The center of gravity is right behind the drum, this is great whilst Bourbon has all 4 wheels on the ground, but when flipped over there is very little weight on the rear wheels. These rear wheels were designed to protrude above the top/rear armour and would allow me to still drive in the event that Bourbon was flipped over. However in reality as the center of gravity is so far forwards they cannot gain any sort of traction and just spin not moving Bourbon around the arena.


To fix this problem I'm planning on moving the drive motors from behind the drum to the rear of the robot. This means that I need to completely rethink the belt path from the original Bourbon. Other parts of Bourbon will be remaining the same: the drum is working excellently and the armour is resisting opponents successfully.


Thanks to my access to the Makespace laser cutters I was able to knock up a quick and simple little test side rail to see how the belt would go together. In the above picture I'm using one my original wheels as well as a new custom 3D printed wheel to get a feel for the belt tensioning. I'm still working on this but hope to have everything ready before Christmas.

Tuesday, 24 July 2018

The Sinclair C5 - Part 2

From the previous post I had a list of things to tackle:
  • Replace the busted rear axle bearing
  • Add a brace for the rear axle and chassis
  • Improve the rear drum brake, possibly swap it for a disk brake
  • Add a display to report on things like battery capacity and motor temperature
The busted rear axle bearing was a simple fix; buy a new HK2016 needle roller bearing, and press-fit it into the axle bracket. In fact this happened so quickly I completely forgot to take a picture of the process. I did notice that this new HK2016 bearing had a metal bearing cage rather than the plastic one of my first bearing. Hopefully this would prove to be a little more robust.

As I didn't have access to a welder at Makespace I'd have to bolt a brace across the rear or the Sinclair C5 chassis to make up for the broken spot welds and to avoid breaking the spot welds on the other side.


This was an aluminium tube 4" by 1" and approximately 600mm long. I fitted it between the two straight sections at the back of the chassis, directly underneath the axle. I of course measured twice, cut once, and found it didn't quite fit so the holes were turned into slots. And now it's a feature!


Being a nice rigid box, it would stiffen up the rear of the chassis nicely and not add too much weight. It would also add a bit of protection should the wheels fail catastrophically as well as a wealth of mounting points. Unfortunately due to it's position it slightly interferes with the pedal chain so to avoid carving a hole in the aluminium tube with the steel chain I added a block of HDPE that was quickly shaped on Makespace's bandsaw and zip-tied in place. It helps to both protect the aluminium as well as guide the chain round the rear cog.




However in the process of mounting the aluminium tube I was fiddling with the rear drum brakes and managed to snap them. 

Oops.


From the way it had snapped I was quite happy I'd broken it while it was in Makespace, if it had happened at speed bad things could have happened. But it was still broken and considering my C5 was quite fast after the gearbox upgrade, I'd really like to stop when I wanted it to. I tried searching for a replacement, initially finding a Sturmey Archer drum brake for bicycles that looks like it would fit. But unfortunately it was for 70mm diameter drums, the Sinclair C5 brake is for a 2.5" drum, or 63mm for the less imperially minded. Very luckily I'd recently joined the Sinclair C5 Owners page on Facebook and found someone who could sell me a replacement. After a very pleasant transaction with a guy called Marc I was the owner of a new drum brake and fitted it the second it arrived.

The final remaining task is to reconnect the brake cable. When I added the rectangular aluminium  tube brace it interfered with the usual placement of the drum brake lever. I think I've found a way to get it reconnected and working but I do actually need to fit it.



Looking around on the internet were a few examples of people who had screwed disk brakes to their drum brake wheels and made an aluminium disk brake caliper mount out of aluminum bolted to their C5 chassis. Now while this would be a a vast improvement I didn't want to damage the original parts so I'll add disk brakes to a future set of upgrades involving a new chassis and wheels.

The pillar allows me to mount the rear drum brake cable so it pulls the lever that actuates the drum brake. It's a small bit of aluminium channel I found at Cambridge Makespace and then attached with 2 pop-rivets. It's not super robust but it held up well to my attempts to break it. A quick cycle round Makespace to prove it worked and then onto the next task.

In the previous post I mentioned I'd managed to get a load of working eBike batteries for a bargain. Unfortunately the largest batteries wouldn't fit in the Sinclair C5, they were too long and interfered with the pedals. However I figured that by dismantling them and re-arranging the cell packs I could get them into a more suitable format for a Sinclair C5.


In that .50 cal ammo box there are 2x 10s, 36V 16Amp hour batteries arranged so I have a single 32 Amp hour pack. I'm using an ammo box because it was a quarter of the price of a suitable outdoors rated plastic box. In addition as it's metal it'll handle accidental impacts much better and should bad things happen, it'll contain the fire much more effectively.

