Would you believe me, if I would say to you that two electric propulsion specialists, Chris Jones and Mike O'Hanlon, have built one of the fastest electric motorcycles in their garage?
7/8/14 7:58 am chumakdenis 1
Would you believe me, if I would say to you that two electric propulsion specialists, Chris Jones and Mike O'Hanlon, have built one of the fastest electric motorcycles in their garage and Danny Pottage won on this motorcycle Australian electric Formula Xtreme Challenge?
I do know that sounds like a mystery, but that's true.
Australian electric Formula Xtreme Challenge
Round 1 of the 2014 Australian electric Formula Xtreme Challenge was held at Queensland Raceway
Five electric superbikes made the record grid with first time entrant O'Hanlon Electric Motorsport winning all four heats with their new Voltron Evo ridden by Danny Pottage.
Built in Western Australia by Dr Chris Jones, the new bike is based on a custom frame fitted with Turnigy 40c LiPo RC batteries that provide 700 VDC to an Evo Electric AFM-140 Axial Flux BLDC motor good for 400 Nm peak, 250 Nm continuous.
Wow, I'm stunned. But how could they get such awesome result? Well, let's go back to the very beginning and read history from one of those guys.
Сhris Jones story
I have a pre-build thread for my e-CRX conversion here on AEVA but I don't have a build thread for my new electric race bike. It's been a long and expensive journey to this point so I thought its time to share my experience. I do have a full build thread on Endless-Sphere, which you are most welcome to read, but for this forum I want to share not just what I did, but the reasoning and planning behind each design feature. So let's make a start, shall we?
It all began back in 2011 when I'd built Voltron the electric Suzuki RG. It had a pair of Agni brushed DC motors fused at the shaft, they were wired in parallel from a 1200 A Kelly controller and 6 kWh of A123 lithium powered the 160 kg bike up to 180 km/h. Together with Dan and Jason/Jon, we made a bit of Aussie electric racing history. And it was a piece of sh*t. #89 was 'unique'.
My building skills leave a lot to be desired, but the point is, I built it, I raced it, I had some wins and I had some DNFs. But boy, did I learn a lot from the experience. The bike handled like crap around corners, the motors were prone to overheat and throw molten solder, and the while bike was a safety risk should it ever get wet. I knew I had to do better. If I was to build another electric bike, it needed to be faster, have a higher thermal budget and be more 'modular' - that is, I needed to be able to pull each part off and put it back on again in minutes, not days.
What the new bike needed to be
1. It needed to be build from a proper racing chassis. Not some agricultural experiment from 1985. I had contemplated using a modern day sports bike as a basis - either a CBR or a GSXR. It would need to be based on a 1000 cc equivalent as these had the room and the strength.
2. It needed a single, very powerful motor. Not a pair of DC Indian smoke machines. AC was most attractive as brushes are a real weak point at high motor speeds. I needed it to be competitive with at least a 600 cc petrol bike - meaning I need at least 90 kW at the rear wheel.
3. It needed to be modular. I need to be able to hoist the battery out in one go. I need to be able to drop the motor out in minutes, or pull the controller out if I had to work on the suspension linkages. Essentially, I want to be able to do a full stripdown of the key parts, do a service and put it all together between races, not between seasons.
4. It needed to be rugged. IP67 type rugged. This bike is for racing, so it will be crashed and thrown up the road at 100 km/h. I'd like to be able to wheel it back to the pits and replace the broken bits and get back on track. Motorcycles can survive some fast crashes if they do it right, but not if the important components are held in place by compression straps and gaffa tape. I also wanted it to be raced in the rain without fear of shocks.
I'd not long had some success at Winton raceway when I started to make plans for the new bike. Where to start? Chassis, or components? Chassis selection was going to be tough as it limits your choice of components. I sought out donor chassis which had already been race-prepped. Eventually I settled on the GSXR1000. The 2004 GSXR had a huge main frame with enough room to fit stacks of cells and components. I even measured up a 2007 GSXR and it was even better - vastly improved handling with a big frame. I found a 2005 GSXR1000 for sale in Melbourne, and after much diligent saving, I bought it and had it shipped over.
