Dirt-E Bike

Kawasaki KE-175 Conversion - Part 4 ...

(Homebrew Meters, Motor Rewinding, Brush-Plate Rework, & GPS Testing)

Section #3 covered paint, assembly and load/no-load testing of the stock MY1020Z 750W motor at 72V.

This section will try to address the heat issues that were identified with the stock motor powered by the BFx Controller (from section #2).

Given the length of this page, the current state of the bike is shown here for those with slow connections, or those that only have a passing interest in this project.

Homebrew  Analog Amp Meters

As mentioned on previous pages, I've been chewing through digital meters like crazy, 4 (four) in the last year alone.

This is a cheap ($11.00) meter with a 10Amp scale that has been wired in line with the positive lead of the battery pack.

Although the image is a bit dark, I wanted to highlight a custom set of test leads that are heavier than the stock meter's wiring and crimped/soldered such that the meter can be inserted for "Hands Free" operation. 

The limitation of a 10Amp scale with a 100Amp controller is addressed below where a pair of custom analog Amp meters are made out of some AWG#14 wire, some old speaker wire, a plastic tub out of our recycle bin and some crimp connectors.

There are numerous excellent tutorials on the web on how to calculate and make virtually any Amperage or sensitivity of analog Amp meter. But the one I used as my guide of the math was Tony Van Roon's page linked below...

http://www.uoguelph.ca/~antoon/gadgets/shunts/shunts.html

Mr. Van Roon's site offers a tremendous amount of information, from electronics tutorials, through hundreds of circuit diagrams with very comprehensive details on assembly through function.

As well as considerable info and guidance by members of the Endless Sphere Electric Vehicle forum (a sincere "Thank-you" to R. Fechter for taking the time to answer successively less dense questions both on the forum and via Email, TD, V-Ice, Mcyc, CraigU & others for clarifying motor re-wind issues on the forum).

These are the materials I started with...

Using the math from the link above, I determined the length of AWG#14 required to deflect the 1ma analog meters full scale for a 40Amp draw through the shunt (formed by the length of the AWG#14).

The wire is stripped from it's sheathing, and cut at 16 1/4".

The wire is then coiled onto a either a pencil, screw driver, what ever will allow it to take-up a smaller space for mounting in the plastic tubs.

Two lengths of AWG#8 are crimped and soldered securely to the shunt wire with ring connectors again to allow the meters to be wired in circuit for hands free operation.

Given that the current flows through a circuit the digital meter which is essentially the reference for calibration is in-line with the meter that is being calibrated.

The Dirt-E bike is run at a steady 10Amps via the throttle and rear brake to load the motor enough that it draws the required current to calibrate.

My understanding is that the meters should behave in a linear fashion, so a 1/4 scale deflection should still be reasonably accurate as it approaches the upper 30 to 40Amp region of the scale.

If the meter was calibrated at say 2 to 5Amps, I would have concerns that any error would be multiplied given that it is such a small reading compared to the full scale.

Ideally I would have shelled out the extra $5.oo for a digital meter that reads a 20Amp scale, but there were none to had in the Town of Renfrew.

This image offers a better view of how the meter is calibrated.

The test leads (red & black) are the sense wires that are easily positioned along the coils until the Digital meter and the analog meter read the same values.

Once set, an ink mark is placed on both sides of each sense wire carefully leaving about an 1/8" for the permanent sense wires to be soldered to the shunt. 

The meter, shunt and leads are assembled into the plastic tub, and hot glued in place.

The lid is popped onto the tub and it's re-tested one last time, then ready for use.

The meters are placed in line with the source (the battery pack), and in line with the load (the motor).

Intuitively I expected that the meters would read identically in lock-step given the serial nature of the circuit.

But they DO NOT...

The motor reads consistently higher current draw than the battery, by a factor of 2 or 3 depending how aggressively the throttle is operated, and the instantaneous speed of the motor at the moment of throttle application.

No this is not an over-unity/perpetual motion/zero-point energy discovery, but rather an anomaly of the PWM (Pulse width Modulation) effect on the motor and (I think) the nature of the Back EMF generated.

Once the motor has achieved it's maximum RPM for the given level of throttle, the 2 (two) meters read near identical current flow.

As an aside, given that the meters do offer a fairly crude scale, it's hard to say how efficient the BFx Controller is at transferring energy to the motor, but given that it exhibits 0 (zero) temp gain on the heat sink it appears that it's quite reasonable.

Through the entire month of December (2007) this image could have been taken on any day but say 5 or 6...