I've been jokingly saying "if seat warmer engages, exit the vehicle and run!" but that really is the procedure.

I tidied up the mess of wires in the above pictures and that was good enough to drive to the pub but I also ordered a new Battery Management System (BMS) board. The new one was capable of switching 100 Amps and also had some intelligence and would communicate over serial, or Bluetooth with the included dongle. The plan is to connect this to some sort of display so I can get battery level information as I zip around town. No more running out of battery on the wrong side of town again, hopefully...


The above picture shows the combined battery packs with all the extra wiring removed and I'm just checking that the BMS will actually fit in the ammo box. I was planning to solder on the automotive in line Maxi Blade Fuse holder as well as thicken up the negative wire, in the above picture it's 14-12AWG which is too light.


I've also added the balance wires that let the BMS check how individual cells are getting on. The purple thing to the bottom right is the 100A Maxi Fuse, while the BMS should disconnect when it sees too much current flowing the fuse acts as a fail safe.


The main attraction of this BMS was the USB/Bluetooth data link. It appeared to be some sort of serial link and I'm planning to use a Raspberry Pi as a display module to show this data in real time. I should be able to get the battery status, discharging and charging rates on a small 2.4" LCD for when I'm driving along.


With the Battery box functionally complete and the rear brake fixed I figured it was time for some test driving in Cambridge, UK. I went to meetup with some friends and ate sushi outside Kings College while relaxing in my Sinclair C5. I got quite a few interested people asking me questions which was quite nice.


I took it for quite a few drives as well as another high speed run however my top speed didn't really improve much, only managed to increase it to 28 mph. I think this may have been due to the brakes not quite being tuned correctly and the subsequent friction stopping me from hitting the goal of 30mph.

But the main thing is that the Sinclair C5 is in a good place for EMF Camp at the end of August, in fact I may also make the inaugural Hacky Racers event, The Scrumpy Cup!

The few remaining tasks are:
  • Tune the brakes
  • Add a display
  • Research all metal rear-wheels

Tuesday, 19 June 2018

The Sinclair C5 as it Should Have Been


It began in 2013 when I went for an interview in Cambridge, UK and stopped over at a friends house to catch up. He’d gone and bought a working Sinclair C5 from a local collector for £250 and I had a ride around his road in it. Despite my head being at the same high as the wing mirrors of all the cars on his street it was hilarious, to be zipping along, that close to the ground, with such little effort. It was then I knew I wanted a silly electric vehicle to call my own!

One of the things that really struck me was how futuristic, and battered, but mainly futuristic it still looked despite being released in 1985. As my friend remarked, it looked “very future”. However that was only skin deep. The original motor had been the best in DC motor technology at the time but these days brushless motors are the ‘in’ thing. With their compact size and high power-to-weight ratios I knew I needed one in my silly electric vehicle.

I’d been wanting to build a go-kart for a number of years but finding the space and time to get started was difficult. I also lacked a welding setup to get the chassis made up. I decided that buying a Sinclair C5 was a nice way of getting a rolling chassis very quickly and I’d get a very cool looking vehicle I could upgrade.

My first bit of luck came when I found an eBay auction for 26 eBike batteries for spares or repairs. These broken batteries, a mix of 36V 10Ah, 12Ah and 16Ah packs, would be ideal for electric vehicles assuming enough working cells could be rescued. I won the auction for £125 and trekked over to the otherside of Milton Keynes to pick them up. Once I got them back to my local hackspace, Cambridge Makespace I started to test them and found that aside from 2 dead packs the rest only had damage to the enclosures, the batteries were fine and in some cases completely unused.


Now I had a power source I still needed the chassis. It took quite some time before a Sinclair C5 went up on eBay for a price I was willing to pay. As I was planning on replacing all of the electronics I was hoping I could get one for less than £150. Eventually, after a few failed bids I won an auction for a rather disheveled Sinclair C5 for £72. It was missing the lights, the storage boot and all the electronics but I figured it was just right! I ended up buying the boot a bit later on for £100 and the rear light for £20.


The first thing I replaced on the chassis was the front wheel. The original had standard bicycle caliper brakes and a plastic rim wheel. If you used the brake too much the plastic rim would melt and warp. The tyre would then blow out and you would have a very bad time™. So thanks again to eBay I bought an aftermarket modification that would give me a metal rim wheel and a disk brake.


It was now May 2016 and I wanted to get the C5 running in time for Electromagnetic Field, a hacker camp in the UK countryside held in early August. I already had a suitable motor, an Alien Power Systems 6374 170kv 3.2kW motor however this motor would be too fast.

