EtekChopper: Electrical Mega-Update

Brace yourselves, this is a Mega-Update!

Last we left off, the mighty brushless Etek/MarsElectric/Motenergy motor had its sensors replaced. It's now time to start putting everything together to start rotating these tires.  

The 35-pin AMPSeal connector required by the Sevcon Gen4 came in, as well as the proper crimp terminals. It's a beefy connector, with rubber gaskets and other protection to ensure the leads do not short and that rain stays at bay.

Looking at the Sevcon wiring diagram in the manual, it dawned on me how complicated this setup was going to be. I need a Main Throttle, a Regenerative Braking Throttle, a Key Switch for logic power, a Contactor to enable supplying motor power, Hall Sensor inputs, an Enable Switch, and CAN Bus inputs for programming the controller with the motor settings and desired configuration. Ultimately, I identified the exact pins I would require: 

• 1: Key Switch In (Supplies logic power. There are three of these for convenience.)
• 2: CAN Termination - Short to pin 24 (CAN Bus)
• 3: Contactor 1 Return
• 4: Contactor 1 Supply
• 5: Encoder "U"
• 6: Key Switch In
• 10: Key Switch In 
• 13: CAN High
• 15: Encoder Return 
• 17: Encoder "V"
• 18: Forward "Enable" Switch
• 22: Main Throttle Wiper In
• 23: Braking Throttle Wiper In
• 24: CAN Low
• 26: Encoder V+
• 29: Encoder "W"
• 34: Main Throttle Supply
• 35: Braking Throttle Supply

I started wiring some of this for an eventual initial test, but first I wanted to check whether the key switch on the motorcycle was sufficient for switching 72V at low current. 

The Key switch in the OFF position does not switch anything, nor does the Lock position, which locks the handlebars of the motorcycle to prevent theft. ON short two of the leads emerging from the switch, and P (Presumably "Park") shorts two others. It seems beefy enough to switch 72V at a low current. Looking further at the bike, I took to figuring out what parts of it to keep or throw away when it came to wiring and indication. 

The original HUD for the bike has a speedometer, engine tachometer, left and right turn signal indicator lights, odometer, brake light, oil pressure light, and high beam indicator light. While somewhat useful, I plan on using the Cycle Analyst (More on that later) as my speedometer and odometer, I can see whether or not the turn signals, brake, or high beam, are activated, and an engine tachometer is not useful if I'm not going to be shifting gears. 

This handlebar-mounted switchbox seems really useful, however. It sports a high-beam light switch (Low beams are always on on a motorcycle by default), Left and Right turn signal switches... 

And a horn button, all in one nifty package. I'm keeping this!

I also received the Cycle Analyst High-Current edition and 0.5-Ohm shunt in the mail. It has a lot of great features, and will serve as my battery voltage indicator, current/power draw indicator, "Gas" (capacity remaining) indicator, Speedometer, and both trip and Universal Odometer. The Cycle Analyst even has more advanced features  to limit current draw or maximum velocity (by putting it between your throttle and motor controller). Because I'm using such a smart motor controller that can already do this, I am foregoing these features. 

The screen is HUGE, and the buttons feel great. It's backlit, waterproof, and runs off of battery voltage. 

It comes with a magnet,which attaches to the bike wheel, and hall sensor which attaches to the front wheel fork. This, coupled with a programmable wheel size, allows the Cycle Analyst to accurately measure velocity and integrate it to act as an odometer. 

The High-Current edition of the Cycle Analyst requires the purchase of a separate shunt resistor for current sensing. This is a 0.5 Ohm shunt, and it's beefy, and can apparently handle 150A continuous (400A peak). 

  The instruction manual is well put-together, featuring informative diagrams and instructions. There's even information on how to hack the device by adding throttle input-output, modifying throttle limit curves, hooking up to a computer for data logging, and even uploading your own firmware. But more on this later, there were more wiring things to test!

Like the lights! While I had the motorcycle inside, I decided to check every light I had. Thanks to one of our favorite awesome battery company's bricks, I verified that the signal lights work, but weren't blinking (yet).

Here's the rear light that's always on. 

And here's the rear light + the brake light in tandem. Sweet. 

Now, because the previous owner of this bike purchased it to remove the exhaust pipes and take the front headlight and turn signals lights. Because of that, he gave the front windshield/headlight/turn signals from another motorcycle. I didn't plan on putting the entire assembly on my bike (It was kinda broken, and I'm not sure it could even mount onto my bike) so I decided to take them all apart and test them. I'll mount them individually with my own adapter hardware if necessary.

