Mechanical bits are no good without the electrical bits to run them. It's a well known fact that batteries are the biggest limiting factor in an electric vehicle
. Electrochemical technology alone cannot even begin to compete with the raw chemical energy stored in liquid and solid fuels at this time. Often, a good percentage of the weight of an EV is made of batteries, so if you're converting from scratch, make sure to take this into account. For a small vehicle like a scooter or bike, it's easy to tack on 30% or more of the original vehicle weight just for batteries. Chances are, you will have to either upgrade or replace any existing electrical system
. You've heard the whole talk about batteries - lead acid is cheap and heavy, lithium ion is expensive but awesome, and nickel chemistries float somewhere in between. The usual suspects in electric scooters and bikes these days are still lead-acid - usually the big 7-20AH types, and not particular high-amperage types (Just how many amps can you pull through a little 1/4" tab
When upgrading to a BLDC system, it's sort of counterproductive to keep the old, heavy batteries. Although this is easy to say, a good equivalent lithium polymer battery pack to a 12v, 7AH SLA battery (roughly 3 Li cells in series) will still run over $150-200. However, I think the expense is warranted. A high-discharge rated LiPo battery will supply more
of its rated capacity at sustained high current draw than an equivalent SLA - the chemistry itself is simply more efficient. You will get vastly improved battery life for an equivalent capacity simply by switching to a chemistry that can stand the discharge characteristics of an EV (heavy burst, moderate continuous current).
At the very least, I would either start with or upgrade to nickel-cadmium cells, of which there is a wide selection. The Sanyo N3000CR cells I used on this build are a classic, and can dump 80 amps continuously. If you want more energy density, look for nickel metal hydride cells. If you want ultimate performance, check out a set of lithium polymer or lithium iron phosphate
, which, although even more expensive, won't run the risk of catching on fire.
Not to fear-hype, of course, since LiPos are good at being stable if you charge them correctly. That's a good side-note actually. Buy a good charger
if you're switching chemistries! Here's a site that has a slew of single-duty dedichargers
. Many of those are OEM for companies like Razor or Currie, and while they don't have all the bells and whistles, tend to be plug-in-and-leave.
Here's a neat section of the RobotMarketPlace that lets you design and order your own battery pack
Pictured below is the stock 2.8AH, 24v sub-C pack that came with the Electric X2. As you can probably tell, it's neither 24 volts nor 2.8AH any more, as almost all the cells have corroded in some way. If you resize, you have to resize according to the predicted increase in power
. This should pretty much be common sense. If you can afford the space or cost, it might be better to go for something with more overhead so you have space to grow for future upgrades.
For this project, I splurged on a 100 amp, 44v brushless ESC
with a programming card. This was a full 100% nameplate rating over anything the rest of my electrical system could support at the time, but with the cost of large lithium batteries falling, I decided to give myself that room to expand.
I investigated several alternatives. One of them cost only $70, but was limited to 10S (about 37 volts), and came without mounting tabs. The rest were limited to 6 lithium cells. There was a decision to make - go with the 6S, inexpensive controllers and save on controller costs and battery charger costs, or go all-out with the 12S controller and as many cells as I could fit into the scooter.
I heeded the old adage "buy right or buy twice" and went with the latter option. Electronics are some of the most important parts of any project, and you should never skimp on them. The worst thing that can happen is to have an overloaded part fail while you're on the vehicle, because unless you have a good contingency plan, Really Bad Things will happen. Especially with R/C hobby equipment, which is often of dubious quality and stretched capacities and ratings, you want to give yourself some space, and never run components at their "maximum" power. Thus is the tradeoff between industrial and hobby parts. Make sure your new stuff can talk to your old stuff
. One of the biggest challenges with upgrading anything is backward compatibility (you software guys ought to know this well). If you are only upgrading one or two parts, for instance, the motor or controller, you will probably need to modify something unless it's a factory OEM part (then what's the fun in that?!). The biggest challenge by far is not interfacing physically, as wires can be swapped and connectors switched, but signalling.