Brushless motors are usually described with a ‘kv’ rating, this stands for Motor Constant which is the RPM per Volt. By powering my 170kv motor from my 10 cell battery packs I would get a nominal 6290 RPM. The original C5 had a reduction of 1:3.8 with 16” wheels which would theoretically get me up to 80 miles an hour. Except I just wouldn’t have the torque to get to that speed. So my urgent requirement was to build, and test, a mount for my motor that would fit the Sinclair C5 as well as reduce the motor speed by approximately half.

It took a few late nights but eventually I came up with a mad idea, the outside of the gearbox could fit into the original motor mount, and as long as the output shaft wasn’t central that would provide an easy way to tension the drive belt. The second mad idea was that the outside of the gearbox didn’t actually have to be metal, I could just make it from stacked laser-cut wooden sheet.



I found some 9mm laser plywood lying around the hackspace and I knocked up the design in Fusion 360. I integrated motor mounts, hall effect sensor mounts, holes for bearings, lightening holes which doubled as hand grips and securing holes. It held 2 steel gears together, one mounted to the motor shaft, one on an idler shaft. This gave me an 1:1.83 reduction ratio which would reduce the top speed to a slightly more achievable 43.3 miles per hour.


During EMF Camp this worked really well. I used the C5 off road every day of the camp and it handled it amazingly. Occasionally a few bolts needed re-tightening and the bearings required some bearing restraints stapling into the wood but that was it for maintenance. Once I got the C5 back to Cambridge it managed to get up to 27 miles per hour, 7 mph above the UK speed limit for electric bicycles. However the gearbox wasn’t sustainable. Even with the gear reduction I had to pedal start the C5 otherwise the drive belt would slip on the tiny drive pulley and make an outrageous farting noise. I started plans for a larger gear reduction, possibly even a planetary to handle most of the speed reduction so the belt would become more of a way of coupling drive power to the wheels.

But it was at this point Robot Wars was resurrected by the BBC and as something of a lifelong dream of mine I dropped everything to become a part of this. The Sinclair C5 project was on pause.

Eventually I managed to pull myself out of Robot Wars preparations. In the meantime I’d managed to buy a NEMA 23 5:1 gearbox with the intention of fitting it to the Sinclair C5 as the main speed reducer. I took it apart and built a model in Fusion 360 so I could accurately model brackets and adaptors. This was really helpful when I needed to make the first custom mount, something that would fix it to the 6374 brushless motor. Unfortunately I made a mistake and 2 of the 8 holes on the adaptor ring are drilled in the wrong places and just won’t line up with their counterparts, 6 bolts out of 8 is enough right?


This new gearbox promised great things, by having the main reduction done by gears before the drive belt there was a good chance I’d be able to do powered starts as well as quicker starts in general. But in order to realise this I needed a new motor mount bracket for the Sinclair C5. I spent a long time procrastinating over this; how should the belt be angled, what if I wanted to add a neat design feature etc.

It wasn’t until Tindie/Hackaday announced a meetup in Cambridge that I finally decided to get on with it. I wanted to show the Sinclair C5 off, so I just had to make the bracket as is. I did some work on it the week before the meetup but it was only until the actual week of the meetup arrived that I really buckled down. I worked pretty determinedly, eventually taking the day of the meetup off work so I could really get it working in plenty of time. Everything went pretty well, I had to restrain my desire to make it completely neat and polished simply to get it working. Occasionally I had to abandon my theoretical plan and just bodge a few things, but luckily this was kept to a minimum. However come 4pm of the day of the meetup it was done!


I hopped in, zoomed out of Makespace, and drove across Cambridge, UK attracting quite a few stares and made it to the meetup! The power from the motor cut out a few times on the way there, at the time I felt that the motor had just overheated and needed a bit of a rest. Of course this wasn’t the whole story but I wouldn’t realise this until later...

It turned out that I’d got to the meetup a couple of hours too early, oh well, I’m in a pub, waiting for a meetup, time for a pint! After an excellent meetup hosted by Jenny List and Jasmine Beckett of HackADay and Tindie respectively where I met lots of equally mad people it was time to head home. The motor wouldn’t work the whole way home and so I had to arduously pedal the C5 back to Makespace. Pedaling the C5 is not like pedaling a normal bike, it sucks, really really sucks.

Finally I made it back to Makespace and it was time to inspect the chassis and see why it’d stopped working.


Oh, that’d do it...

I'd knacked the chassis, bent it like a banana, the cheap, weak Sinclair C5 chassis. It turns out a motor with 14 times more power that the chassis was designed for is a little brutal. I’d also mashed the rear axle bearing in addition to the chassis which explains why the rear axel had completely locked up.