It's a shame, it's a really nice front windshield assembly. But hey, onwards. 

I found the part of the motorcycle that makes the turn signals blink! It's a relay-looking thing that probably operates on the same principle as thermally PWM-ing electrical stovetops. Wicked.

I decided to try and start fitting components to the bike while I had it indoors. They... kinda... fit? I'm gonna have to massage the bike a bit to fit those batteries. Chances of me fitting twice the amount of batteries are pretty slim...

 WIRES! I found some of LOLriokart's leftover heavy duty wiring somewhere in MITERS. I need to ensure all my wiring can handle the 350 Amps peak without melting, so 2AWG it is for me! In a hardheaded attempt to get the motor spinning RIGHT NOW, I bought the local True Value Hardware out of all their 2AWG ring terminals. All 6 or so of them. 

 One soldering iron was not enough to properly solder the terminals onto the wire that was about as thick as my thumb. Two irons were also barely enough. So i settled for a propane torch, which ended up working out quite nicely. 

 I dig the result.

 I then acquired the second of the 12S8P A123 packs, and machined an aluminum bus bar to strap between the "+" of one battery and the "-" of another. This left me with ~72 Scary-Ass Volts of A123 goodness. The whole thing was like 30 lbs. I can't even fathom how heavy that would be in Lead Acid packs...

The Tyco Kilovac EV200 is a 12V-activated contactor which can deal with a metric-ass-ton of current. That circuit you see (it's normally hidden beneath a plastic panel) is an Economizer, which ensures low power consumption after the initial contact and current surge. However, according to this post, the Sevcon has built-in economization, and actually cannot work with the Economizer of the EV200. I removed it and soldered the red and black leads directly to the two black leads. 

And here's everything needed to test the rig, almost ready to rock!

It will rock soon, I swear. 

Just hold tight.

Replacing a Brushless Etek/Mars/Motenergy Motor's Hall Sensors

Meet the brushless Etek/Mars/Motenergy motor (EMMM? EM^3? EM3?). This one has dead hall sensors. How do I know?

Oh. That's how. I also ran 10V to the the red line, tied the black line to GND and scoped the three other colored leads. Nada. I'm going to need those up-and-running if I'm going to be using the sensors-required Sevcon controller. 

I decided to call up Motenergy, the current makers of EtekMarsMotenergy Motors, and I have to say their customer service is absolutely top-notch. John Fiorenza, President of Motenergy and I believe one of the original designers of the Briggs and Stratton Etek motor, picked up the phone himself and seemed genuinely interested in my project. 

He offered to send me the replacement sensor board free of charge, but while I was in contact with Motenergy I purchased a replacement fan and fan cover for the back of the motor. 

It's also on the phone with John where I found out officially my model motor is one of only about 200 manufactured for a special order (someone at MITERS must have picked it up wholesale) and that it is, in fact, more powerful than an ME0907. Sweet. 

Now the issue is getting to the sensors. They are deep inside the motor, attached to the inside face of the back plate, past damn powerful magnets, and a shaft well press-fit into a bearing. And that's after you get past the fire-breathing dragon and beachhead, then perform an animal sacrifice. Taking the motor casing apart is like trying to pull apart two ultra-magnetic plates: very difficult. 

I got to work making a makeshift mount out of leftover aluminum. My plan is to put the motor in the mill and do...something with it. But I can't just clamp onto the tangential points of the cylindrical motor. The flat edges on this aluminum plate (which is mounted on the dedicated motor mount points anyway) will work perfectly. 

Man, I haven't machined anything in a while... 

Well, 3 out of the 4 holes ended up lining up. That's what I get for eyeballing it.

I mounted the motor on the mill on top of high parallels to allow the output shaft the room it needed. 

Using a length of steel cable, I tied figure-eight loop (not a smart idea with steel cable you want to reuse, but hey it was free and lying around) to one end and snaked the other end under this hole in the motor, over the head of the mill, under the opposite hole in the motor, then back over the mill. After that I pulled the standing end through the loop, tensioned it, and hitched it a couple times for a full hold. 

Then it was time to lower the Z axis. I slowly lowered it until...

With a SHUNK, the stator and back half of the motor came free!

I can see why these suckers were giving me trouble: they're HUGE. 