Many stock DC motor controllers on bikes and scooters tend to be very simple devices. They are controlled by a 1 to 4 volt analog voltage, from any source. They often have no microprocessor in them at all, only a series of op-amps which are attached to a PWM generator that directly translates the analog voltage to an output. This 1-4v analog control is a standard in industry, and even large (and smarter) golf kart and forklift controllers use them (see 4QD
controllers, famous for driving many Robot Wars and Battlebots entries to victory). With that said, most of the throttles that interface you to said controllers output a plain 1-4v analog voltage, and is usually a big potentiometer with a spring loaded grip, pedal, handle, or whatnot. Simple, rugged, and proven.
When using an R/C controller, the signalling is very different. R/C servo control uses "pulse width modulation", which means the controller is driven by a series of digital pulses. The industry standard is 1.5ms (millisecond) long pulse for neutral, 1.0ms for full reverse, and 2.0ms for full forward, with pulses every 20 milliseconds or so. This is vastly different from what alot of electronics engineers think when they hear "PWM", which can also refer to a fixed frequency with variable duty cycle
whose intent is to generate an analog voltage out, after being filtered.
So chances are you have to upgrade your bike or scooter throttle so it can put out a servo pulse. Problem? I have never
seen one like that commercially! You will most likely have to rig your own signal interpreter, and there are numerous ways to do it.
. If you're down with programmable controllers, you can whip up a very quick servo signal generator using an ADC. The upside to this is that you can also include a slew of other features along with it - variable acceleration curves, battery monitoring, built-in datalogging, whatever. The most flexible option. Take a reading, scale it, and spit out a 1.5-2.0 (or 1.0-2.0, depending on your motor controller's tastes!) millisecond pulse every 20 milliseconds.
2. Commercial servo tester
. These usually appear as boxes with buttons and knobs. If you want to operate an EV with it, you want the most drop-dead simplest type - a box with a knob that outputs a 1.0-2.0ms signal. You can then rig this to your own array of levers, springs, and whatnot to make a trottle. Most of these are even 5K potentiometer based, and so you can even rig your old bike / scooter throttle into the circuit.
Here's one example of an ultra-premium one
for ultra-premium luxury servos with Corinthian leather. You do not want this.
This is a bone-stock basic one
very much suited for throttle conversion.
The downside? Fixed, linear throttle curves and the requirement of mechanical trims. If your throttle neutral isn't the controller's neutral, you're out of luck unless you can physically adjust the throttle's travel limits! If you're a good e-hacker, you could twiddle some components and allow for adjustment, but if you're going that far, might as well go with option 1.
3. Brute-force DIY servo pulse generator circuit
out of discrete components. This is the route I took, and I will never do it again.
At the time, I had no programming facilities, but did have piles and piles of components. So I decided to just get some perfboard and make a pulse generator from a 555 timer, which is pretty much the most common IC around.
There are many circuits online for simple servo pulse generators using a 555. Some use two (or one 556) to get a more stable signal. The simplest I found (and the one I ended up using) is here
. 7 parts. You could probably make it hard-connection-to-hard-connection and wrap it up in a ball of duct tape.
I got a little more fancy, however, and laid all the parts out on two square perfboards (I didn't have one big one!) and also dropped in a 12 volt DC-DC regulator for future expansion accessories. You could mount this regulator elsewhere or not even use one at all, but I put one on just in case (And to run the underglows which never materialized)
Downsides? Although it's electronically trimmable (by putting a small trimpot on one leg of the main throttle pot), the components will be subject to temperature drift. Neutral in my room wasn't neutral outside, nor neutral in Boston. This is especially true if you are not using precision components. It's complicated, there's more parts to fail and solder joints to mess up. Don't do it (Unless you love putting yourself through pain).
I would really love to see an R/C type controller that can take a straight 1-4v input beucase the their power densities greatly exceed industrial controllers of the same type. How complicated can it be? Most of the airplane controllers I see have fat Atmel ATMEGA32 chips running them, which have onboard ADCs!
Someone want to hack one?