Everything looked recoverable but I’d have to buy a new bearing and while I could bend the chassis back into the correct shape I’d actually popped 2 of the spot welds so it would just bend out of alignment again.

My list of things next to upgrade now included:

  • Replace the busted rear axle bearing
  • Add a brace for the rear axle and chassis
  • Improve the rear drum brake, possibly swap it for a disk brake
  • Add a display to report on things like battery capacity and motor temperature

As can be seen I only seem to get any productive work done on the Sinclair C5 when I have a deadline so I will be aiming to enter Hacky Racers at EMF Camp 2018.

Acknowledgements

Thanks to Makespace for the facilities and storage
Brian Corteil for the transport, advice and mocking
Mark Mellors for advice and mocking as well
Alec Wright for the idea

Sunday, 11 February 2018

Bourbon at the UWE Beetle Brawl Feb 2018

The University of West England held a Beetle Brawl on the 10th Feburary 2018. I managed to secure a place for 'Bourbon' and competed. This was to be the first time I'd get to use my newly designed wubber drum. A drum spinner with the motor driving the drum via a rubber wub to isolate the motor from impacts and shocks.

The Heats

Bourbon's first fight was against 'Puns & Roses' a robot I'd helped to build. As 'Puns' was a wedge bot it was my drum bot's natural enemy. We were both quite evenly matched in our drive power so couldn't out push each other. Eventually I managed to attack from behind and get some bite into the HDPE chassis of 'Puns' resulting in them being flipped out of the arena into the trench.




The second fight in my heat was against a student bot 'Critical Failure'. This bot had two spinning blades but they were too high up to actually hit 'Bourbon' so did no damage. Due to the thin mild steel chassis and vertical sides I was able to quickly get a good bite and throw it around. It went out as I managed to beach it on its back immobilising it.



After winning both my fights I was though to the 2nd Round

The 2nd Round

The first fight was against another student bot, 'Crabsolutely Clawful' which was my favorite bot in the competition. Even if I won I'd be knocking an awesome bot out! It was a HDPE chassis with 3D printed claw arms, I managed to throw it around and eventually broke one of the claw linkages when the driver tapped out. Luckily I hadn't caused any damage he couldn't repair so everything was okay!



The second fight was against 'Dr. Bees'. I knew this fight was going to be nasty as 'Dr. Bees' was a powerful horizontal spinner with a large Hardox bar. At the time I figured that by aggressively attacking with my new wubber drum I'd cause more damage to him than myself. This was partially correct, I managed to stop his spinning bar and stop one of his drive motors at the cost of flipping 'Bourbon' over, jamming the drum and dislodging one of 'Bourbon's Hardox teeth. I won due to immobilisation as I was able to still "move" (more like crab) around the arena.



However this is the fight that caused the biggest problems. I needed to strip 'Bourbon' down and see why the drum stopped, fix it and put it back together. But due to the event running rather late I was pressured in to quickly throwing my spare, old style drum, onto 'Bourbon' not checking if it worked before going back into the arena.

The Finals

I was up against 'Incomplete Control' a ridiculously good 4WD pusher. Honestly I wasn't expecting to win this fight. The driver was better than me, 'Incomplete Control' can out push 'Bourbon' and I was rushing due to almost no fixing time. When the match started I attempted to spin the drum up but due to my rush I had tighten the bulkheads too tight and the drum was unable to spin. The hope was I'd be able to flip 'Incomplete Control' away enough times to give me a fighting chance, instead I was quickly pushed in to the pit for a very unsatisfying end.



However all was not lost, I then fell though to a losers melee where I could redeem myself and fight off two other bots for a return to the running for 3rd/4th place. I went downstairs for a chance to look over my robot and try and get the drum running before I was immediately called back to the arena. Luckily I was able to get the drum working and actualyl had a weapon for this match.

I managed to get some really good hits in before getting flipped by a poorly considered attack on 'Swag Daemon'. The drum had dislodged on the last hit and kept me propped up such that I was unable to move. The remaining bot in the melee 'Mr Cat’s Mouse House' took the victory as the remaining moving robot.




Conclusion

'Bourbon' is a very nasty little drum spinner that can both deal out massive hits as well as take them. However its biggest weakness is being flipped upside down which means it has poor maneuverability. I didn't have any major drive problems and the wubber drum worked superbly. In fact once the competition was finished I was able to inspect the drum properly and found that it was still operation but the idler bearing had dislodged which had jammed the drum.

In the end I came in 6th place out of 36 robots so I am very pleased with that result as I felt I ought to get into the top 10, and had a good chance at the top 3. On the other hand I am very frustrated that I was given insufficient time to properly inspect and repair 'Bourbon' after 'Dr. Bees' and so ended up fighting a match without a weapon.