And f**k me the frame was too small . Turns out the 2005 GSXR was the only bike that decade to have the smaller frame. Oh well, I thought, at least I have a mean 170 hp ICE bike. If I did try to build an electric bike from this frame I would be struggling to fit even 7 kWh worth of cells and who know what kind of motor.
It was looking increasingly like I need a fully custom racing frame - one without exhaust header bolts and countershaft mounts...
Around about this time I was trying to decide which motor I should use. I wanted it to be powerful. Stupidly powerful. More power than I would ever use in a race. Why? Because I needed my racing skills to be well clear of the thermal limits of the motor. My DC motors would have put out a mere 45 kW on a good day, yet they were always on the brink of incineration. Having more and using proportionally less of it eaves me with enough headroom to push my limits before the motors limits were pushed.
I learned that there was a certain power envelope which was absent in the electric motor world. This happened to be about the 60-90 kW region. It has since gotten better, but at the time I was a bit stuck for options. I could choose between the UQM Powerphase 125 motor, the Carbon motor, an Emrax motor, Remy, Yasa or Evo. One thing was obvious - power to weight was going to be the main driver here. And continuous power, not peak. When you are racing, you are hard on the throttle 90% of the time, so 'peak' just isn't good enough. Based on continuous power output, I needed something in the order of 2 kW/kg.
I'd narrowed it down to UQM, Evo or Yasa. Yasa was simply too expensive. It was a $14,000 motor, although on paper it's power to weight of 3.8 kW/kg might be worth every penny. UQM was also expensive - $22,000 but that included the inverter. It's power to weight was a bit low for my liking, and the size of the motor was hard to accommodate. Evo made two sizes - the AFM130 and the AFM 140. The 130 was 300 mm in diameter, while the 140 was 400 mm across. Remarkably, they were barely 115 mm wide. Axial flux is where the magic happened. Seems you can get much better power to weight by intersecting magnetic fields axially rather than radially. And torque! This motor has it in tiptrucks.
Lots of things lead to my decision to get the Evo motor. The $11,000 landed price tag was a bit hard to swallow, but it's certainly cheaper than the alternatives. But it had the high continuous power I wanted. The 130 or the 140? I settled on the 140 despite it's large diameter. On paper it claims 75 kW all day long, but the guys at Evo have tested it at 100 kW for several hours. They doubted the 130's ability to do this, but were confident the 140 could deliver. I committed to it
It needs high volts.There were winding configurations to choose from, and I choose the AFM140-4
AFM140-4 - a means to a custom frame end
It was clear the big motor would need a custom frame. I tried to visualise all sorts of ways of getting the motor in there, and while they could be done, it would mean a compromise on the battery space.
Something I've always felt electric motorcycles were ripe for, was the concentric, or at least near-concentric swingarm. They are an engineering headache, but they get the mass of the motor down and out of the way. Contrary to popular belief, you do not want a low centre or gravity on a race bike. Need proof? Ride an E-max 110s scooter as fast as you can towards your favourite corner and attempt to throw it in (biker slang for 'make the turn as fast as you can'). It will push the front wheel to the point of loosing traction and it will buck like crazy. You need weight up high so you have control over the countersteering forces. But I digress...
The swing arm would need to go around the motor. Otherwise the motor would be too far forward resulting in a long chain deflection and no room for your batteries. Being a large diameter motor you couldn't run a jackshaft to the drive sprocket very easily either. Also, where would the shock absorber go? It needs to be set up as a cantiever, above the motor where it still offers plenty of travel but clears the motor. So I came up with some sketches to see how it would work.
Eventually I came up with something I felt would accommodate the motor, satisfy the suspension needs, have a removable seat subframe (the first thing to get damaged in a crash) and be relatively easy to fabricate.
The large frame meant I could have a single removable battery pack while the cantilever swingarm would work around the motor. The motor could be removed by dropping it out the bottom. It also meant I could access the drive sprocket fairly easily, and ideally without having to remove anything large and heavy.