The test road that I'd plowed kept getting blown in, and eventually it was abandoned as I was spending so much time just trying to keep our driveway open, and access to the barn & foundry. 

This image was taken after I had just busted the loader on the tractor and was faced with some hard decisions (none of which was going to be inexpensive).

Motor Rewinding - A practical approach to understanding the theory

I want to be clear, nowhere on theworkshop.ca do I claim to be an expert or even marginally knowledgeable about anything, beyond a healthy curiosity. (please read the disclaimer)

In the course of testing the motor at 72V at various loads I started to make Temp over Time charts that graphed the efficiency of the motor (in a rough way, as energy not used in the rotation of the motor would be lost as heat).

The data that I accumulated indicated to me that the motor was grossly inefficient (unless I was aiming for a way to heat the shop, with wheel rotation as a by product.)

As the motor in question is rated at 36V, 27 Amps and 2800 RPM, just flipping the terms to 72V at 13Amps won't work, the motor RPM is pretty darn high.

For reference, the numbers listed below supersede any previous numbers posted either on theworkshop.ca or Endless Sphere forums (dated Jan 11/2007)

There is a definite trend that shows higher RPM as the Voltage is increased. By extension it is fair to say that the heat losses are rising as the higher voltage is encountering the same resistance of the armature coils... The general rule is called I2R losses where I (Current) is squared times the R (resistance), more voltage equals more current, against a static resistance has to equal greater losses as a result of heat.

The Temp/Time charts supported this theory as well.

Remember the snowy picture above, it's not like I'm losing a lot of riding time, so I may was well try and figure out a strategy to deal with making the cheap MY1020 perform like a motor that typically would be 3 or 4 times the cost. Also I've found that there is a gap in the market between the low-end MY-1020 (1 - 1 1/2 HP peak) range and the ETEK 8hp - 15hp peak.

With a poorly defined objective in mind I decided to tweak the only variable on the motor that is available, the armature windings and see where it leads.

Chronologically we are right in the middle of the Xmas/New Years holiday season, so I couldn't purchase magnetic wire if I wanted to.

This motor is from theworkshop.ca furnace and was having trouble starting-up so it will be gutted to rewind the MY1020. 

This is a miserable job, and I only endured it in preparation for the Post Apocalyptic Era that is on the horizon.

The wire is carefully unwound (avoid nicks or cuts in the enamel coating that isolates each coil winding from it's neighbor).

As this wire is removed it is being spooled up loosely on a length of wood to keep the operation from kinking or snagging into knots that would require that the wire be cut.

This is the factory wound MY1020 armature.

The Blue globs are to balance the weight of the coils.

Each slot is lined with a heavy black fibre material and the ends of the slots are closed with narrow lengths of fiber glass to keep the coils from lifting due to centrifugal force.

The coils are made up of 14 turns of AWG#20 0.032" dia wire.

Each coil is electrically wired to the commutator by the small tabs shown to the left.

The tabs are CAREFULLY lifted to allow the wire to be removed.

If a tab is broken I think the armature is garbage, as the coils can't be soldered reliably to the commutator.

I have to admit I was intimidated by the prospect of tracking tabs, slots, turns, the direction of the windings etc... But once I started to un-wind the factory motor it did turn out to be very simple.

Have a pen and paper handy and make sure you have detailed notes and some simple diagrams to guide you in the re-wind.

I thought it best to mark the start/end tab and the associate slots that the first coil is wound into with a dark marker.

Don't worry about the ink, it can be sanded away with a light swipe of 400 grit sand paper.

This is the last Coil to be removed, and an important image for me, as it will be the first coil to be wound with the new wire.

Keeping with the basic idea that I want to lower the resistance of the coils to reduce the I2R losses that result from the higher Voltage, I assumed that heavier gauge wire would lower the effective resistance of the coils.

Just by chance the Wire recovered from the furnace motor is AWG#17 0.045" dia and does have a predictably lower resistance as a result of the greater cross sectional area of copper for the current to flow through.

But given that the armature is fixed in it's size the number of turns that can be wound per coil is limited. I calculated the total cross sectional area of the coils (14t of #20) and found that 8 turns of #17 would fit.

While winding the coils, as I came to the end of a coil, I'd hook the wire into the tab, remove it and file the enamel coating CAREFULLY to ensure that a secure connection is made.

The bare wire has to be small as the coil leads are overlapping as they connect to the commutator.

The leads should be positioned so that they don't physically touch, but keeping the enamel coating as close to the tabs is ideal as coils and leads will shift as you wind the rest of the armature.

