Design Evolution

Design Considerations

This section is not about any one specific change. It's about my process of designing electronic clutch enhancements.

Even though it's possible to entirely eliminate the clutch with an electric motor, this “advantage” becomes a drawback when you get back on a normal motorcycle. Maintaining the standard motorcycle control methodology is of paramount importance to me.

The problem of designing a Progressive Electronic Clutch comes down to optimizing four considerations:

  • Cost

  • Feel

  • Ease of designing/fabricating the mechanism

  • Ease of designing/fabricating the electronics

I guess there's also a fifth consideration – weight/physical size. But since this will be a prototype, I did not actively deal with that initially. But it's always in the back of my mind.

A Bit of Theory

The theory behind the electronic clutch is simple, it should behave just like a normal clutch. That is, it can only diminish the torque requested by the throttle, not increase it. Mathematically, think of the electronic clutch as a function that multiplies the throttle's torque requested by a number between 0 and 1. With the lever pulled to the bar, the multiplier is 0, and no throttle setting is allowed to pass. With the lever fully released, the multiplier is 1 and the throttle is in complete control. Anywhere in between, the throttle request is linearly modulated by lever position. This can produce a very fine level of control – perfect for the precise application of torque at slow speeds.

Where things get a little uncertain is what exactly happens when the clutch is “popped.” This will depend on the behavior of the motor controller. I know that with my simple binary modification to the original 5.7 electronic clutch lever, I could not really get much of a “pop” out of it. However, there are some programmable “smoothing parameters” in the controller setup that may benefit from tweaking. I have not yet experimented with these.

An electronic clutch lever has the advantage of begin able to actually modulate the power output down to zero. This is not always the case with a mechanical trials clutch as their behavior is often temperature dependent. Even in the best of situations they can exhibit creep and drag. Furthermore, it would even be possible to tailor the action of an electronic clutch to a particular rider's preference with the appropriate signal-processing electronics (more on that later). However, with analog electronics, it's easiest to provide completely linear action – and I expect that's what I would prefer anyway.

Prior Art

Prior to commencing design work, I like to see what already exists. A search uncovered a plug-and-play solution proposed in a 2017 YouTube video by a guy in the Czech Republic.

Apparently, the price was $500, shipped. However, an inquiry sent to the email address in the video went unanswered.

Screen capture of electronic clutch lever shown in video

Signal Conditioning Electronics

The 5.7's motor controller provides 5 VDC at a maximum of 40 mA for external uses. This voltage source provides the reference signal for the throttle. The stock Mode Switchbox (Novice, Trek, Trial) controls the maximum command voltage that can be produced by the throttle. As stated in a prior section, the Dry Trial mode produces the largest command voltage (most torque) and is set to approximately 2.66 volts. Interestingly, the Kelly controller is factory-configured to accept a maximum command voltage of 3.0 volts, so there's room to increase the maximum torque. But I'll leave that alone until it proves to be a limitation. Of course, zero volts equals zero torque.

There are two approaches to building signal-conditioning electronics: analog (based on one or more op-amps) and digital (based on a microcontroller). Analog offers the possibility of the simplest solution (assuming the mechanism can be designed to accommodate its limitations). Any solution utilizing a microcontroller is complicated by firmware (writing, testing, and possible bugs) however it does offer a fairly straightforward way to do complex manipulations that would be burdensome to achieve with analog electronics.

Since only a unipolar (+5V) power source is available, I wanted to avoid any solution that required the analog circuitry to “invert” a signal. (Although a DC/DC converter or virtual ground could solve those problems, albeit with added complexity.) As mentioned previously, this means that the behavior of the clutch mechanism needs to be consistent with that requirement.

Using a potentiometer as a voltage divider is the only way I could see to achieve the desired simplicity. (In fact, my initial design idea was to use a very nice aerospace-quality potentiometric pressure transducer I owned that was collecting dust. But its range was limited to 0 - 30 PSIG and used an AN fitting, both of which complicated the hydraulics.)

Because a potentiometer is symmetrical (either end can be connected to the reference voltage), it can be configured so that the wiper voltage can either increase or decrease as it's moved through the range of motion.

But, it's also imperative for simplicity that the full range of the potentiometer can be used, and this leads to some mechanical constraints.