I removed the retaining ring from the remaining shaft, but it still would not budge out of the bearing. Time to bring in the big guns...

I began to push it out with a small arbor press when...

The rotating steel bottom of the arbor press flew up and into the motor -_-. It took Nancy and I 10 minutes of prying with a hammer and tapping with screwdrivers before we could get it out. 

And so I found parts of Chibikart's old steering assembly and used them as sort of "magnet condoms", due to their aluminum-non-magnetic-ness. After much pushing down with the arbor press, the shaft came out with a CRACK. 

Ooh, the inside of the EtekMarsMotenergy Motor! And there's the sensorboard for the taking! 

I wasn't expecting it to be all plastic. Then again, I imagine that's to ensure it is weatherproof. Unlike Cruscooter...

Curious, and knowing I would get a new one in a few days, I took it apart. There were the three hall sensors in some kind of SMT package. 

Out with the old and in with the new! Here is the new sensor board, fan, and fan cover I ordered.

And it fits like a glove, as expected. Time to repeat the above process in reverse order!

I have to press the shaft and magnet housing back through the back bearing. This time I needed a larger arbor press, so I headed to the FSAE/Solar Car shop.

I re-tied the back plate to the mill and prepared for reassembly. 

Before I could reassembly the whole thing, I needed to get the sensor leads back through to the rear end of the motor. This part was tricky, as the hole was teeny and there was a large wound stator in my way. I passed a small piece of twine down through the hole and tied a friction knot to the cable assembly. 

I then pulled it up though, eventually needing to remove the rubber gasket to fit the thick cable assembly up. 

I then raised the motor up to the rear cover, taking care to ensure they were properly aligned for when the force of the magnet took over. 

CRACK! 

The magnets grabbed it hard when it was close enough. Right on target. 

Time to test the new sensors...

YAY! All three channels worked properly. One last thing to do...

Sweet. See you next post!

EtekChopper: Drivetrain Design and Parts Selection

Time to get down to the nitty gritty and make some drivetrain decisions. I need to shuffle around the basic variables of my system in order to finalize the specifications and start buying parts. Decisions will be bold and underlined.

So far all I know is I am using a Sevcon Gen 4 motor controller  but I still need to choose a battery voltage, design or pick a battery pack, choose the motor, choose the gear ratio, etc. The constraints I'm working with are those of the Sevcon Gen4 controller, the space in my motorcycle, and the tug-of-war between performance and my wallet, among other things. 

The Sevcon Gen 4 wants a 72-80 Volt system. 72 Volts seems to be a standard voltage for things, so I will run a 72-Volt system. 
What does this mean? My battery will be a 72V  nominal. Assuming I'm going with 3.3 volt LiFePO4 cells, I am running a "24S" system, meaning 24 3.3V cells in series. Battery systems made up of smaller cells are usually defined by "XsYp", or X in series, Y in parallel. Parallel batteries allow for increased charge capacity and greater allowable output current. The more cells your have in parallel, the less electrical stress there is on one cell. 

Here is a CB750 frame solidworks model I found online, a motor for show, and four  A123 12S8P battery packs with included Battery Management Systems (BMS) that my friend Dane has access to and may give me. The BMS balances the charges of the many small 3.3V batteries to ensure the entire pack is healthy. Issue is, they are HUGE. I don't think I can fit more than two of them (for 12S8P) comfortably. I can probably fit the other two (12S16P) on sides, or as saddlebags. Maybe? Hmm. 

Or I can make my own custom 24S battery pack with some of the infinite 3.3V A123 cells, maybe 10P? That's 240 cells. Adding more cells in parallel will increase my maximum range, but how far exactly will that take me? 

ElectricMotion.org, the website build log for Lennon Rodgers' eMoto has a nifty spreadsheet available for download in the Specifications section where you can modify characteristics such as weight, battery size, voltage, gear ratio, etc and see projected top speed, power dissipation, range, etc. 

I took the liberty of uploading my own version of the Spreadsheet onto Google Docs. The most important parts of the document (to me) are the top speed and range. By changing the cruise speed (I usually have it set to 55mph, which is not energy efficient), drag coefficient, and other parameters, the spreadsheet spits out a maximum range estimate. I should be getting 15 to 30 miles of range on 24S8P of 2.5Ah cells, double that if I can fit 24S16P. By fiddling with the max voltage, the max motor RPM at 72 volts, and the gear ratio, I can find an achieve a max velocity that can keep me comfortable on a highway, so around 65mph. 