I asked around a few Australian firms to see if they would build the frame for me. Most said no, some said maybe but at a price (a very large one) and a few said it was 'impossible' to build an electric race bike. This post is for you then buddy
Eventually I found these guys: www.framecrafters.net
The father and son duo of Randy and Karsten Illg have been building custom frames for motorcycle racing for many years and are always pushing the boundaries of design. Their bikes are amazing and they win races too. They are explicitly NOT chopper guys. I got in touch with Randy by email and phone and he was very excited to be involved. There was a slight problem - they are based in Union Illinois. Shipping a custom frame to Perth won't be cheap, but I know the frame will be perfect so it's worth it.
I placed my order back in November and work has continued steadily since. I had to get Randy to build a mock-motor, as he really needs it to know where to place mounting holes and how to work around it. So he knocked this up based on the dimensioned drawing from Evo.
I had more or less settled on the dimensions of the battery pack too. Not certain about the height, but the width and length were known to be 320 mm x 220 mm respectively. With these figures he got to work on the frame. The swingarm could have been a custom job but I found that a Ducati supersport 1000 allow swing arm would almost fit the bill perfectly. I bought one off ebay sight unseen and had it sent to Framecrafters. The motor plates were machined and fitted to the mock motor, and these were set inside the swing arm.
These were ready to be fitted to the trellis frame. Their bi-metallic frames are epoxied and bolted together nice and tight. And they look amazing. Here it is in it's latest incarnation:
Bit of a fatty, but no fatter than a typical non-Honda Japanese litre bike.
Looking nice and shiny
Danger! High Voltage!
Buying a really powerful synchronous AC motor is pretty straight forward until you start to read into it. Typically it's not the motor which is all that hard to decide on; it's the inverter (controller, variable frequency drive) to run it that seems to be the sticky point. You need an inverter which is rated at or just below the power output of your motor. Any more and you risk cooking the motor, any less and you will cook your inverter. Inverters come in three sizes - small, big and epic. If your motor is between these sizes, it's not a big problem, you just won't get full use out of it. I found many inverters out there which are able to do the job, but which one was worth it?
I'd considered inverters from UniTek in Germany, Sevcon in the UK, Rinehart in the USA and Tritium here in Oz. Given my motor was capable of 100 kW continuous, and up to 170 kW peak, I really needed an inverter capable of more than 100 kW. The Tritium Wavesculptor was a good contender as it was much cheaper than the rest, and still promised 165 kVA. With a maximum battery voltage of 450 VDC and peak motor currents of 300 A rms (root mean squared) it was looking good. However I worked out that it would fall short of what the motor was capable of. A 450 V maximum would mean about 400 V from the battery nominally. Given the sine wave nature of the power delivery to the motor, your motor would only see 283 V rms. So motor power is actually going to be 283 V rms * 300 A rms * 0.77 (motor power factor) * 0.91 (motor efficiency) * (sqrt 3)= 103 kW peak. This is a lot of power by anyone's measure, but not enough to really make use of this motor.
I needed more like 150 kW peak from the motor, given the continuous needs of racing. So the Rinehart PM150 series will fit the bill. These inverters come in two flavours - 360 VDC maximum battery pack, or the 720 V maximum. Hmm, these voltage sound kind of high...
This is where the motor winding comes in. I asked Evo which one I should get and they suggested either the 3 turn or the 4 turn motor. The 3 turn motor would operate on a lower base voltage (450 V DC bus) but it would have less inductance, resulting in a choppier wave form, more inefficiencies and thus heating. Heat would damage the permanent magnets in my $11,000 motor so I was cautious... The 4 turn motor was going to be better in this regard, but would call for a >700 V DC bus. Wow. OK, do I go for the low base voltage motor and accept a motor spinning slower than I want it to as the inverter is only capable of handling up to 360 V DC, or go all the way up to 700 V DC and use all of the silicon available?
It comes down to field weakening, and the inverter's ability to adjust it's timing such that the back EMF being generated by the spinning rotor is defeated by changing the timing and duration of the stator magnetic field. This allows the motor to spin faster at the expense of torque. At least this is my crude understanding of it. The 4 turn motor was better able to spin up to full speed by means of field weakening than the 3 turn motor given the higher motor voltage.