AJP master cylinder from A Sherco 200 with pressure gauge

Hydraulic Pressure / Lever Feel

The adjacent photo shows a test to see what typical hydraulic pressures “felt like” in a clutch lever.

It's possible to generate 200 psi with a single finger. There is some “give” in the system which I attribute to the spring in the mechanical pressure gauge. I suspect there would be no “give” at all with an electronic pressure transducer and some sort of hydraulic accumulator would be needed.

I once bought an MX bike that had an aftermarket “antilock” brake system attached to the rear.

In negotiating the purchase price, I was happy for the previous owner to keep that device.

I don't recall the manufacturer, but a Chinese knockoff is available today on eBay for around $10. This may be a method to give the lever some “feel.”

[ Not really - see the Regn Experiment in the Pure Race section for more info on this device. ]

Chinese knockoff of a mechanical ABS unit

Pressure Transducer

A zero to 300 psi Chinese-made pressure transducer may be purchased remarkably cheaply (about $14) via eBay.

However, a pressure transducer complicates the electronics in two ways. Firstly, there's an offset at the low-pressure end of the scale. When powered from a 5V source, it emits 0.5 volts for zero psi (and 4.5 volts for 300 psi.) Secondly, it's the opposite of what I need in that pulling a lever to the bar produces a higher pressure and therefore a higher voltage.

The AJP clutch master cylinder shown above is from Cindy's Sherco 200. It has an M10 - 1.0 thread for the banjo bolt. The problem with a generic left-side master cylinder is obtaining a fabulous lever. An AJP trials lever would probably be acceptable.

Although a standard trials hydraulic master cylinder and lever is okay, we can do better – especially if you want to extend the range of modulation. Ideally, you want to be able to pull the lever to the bar without trapping other fingers.

Clever Lever

I originally designed the setup shown in the adjacent photo for my OSSA, but it would not completely disengage the clutch. I ended up installing it on Cindy's Sherco 200 where it became her gold standard for clutch control.

An eBay Chinese Magura clone master cylinder that uses mineral oil is $54 delivered from within the USA. This would allow Midwest Moutain Engineering's Clever Lever ($83) to be used.

The Magura master cylinder exhibits about 9 mm of piston travel.

However, a potentiometer is preferable to a pressure transducer because this simplifies analog electronics.

Magura master cylinder with MME Clever Lever

Yamaha servomotor as a cable-operated rotary potentiometer

Rotary Potentiometers

This next concept comes by way of Yamaha's YPVS (and later EXUP) servomotor actuator. I've shown it with a simple “return spring” to give the lever something to work against.

The Yamaha servomotor has gears inside. Partly this is to get increased torque out of the servomotor. But partly it allows more travel (and ultimately resolution) from its internal rotary feedback pot. The pot itself has no mechanical stops (probably as a failsafe if the push-pull cables are installed poorly/improperly). Electrically, the rotary pot would be easy to implement (and I don't have to butcher any hydraulic master cylinders). But it would not be a trivial task to manufacture a good one. Positioning it on the bike is not really a problem, as there's plenty of room above the battery.

Unfortunately, the useful range of the potentiometer is not sufficient for my application (which complicates the electronics).

Bicycle Cable-Lever

I figured that if a cable can operate a valve, a bicycle lever could operate a rotary potentiometer. Furthermore, a bicycle lever is a reasonable choice in that it's much shorter than a motorcycle clutch lever (won't trap other fingers) and we don't need to exert much force.

The next photos show a mock-up of the rotary pot with an Avid “Speed Dial” lever for a bicycle. This lever provides an adjustable leverage ratio. At one extreme, it provides 1.25 inches of cable travel. At the other extreme, it provides 1 inch (which is more than enough).

However, a low-friction mechanical implementation is difficult to achieve with the pot's shaft supported at only one end. The correct torsion spring is also very important for feel (and not easy to change in this mockup without complete disassembly). The cable itself is a source of friction. There's a reason hydraulic clutches are popular.

One disadvantage of a bicycle lever is that it can't be installed/removed with the handgrip in place.

Cable-operated rotary potentiometer #1

Cable-operated rotary potentiometer #2

Avid Speed-Dial bicycle lever/cable for control of a rotary pot

Credit: eBay

Hydraulics to Linear Motion

It would also be possible to use hydraulics to move a linear-motion pot. Shown adjacent is an aftermarket part available to convert a cable-operated clutch system to hydraulic actuation.