As the coils start to build up in the slots, I had no choice but to offer a bit of encouragement to keep them in place so that the fiber glass strips could be inserted.

This is a very light mallet with a brass face, and very light taps are being applied to work the coils into place.

Similarly the coil ends near the front bearing and at the commutator are carefully tapped to keep the assembly neat.

I couldn't tell you if this is a neat or messy rewind as it's the first I've ever done...

Once it's re-assembled and tested I'll see if it's a success.

Oh, it's a success given that there are no shorted windings and the motor runs like a banshee.

Generally I prefer to avoid subjective metrics of operation like "Banshee" but the motor drives the rear wheel of the Dirt-E so fast that everything is shaking and I have concerns that the armature itself will start to come apart. I just don't have the nerve to bring the motor up to top speed even with the 20lb or so load of the chain and rear wheel for more than a second or so.

Given this is the case, there is no point trying to read the armature speed with the mechanical tach as it would have 0 (zero) load and have to run for 60 sec at least 3 times.

I posted these results on the ES Forum and had answers in a day or so...

The number of windings is inversely  proportional to the speed of the armature, more windings equals lower speed and higher torque, fewer windings equals higher speed and lower torque.

Given that the R of the coils has been dropped so heavily by the combination of fewer turns AND heavier wire, there is also tremendous current flow for the no load condition.

Essentially I went in the wrong direction with this rewind.

I need high torque and low speed (as noted on a previous page, I'm happy to putt along at 20Km/h as long as I can climb any hills on the farm).

Once the holidays were over, I found that the local motor repair shop would sell me wire by the Kilogram (approx $25/kg)...

So I purchased 2 Kg of AWG#18 and 2 KG of #19 wire. 

This image is of 10 turns per coil of #18 0.040" dia wire.

The goop shown is body fill to help balance the armature.

Again I had similar results, the top end is still too high...

This is the 3rd re-wind of a different but identical MY1020 motor with AWG #19 0.036" dia at 13 turns per coil.

I had suspected that there was a bit of unused space in the slots of the armature on the previous re-winds, so added a single turn above the calculated 12 turns that was equivalent to the 14 turns of #20.

I had concerns that the previous motor would start to drop the commutator tabs if I tried to lift them one more time.

All three re-winds had to have each segment of the commutator checked for shorted coils and poor or open connections.

I never found any shorts, luckily as that would entail starting from scratch.

But every re wind had 2 or more poor connections and one coil leg that I forgot to remove the enamel from.

Well after all that work into the BFx controller, I had to drop the voltage back to 48V just so that I could get some accurate no load RPM readings to prove-out the theory concerning the number of turns of the armature winding.

The chart above summarizes almost 2 weeks of work going back to just about the last week of December 2007.

The #17 rewind was disassembled and rewound as the #18 motor, and has been put aside for the recumbent trike that is next on my list of projects, and likely will be a high current 36V implementation.

Per the responses from the ES forums it would appear that there is a limitation of how much power you can squeeze out of a motor relative to the mass of steel laminates that make up the physical armature.

The image on the right illustrates the difference in size between the 750W and 1200W (stock) MY1020 models.

I should note that he wire on the 1200W (48V) motor is lighter than the stock 750W 36V motor, as well as being 1/2" longer.

Brush-Plate Modification

Along side of all this winding, testing, posting questions and reading replies (not to mention having to make the effort to understand concepts that were counter to what I expected), I wanted to address the frail execution of the brush-plate assembly given that the motors were being called upon to deliver more than their specified design.

I had to drill the tops of the cast posts that secured the brush plate to get it free.

Before I cut any wires I applied a test blob of solder to ensure that I could make an electrically secure connection to the brush housing(s).

A witness mark and notes document +ve and -ve brushes as well as orientation to the rear motor casting.

All wiring is removed as well as the brush springs to make work easier.

The original wire was AWG# 16 stranded and seemed a bit on the light side.

The replacement is consistent with the Brute-Force Eng ideology that guided the BFx controller upgrade.

AWG #14 solid core paired and soldered.

This is a tedious job and required a measure of effort to ensure that the wires were accurately measured/formed and installed so as not to preclude re-assembly.

All things considered I think that I've done no harm...

The minor burn marks on the fiber glass board are from the soldering of the twin leads.

The brush plate is re-assembled into the rear casting and secured by a pair of self tapping sheet metal screws.

The temp sensor is re-located from the front of the motor (where it was measuring the exhaust temp from the rear blower) directly onto the brush plate itself.

The last step prior to testing each re-wind was to set the motor to it's optimal timing.

There are 4 to 5 degrees of brush-plate rotation between the magnet segments inside the motor.