So I wasn't making a simple decision on the winding configuration, I was making a decision on the motor, the inverter and the battery voltage all at once! Why buy a motor designed to run on a lower, yet safer voltage if you aren't able to take advantage of the power available from the inverter? Volting up was not impossible, but certainly challenging. I have settled on the PM150DZ and hope to purchase one this year. Not sure what their price is, but it's close to $10k.
I settled on 700 V DC at top of charge - the maximum allowed by TTXGP and about as high as I would ever want to go. This meant the PM150DZ inverter was going to be able to spin it up to full speed (5000 rpm) and I could take advantage of most of the silicon available in the inverter. After all, doing the motor power numbers on this inverter gives me 158 kW at 640 V DC nominally.
So that's like, 168 cells in series!?
Not only was it a lot of cells in series, it had to be this number so that I could build a convenient battery pack from the LiCo cells I solder together. My pack measures 320 x 220, which is two lots of four cells side by side. The number of cells in each quarter pack had to be a number which can be divided by two to give an odd number - an odd number means the cell terminations can turn around and come back up the other way. Confused yet?
So I have four blocks of 42 cells with the negative at the top left. The cells snake down to the bottom, and at cell 21 link across the the negative of the adjacent stack. This then snakes it's way back to the top where the positive of cell 42 is on the opposite side as the negative. However it is now able to snake back down the second block of 42 cells. This way I have all of the cell terminals at the top of the pack where I can isolate them, put fuses and contactors etc. It is probably the most space efficient way to stick 9.2 kWh worth of cells into a 50 litre volume.
I have the motor's needs met by the inverter, and the inverter's needs mostly met by my battery. The battery's needs are met by the space constraints of the bike. All that's left now is to design the crash-proof, water proof enclosure for my cells.
Designing an isolated battery pack
One thing about high voltages is that they carry higher currents for a given resistance. The resistance of humid air is pretty high, but at 700 V top of charge, there is a real risk that someone could get a shock without even going in search of one. So isolation was important.
I needed to be convinced that electrically isolating the pack into lower voltage units was a good idea. Thankfully the AEVA has some great folks who are very knowledgeable on these things, and the Webber/Coloumb duo with their wealth of experience of trying to make a high voltage MX5 could offer plenty of sound advice.
I wanted to keep the 42 series blocks of cells separated from each other physically as well as electrically, so I designed a polycarbonate box with dividers. The terminals are at the top, so mounting isolation contactors and fuses up there will be easy. I also wanted to design the pack so it was fully self-contained. All isolation, monitoring and charge logic could be done from inside the enclosure. Also, I wanted it so you can't get power out unless you put 12 V into it. It was a tight fit, but through the use of a main control box on the outside I could house a shunt/current sensor as well as the main contactor for providing power to the inverter.
This is what I came up with:
The cells go in the bottom, the isolation contactors and fuses go in the middle shelf and the battery monitoring electronics go in the top shelf. Power is fed via the mid shelf to the control box on the back of the pack. The charge leads are accessible from the front of the pack while the 12 V logic to drive all of the contactors is contained in the control box.
The M8 stainless steel studs are where the battery sub-packs slide in and terminate. They will use forked copper bar just like fork terminals. This enables all of the isolation equipment above to be ready and in working order before you slide the packs in. The studs are fixed in place so there is no need to hold an insulated spanner underneath while you ratchet up the nut with a socket.
The 42 s packs will have use centralised monitoring system. Yes, this means having 43 wires which are directly connected to the cells ending in a large serial plug which plugs into a monitoring circuit on the top shelf. It won't do any balancing via bypass shunts - just sound a warning if a cell is out of range. This board is in development right now.
What does a 42 series wiring harness look like?
They will be soldered onto the sub-packs once they are build and once neatly cable-tied up, will plug into the BMS units on the top shelf, like this:
See how leaving the gap in the middle was deliberate
This is what the pack looks like now that I've finished everythig except the batteries which slide in:
The top of the pack, looking through the BMS shelf. You can see the careful layout of the contactors, fuses, terminal studs and the like.
I have used that 'noodle' stuff wherever a wire might potentially rub on a conductor or edge of polycarbonate. Split conduit for everything else. The charge leads go across the half-packs and together with the charge enable pair, head out the front of the pack to the charge socket.