These are available on eBay very inexpensively. This would allow the use of any master cylinder and any lever, but the system would be bulkier.

Linear Potentiometers

Any of the following potentiometers could, in principle, be integrated directly into a clutch lever mounted on the handlebars.

Below is a photo of a Bourns 3048L linear-motion potentiometer. It costs about $40 in small quantities and measures 0.326" tall, 0.275" wide, with a 2.22" overall length. It has two mounting holes. The shaft has a #2-56 thread. They are available with electrical travels of 0.2, 0.3, 0.4, and 0.5 inches. They all appear to have the same mechanical travel (0.5 inches) and are rated for 500,000 cycles.

A smaller, less expensive, version is the Bourns 3046L. They cost $14 each, have a #0-80 threaded shaft, and are 0.259" tall, by 0.244" wide with a 1.591" overall length. Unfortunately, there are no mounting holes, so it would need to be epoxied in place. Only fairly short electrical travels of 0.15, 0.25, and 0.35 inches are available.

BEI makes a linear motion pot (9605R1.7KL2.0) that's probably a better choice, but not as readily available as the Bourns parts. It is only available with a resistance of 1.7k ohms (which would not be a problem). They are rated for 1 million cycles. The cost is $39 in small quantities and it exhibits 1/2" travel. The housing is only a bit over 1" long. Unfortunately, there are no mounting screw holes.

Credit: Radwell International, Bourns 3048L

Credit: BEI Sensors, 9600-series spring return linear position sensor

Closer to a Solution

Shown below is a 3-wire Chinese device (KTR-10MM) intended for measuring in machine shop environments. It is a 1k-ohm linear potentiometer that costs $18, delivered. It may be a copy of a German product. I've photographed it next to a wrench to give a size reference. It's quite large compared to the Bourns and BEI pots, which is an advantage for prototyping but would be a disadvantage for a permanent installation.

It supposedly has 10mm of electrical travel but measures more like 15mm. The physical travel without damage is even greater. Unfortunately, it would not pull down to zero ohms (which corresponds to zero volts when configured as a voltage divider) as received.

Luckily, it was easy to disassemble the unit and reverse the shaft. Notice the difference in cable orientation (relative to the longer shaft) with the next photo. Now, with the shaft reversed, it pulls to zero and still has mechanical travel to spare. The spring provides a reasonable lever resistance and coil-binds when the clutch lever is fully pulled in, thus protecting the electrical portion from damage.

KTR-10MM linear potentiometer

Bench test of operation with the Avid bicycle lever.

Avid bicycle lever shown with Magura throttle

Visualizing Mechanical / Electrical Interaction

You can get a reasonable idea as to the electrical behavior / physical action of an electronic clutch system by watching an analog voltmeter, but a digital oscilloscope set to roll mode is better.

The next photo shows just the clutch lever operating between 0 and 5 volts. You can see a full-scale transition in about 0.1 seconds. I repeated this experiment with a prototype system comprising the throttle, clutch lever, and some very simple signal condition electronics. I liked what I saw and packaged the unit for testing prior to the end of our trials season.

Behavior of cable operated linear potentiometer with clutch finger

Hall Effect Sensors

I was surprised to learn that EM's Progressive Electronic Lever Switch (PELS) uses a Hall effect sensor.

Although linear Hall effect sensors certainly do exist, the simple on/off switch type is far more common. Fortunately, I'm well acquainted with linear Hall effect sensors and have used them as a “zero cost” gaussmeter for undemanding applications.

The advantage of the Hall effect is that the sensors are small, inexpensive, and non-contact. But any lever designed around them would be forced to deal with their inherently nonlinear nature since magnetic field strength falls off as an inverse-square function.

UGN3503 linear Hall effect sensor on stick

Did EM Have the Same Idea?

I found some interesting photos of what's purported to be the 2012 EM 5.7 on the website. The photos clearly show a cable going into a mode switch-box larger than what was eventually used on the later bikes. The other end of the cable goes to a bicycle lever on the clutch-side handlebar. It appears to be a mechanical cable. Having a mechanical cable pulling on a simple on/off switch makes less sense than pulling on a potentiometer. So, it's possible EM originally came to exactly the same solution I did.

Of course, that begs the question why did EM not take it into production? The cost would have been similar, if not lower, for a better product.