With the digital Amp meter, a 12 V battery is connected to the motor, the rear casting is carefully rotated until the motor is running at either it's highest RPM or it's lowest current draw.

I opted for the meter just to be sure, but the variance in RPM is so pronounced I could have done it by ear.

So with my new found knowledge that I've proofed beyond a doubt in my own mind, I've accomplished very little with regard to the idea of beefing-up a 750W MY1020 specifically for the Dirt-E Bike project.

To be fair though All the testing was under No-load, and I really need to run the bike outside over a range of terrains and get a feel for it's torque/speed combination.

The motor that is presently mounted is the AWG#19 @ 13t per coil.

The BFx controller is still on the bike, but has been hobbled to a fraction of it's potential with the 48V 13Amp limiting. I have no regrets on the BFx design and build as it will definitely move on to more demanding applications in the spring.

As all of the above was going on I was simultaneously looking for a solution to my tractor issues (which go far beyond just a broken loader cylinder.)

The resolution is this John Deere 1050 4WD turbo Diesel.

If I wasn't committed to a monogamous relationship with my wife, I think that this would be my mistress...

I have to privately admit my heart is fluttering right now just looking at this picture.

I don't think that I have to point out that the weather finally broke and we had an early January thaw. This happens most years and knocks down the snow for a few days before we hunker down for another 3 months of winter with a deep freeze in February.

Beyond the travel to look-over the tractor, and then another full day to haul her back on a trailer through the high country (230Km one way), I finally got a chance to run a full charge of the batteries through the rewound Dirt-E Bike and see how it hauls.

GPS Data Collection

This section details the idea of trying to accumulate some more meaningful data about the Dirt-E bikes performance than vague descriptors like "Gee it's swell" or the ever popular "It must be fun, your smiling".

To that end, given that I despise the magnetic wheel mount bicycle computers that I've been using to date, I wanted to graduate to the world of a GPS unit.

It took considerable research, but I settled on a Garmin ForeRunner 201. It doesn't offer turn by turn voice instruction or any meaningful map info downloadable off the web, but is better suited for this task as it's designed to evaluate the performance of runners or cyclists over a set course.

The FR201 has an interface to a PC and can down load it's data for analysis and comparison between one lap and another.

The unit is light, durable and water proof to a reasonable degree.

It's portability is such that I've started to develop my own custom map of every where that I go relative to our home.

Perhaps I'm just slow, but there is a tremendous amount of intellectual overhead required to overcome if I want to get the most out of this unit.

So far I've got it interfaced to the PC, I can mark my Way points, start and stop laps and save sessions back to the PC.

Bear in mind the closest I've ever come to a GPS is one that a friend let me hold (but not turn on) 2 years ago.

The image below is a screen capture of the speed profile of a 1Km run on the Dirt-E.

As this is an athletic trainer, there is provisions for entering the total weight of the device under test including rider (in this case the total Dirt-E and me) and the software calculates the energy expended over the course of the lap.

I can appreciate that there may be inaccuracies due to the algorithm that the caloric count is based on ie; 400lb man running at 20km/h may not equate to 400lbs of Dirt-E at 20km/h... But it would still offer an objective metric that would point to either improvements or losses of performance over a static course.

This graph of the elevation drop and rise is so cool even just to look at, but given that there is an XML export feature, my hope is that I can co-relate the speed profile to the elevation profile and make some extended calculations on torque and EV performance.

This image is of the course that I ran the bike over for the 1Km run.

Once the spring hits I'll establish a standard course that covers several Km's and a variety of terrain. and that will be the bench mark that all EV related projects will be tested against.

So in closing, the BFx controller is really over kill for a little dirt-bike, and I'll return to one of the factory Yi-Yun controllers, most likely the LB-37 48V 50Amp unit just because it supports the brake disconnect and charger port.

The additional batteries will be removed as they did make the bike heavier and less stable, as well as making for a more cramped riding position.

I have a major GPS learning curve ahead and will have to find an XML reader/editor that will easily allow me to do more meaningful analysis of the GPS data.

As an aside the Brush plate temps never went beyond 120F (the limit of the sensor) over the 8 or 9 Km that the bike was run, though the temps were just above 0C outside, every thing is tuned down to 13Amps, and there are only baby hills close to the house the big hills are further out, and I don't trust the bike enough to venture too far in case I have to push it back.

Dirt-E Bike (KE-175 Conversion), 2, 3, 4

Back to Energy

Disclaimer (an unfortunate necessity)
All Rights Reserved theworkshop.ca © June 03, 2008