The right hand side:
With the BMS shelf removed, the packs will simply slide up and onto the supports with the stud in the middle. The charge ring-lug is positioned and the M8 nuts is tightened down with an insulated 13 mm socket. Then the BMS shelf is returned and the pack is ready for it's side covers!
When my frame finally arrived from the USA I dropped the battery in to see how it fit... and it didn't. However, by mounting the main control box at an angle under the tank cover it will actually fit quite well. All is not lost
Lithium Cobalt cells - won't they burn your house down?
Then don't park your bike in the house. Geez.
But what of the cells themselves? These much maligned cells are the backbone of the radio control hobbyist. Built into a pack like this they store nearly 130 Wh/kg and about 200 Wh/l. They can crank out the amps like crazy while the voltage sags very little. I have done a discharge test on these cells at 13 C and the voltage sagged by about 5%.
Like all LiCo cells they cannot be taken past about 3.5 V or irreversible damage will be done. Attempting to charge them from this state may cause excessive heating and potentially a fire. Not what you want when you have 75 kg of cells there. But they are the right price and they deliver the goods.
How are the 42s sub-packs built?
I got Dmitrii in Melbourne to generate Gerber files for me, based on some crappy MS Paint sketches. The cells come from Herewin in Shenzheng, and are the 5 Ah, 40C type. These cells are exactly 50 mm wide and 9 mm thick; again, all part of my elaborate mathematical invention.
I start with a jig, made to the exact dimensions I'm after, give or take a mm. The jig doesn't really serve to compress the cells, rather keep them square and prevent them from slumping to one side.
Then, after trimming the tabs back to about 5-7 mm in length, I slide them into the slots on the board in the right order. If you get it wrong, there's a 2 in 3 chance you'll get a little spark to keep you in your toes. I bend the tabs over firmly with a blunt plastic tool and make sure they are fairly flat.
Then I go through and scuff each tab with the Dremmell. This takes a while but it;s necessary to get some good tinning. I tin the tabs generously with solder and my 80 W iron.
I don't normally put the fly-wires on before soldering the copper bus down, so just ignore them. I usually loop a nice ring of soler wire around the pad and lower a tinned copper block down.
I then lower my third hand down with a piece of insulating material before soldering the block with a pair of 200 W irons. As long as you are quick, this doesn't damage the cells. I use a 5 kg lump of steel.
Then then the bus has been soldered down I can put a G10-FR4 layer down with holes Dremmelled out where the balance leads go. These holes are typically about 12 mm in diameter, so I can fit the point of a hot 80 W iron down into the PCB plated holes. The packs will have the DD-50 already wired up, so it's just a matter of soldering the right leads in the right order. It will look a bit like this:
Except 567 mm long, two cells wide and with 43 leads going to a serial socket. This serial socket will have enough length to go past the switchyard level and up to the BMS level.
The wiring diagram
Isolating the battery into four separate sub-packs of 175 V each (maximum) would be achieved using Gigavac contactors which come with a set of auxiliary contacts. I specifically wanted these as it allows you to run a separate circuit through them to an LED to let you know that it's shut. I think all contactors in use in EVs should have this feature. I wanted these to open every time the bike was turned off, and the only time 700 V was available was when the bike was being ridden. So some clever relay logic was needed. Again, with the help of the AEVA forum a new diagram was made up, incorporating a precharge delay as well as some "charge enable" logic which disabled the main contactor so you couldn't ride off with the charge lead in.
Charging was going to be a pain. Finding stuff that's rated to 700 V isn't easy. So I had the option of either running two 350 V chargers in series and charging at 700 V DC, or I could fully isolate two half packs and charge them at 350 V each. This way 700 V wouldn't be found anywhere in the system. I opted for this latter choice as suitably rated parts can be found, and because the pack was already going to have a half-pack isolation contactor anyway. I just needed to draw this all out:
The relay logic board was able to power the quarter contactors separately to the half contactor, as well as the main contactor. The idea being that upon connecting a charge plug, a pilot circuit is completed. This energises the 'enable' relay coil, which switches the quarter pack contactors on and the half contactor off; creating two isolated half packs.