But before I can finalize all that, I need to find a motor that is happy running at 72 Volts (electric motor speed is proportional to input voltage).  For Cruscooter's Kv=190RPM/Volt motor to be running at 72 Volts, it will be running over 13000 RPM, which is REALLY FREAKING FAST. A quick hop over to McMaster shows that only their highest quality bearings have a maximum RPM that high. And they each cost half as much as Cruscooter's motor. 

So I need a low-Kv motor (Which translates to high torque constant) or one which can deal with being driven at many thousand RPM. I can go the way of Bayley and get myself a hybrid vehicle motor, like his Ford one. The issue is I would have to design my own case for it, like his, and would need to find one, which I can't seem to be able to do in five minutes of scanning Ebay. 

There's a HUGE Baldor Induction Motor Bayley was going to use in his motorcycle before he decided to use the Ford hybrid motor. Apparently Adam Bercu, another MITERS-goer, had gotten it to spin with the Sevcon Gen4 controller, albeit not efficiently. It also weighs like 100 lbs. 

Everyone around MITERS keeps praising the Etek motor, and there's apparently a few lying around up for grabs. 

Here's one I dug off a shelf at MITERS, though Charles tells me (and there's a sharpie scrawl on it) that the built-in hall sensors are dead. A quick check with an oscilloscope confirms this. I'm going to have to replace them. While this particular Etek-style motor was a custom order, not one that was ever available for retail, I am told it is almost identical to the ME0907, though a little bit more powerful

Looking at the datasheet of the ME0907 (Remember, mine is more powerful), it seems to be a decent-enough motor, but not quite taking full advantage of the Sevcon controller's 25kW possible power output. It has a Kv of less than 70 rpm/volt, so at 72 Volts it will spin at about 5000 RPM. It's 90% efficient at voltages from 24-48 (But I'm running at 72). It has a peak current of 250A, but the Sevcon can deliver another hundred.

Using these parameters I can get a ballpark for the absolute ideal peak power output of the ME0907 motor. A fully-charged A123 cell will be at 3.6 Volts. 

3.6V * 24 * 250A = 21.6 kW or 29 horsepower

The original CB750 engine was 68 horsepower peak, so at absolute peak ideal conditions this Etek motor can put out over 42% of the original engine's power. But the original CB750 weighed 500lbs, while I'm looking to cut it down to about 350lbs after gutting the 250-lb engine and adding 100lbs worth of batteries, controller, and motor. 

Power to weight ratio (PWR) is a thing, right? Assuming I'm 185 lbs (Well, I am), the PWR of the original CB750 (my weight included) would be 0.0992 hp/lb. EtekChopper would be 0.054 hp/lb. While I'm not even sure those are the right units (They seem to be inverted according to Wikipedia), it appears the ME0907 motor has over half the normalized power of the original engine. 

Considering the CB750 was dubbed a superbike, half the normalized power of that is not too shabby! Because it's direct drive electric, the torque to the ground is instant. Plus the ME0907 is free, but costs about $500 online. All I have to do is replace the sensors. I am using the ME0907 Etek Motor. (But you already knew that, didn't you! I have been calling this thing EtekChopper after all...)

Now it's time to choose the single-stage gear reduction. After tuning the gear ratio from 3.5 to 6, I settled on 5.8 (10 teeth at motor, 58 teeth at rear wheel) for the following reasons. 

I want decent wheel torque so I'm not sad when an intersection turns green. Most vehicles are defined by a 0 to 60 mph time. My 2005 Mazda 3 sedan goes 0 to 60 in about 8 seconds. The original '79 CB750 went 0 to 60 in about 5 seconds. Given the peak current (Charles claims this etek motor ), Torque constant, gear ratio, wheel radius (13" tire radius! I almost messed this up), loaded vehicle weight, a rough 0-60 time for my system can be found: 

60mph/(300A*0.13 NewtonMeters/A *(58/10) /(13") /((350+185)lbs)) = 9.5 seconds

So realistically it'll be more like 12 to 20 seconds because the above calculation doesn't take into account wind drag. As for city driving I can get to 20 mph in an ideal 3.17 seconds, so I'm content with this. 