This was a bit too much for a 'squished fly' circuit, so I got a friend to generate some gerber files which I could send to China for the production of some PCBs:
Once populated I could do some testing:
hen turning the bike on, a switched +12 V signal closes the very expensive 1300 VDC rated precharge relay, the quarter contactors and the half contactor. Several seconds later the main contactor closes. This is important as the 560 uF capacitors in the inverter will create an impressive explosion if I don't charge them up softly. I plan on using four paralleled resistors giving 500 ohms. Using this awesome calculator:
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/capchg.html I determined that my precharge will be 99% over in 2 seconds. I can set the timer resistor to 23k ohm to give me a 2.5 second delay.
As the gigavac relay is a SPDT relay, I can use the same devise to discharge the capacitors when the power is cut. This is another rule stipuated by TTXGP - if the capacitors are still charged when the bike is turned off, there is a risk someone will get defibrillated should they go near the inverter.
I have spotted several issues with this circuit since devising it. Particularly, if the precharge relay fails for some reason, I have no way of knowing about it. Before you know it, 2 seconds later BAM your main contactor is welded shut. I had originally planned on using a two position key switch where position one is the precharge step and position two is the ON switch. It relies on the user to count to three before going all the way to ON. So what do you do? If I had a voltage monitor across the terminals I would be able to see the voltage quickly increase to full pack voltage before the contactor closed. If you see no such thing, your precharge circuit has failed and you have a chance to turn the key the other way and fix it. However the timer would give you 3 seconds to turn it the other way! Really I need a precharge system which prevents the contactor from closing should the voltage not rise to within a few volts of maximum. I know there are solid state relays out there which can do SPST switching, but the discharge is an important feature which I can't go without. Needs thought...
Here is a video of the relay logic in action -
LEDs are hooked up to the auxiliary contacts of each Gigavac so I can see if one is welded by inspecting the LEDs. The contactor most likely to weld is the main contactor, but at least if it does, there is additional isolation to prevent a fire. Maybe an optocoupler attached to the LED could prevent going from ACC to ON if any LED is illuminated?... One for another revision.
So my main control box needed an appropriately rated voltage divider. This calculator is pretty handy for this job. I sourced a pair of 5 watt, 680k ohm resistors, which I can put in series with a ~9.7k resistor for producing a 5 V signal proportional to the incoming voltage. I want it to be as close as possible to 5 V at top of charge, so I discovered this resistor combinator:http://www.qsl.net/in3otd/parallr.html which lets you come up with something pretty close to your desired resistance by paralleling or series-ing standard R value resistors.
The resistors at the bottom of the board are the voltage divider. Appropriately rated I might add . This sends a 0-5 V signal to the Cycle Analyst which then multiplies it by ~140 to give the pack voltage. The Cycle Analyst itself is powered by a separate 12V supply, however the grounds will be connected. Therefore I will need to power the CA through a small 12 V to 12 V DCDC converter for proper isolation.
Justin from E-bike.ca has been constantly improving his Cycle Analyst, which does all this and more. It still had the problem of running high voltages, so he changed it so that you can read up to about 800 V. He's such a great guy he even made a special splash screen for me!
This has not been a cheap bike to build. I will list them in order of components, and tally them up at the end.
AFM140-4 shipped from the UK - $11,000
Rinehart PM150DZ inverter - about $9000
Cells, busbar and PCBs - about $7000
Battery Monitoring System - $2000
Polycarbonate box fabrication - $1100
200 A, 700 V rated fuses (4) - $600
Gigavac Contactors (4) - $550
High Voltage Precharge relay - $280
Precharge resistors, PCB, ABS box - $160
Balance leads, DD-50 plugs (4) - $150
HV cables, lugs and busbar - $100
Custom frame, motor mounts, linkages - about $8000
Fairings, accessories, 12 Wiring loom etc - $1000
Running gear (brakes, forks, shock, wheels etc.) $2000
So we are a bit over $40,000. It might be a bit less, but most likely a bit more. This is what I expected, so I'm not too concerned. It will just take some time to save up for the bits I don't have yet.
Race bikes run on money you know.
Brilliant engineers Chris Jones and Mike O'Hanlon told us nothing about mass production, but we hope that one day you could see these electric motorcycles on sale for ordinary people.