As for top speed, I want to be able to drive on a highway if necessary, which is usually 55 mph minimum. I also want some wiggle room to be able to pass/dodge cars if necessary, to I shot for a 65mph top speed. Plugging into Wolframalpha the Kv, nominal battery voltage, gear ratio and wheel circumference yields: 

(68 RPM/Volt)*3.3V*24*(10/58)/Revolution*2*pi*13" = 71.8 mph

Sweet. Now, I know what you're thinking. But this time my system can put out 100A continuous. Just check out the following calculation of max motor force minus wind force at 65mph: 

0.13 Newton meters / Amp * 100 Amps * (58/10) /(13")
- (.5 *(1.275 kg/(m^3))* (65 mph) ^2 * 0.8 * .5 m^2) = 13.04 N 

I also need to decide what size chain to use. Too small and I'll be breaking the chain like Straight Razer ate through #25 chain. The original CB750 had #530 chain, which is quite a bit bigger according to gizmology.net's page on sprocket and chain sizes. I've heard of people using #40 and its vehicle derivatives on electric motorcycles, but this post of a broken 420 sprocket convinced me to stick with the larger #50-series of sprocket. 

But which one? What is the difference between #50, 520, and 530 chain/sprockets? This post has the answer, as does the gizmology page. #50 and 530 are nearly identical. Let's go with #50 chain! 

For the small 10-tooth sprocket, I looked no further than Surplus Center for an inexpensive steel one whose keyway and bore diameter was compatible with the 7/8" output shaft of the Etek motor. 

Here I introduce to you bikebandit.com. On this site you can specify the year, make, and model of your motorcycle and look through pages of the manual for the exact part you need. They also keep track of compatible aftermarket or universal parts and ofer those options as well. They SHOULD have the right rear sprocket, right? 

Wrong. Apparently 58 teeth is not available for the CB750. The highest rear sprocket available is 48 teeth, which simply wouldn't be sufficient for my torque/speed needs based on my rear wheel size. I guess I'm gonna have to make my own then! (Which I've done before for MelonChopper, I just have been too lazy to blog about it yet). 

Some quick searches on Google led me to Sprocketeer Online! It takes in the desired pitch, roller diameter, and teeth, and generates both a DXF file and generic CNC code for milling out a sprocket. I then simply imported the resultant DXF into Solidworks and extruded it to the right thickness (0.343"). 

However, In order to make the rear 58-tooth 530 sprocket compatible with the CB750's rear wheel mounting holes, I had to either measure the current one or find a drawing with the proper dimensions. I found a drawing at Dime City Cycles (third pic), and added it to the sprocket. I'll waterjet it out of steel or high-strength aluminum and file the edges until the chain fits. Trust me. It'll work. 

They also had 10 feet of #50 chain (Solidworks estimates I need about 5' 7") and master links/extender links for pretty cheap. I freakin' love SurplusCenter. 

Is that it? I think so. For now. Time to start buying stuff. 

Final Specifications:

• Overall Estimated Specs: 
• Weight: ~350 lbs (no driver)
• Maximum Speed: >65 mph
• Ideal Maximum Power: ~19.8 KW (~26.5 horsepower)
• 0 to 60 mph: >10 seconds
• Range: ~15 miles @55mph, ~30 miles @ 30mph 
• (double that if I can fit the other two battery packs)
• Motor controller: Sevcon Gen4
• Free, Estimated cost $925.00
• 350A peak, 140A continuous
• 72-80 nominal VDC, (39-116VDC Absolute limits)
• Motor: Custom Etek-style, Almost an ME0907. 
• Free, Estimated cost $500.00
• Happy at 24-48 Volts (I will drive it to 72)
• Over half the normalized power output of the original CB750 engine
• 300 Amp peak, 100 Amp continuous
• Batteries: 2x or 4x 12S8P A123 Packs, 
• Free (Thanks, Dane!), Estimated cost $936 each pack
• 72 Volts nominal
• Integrated Battery Management System
• 2.5Ah per cell, 2x or 4x (12 x 8) cells. 
• Charger: I'm liking this one so far: http://www.cloudelectric.com/product-p/bc-sco7220.htm
• $425.00, but I'm sure I can find a high current one for even cheaper. 
• Transmission: Single-Stage 5.8:1 gear ratio #50 (530) chain
• Motor Sprocket: 58 teeth, custom made
• Rear Sprocket: 10 teeth, from surpluscenter.com
• Chain: about 105 teeth, from surpluscenter.com