It keeps getting better

Covering some real ground

With the full battery pack and a instrumentation package up and running, we are now going for longer drives.  Yesterday, we covered over 10 miles on a single charge.

The charging system is working great.  Two charges, both tied to the 220V input.  One is the 1500W 91V Lithium Ion charger for the drive battery and the other a 120W 12V charger for the system battery.  Since both are automatic, they shutoff independently when charging is complete.  I have the 220V outlet on a timer just to be safe.

Instrumentation Pack

Here is a the roughed-out instrumentation pack.  This will be mounted behind the dashboard once it is installed.

1) On the left are 4 CellLog8M battery monitors.  These are set to alarm if any of the individual cells go out of range.  Currently I have them set tightly at [3.0V - 3.6V].  Charging is set to 91V which should be 3.5V/cell on average, so any major drift will be detected.  The cells have sagged below 3.0V on a recent hill climb, but recovered to 3.2 very quickly once the road leveled off.  

2) The top-right rectangular LCD is one Arduino Uno which is connected via the CAN Bus to the Kelly Controller.  It monitors: 

• Top Row: RPM, Battery Pack Voltage, Controller Temperature
• Bottom Row: Drive/Reverse, Calculated Speed (Assuming 2nd Gear for now), Motor Amps, Motor Temperature

3)  The lower LCD is another Arduino which is measures battery current through a 600A/75mV shunt.  It is power separately from the drive battery pack via a 12V adapter.  This keeps the battery system isolated from the rest of the car's 12V electrical system.  This Arduino is always on and maintains a running count of the amp-hours consumed.   The shunt is installed at the battery negative terminal to eliminate noise from the motor and to allow for monitoring charge going into the battery from either the charger or from the motor's regeneration braking.

In a quick experiment, after a 5 mile drive, I then connected the charger.  The Amp-Hour counter returned to within 3% of full.  Please note, I have not calibrated the system yet, so this is very promising.  The system is using an Analog Devices AD8210 Shunt monitor to measure the voltage drop across the shunt.  It is wired with a split-supply configuration, so 0 Amps = 2.5V going into the A/D on the Arduino.  This setup should yield about 1A precision on counting amp-hours.  Given the 20% DoD (depth of discharge) limit, this should provide plenty of margin.

Controller settings

Now that there is decent current monitoring, we tweaked the Kelly KHB72701 controller settings a bit to see if there is a bit more power available.  During the drives yesterday, The battery current never exceeded 200A and the motor current peaked at about 300A.  The spec on the motor is a max of 420A for 1 minute and the batteries should be capable of 400A-600A for short bursts since they are 60AH packs and there are several reports of people safely pulling 8C-10C.  Batteries are rated in terms of multiples of the AH measurement, so a 8C battery can deliver 8x60AH for a short burst. 

With motor blower providing plenty of cooling, he motor's temp peaked about 55C.

New Settings:  

• Throttle Range 5%-80% (this is a mapping to the mechanical throttle position).
• Max Motor Current 90% (this is of the controller max of 700A --> 455A)
• Max Battery Current 100% (700A)
• Undervoltage (Controller will start to cut back power draw at 110% of this value): 68V (cut-back at 74.8V, or 2.87V)
• Overvoltage (Controller will limit regeneration if voltage exceeds this): 90V
• Throttle Ramp: 4 (How fast it will ramp up acceleration)
• Regeneration: Off 

Off for another test drive to the grocery store....

Starting to see some encouraging signs

First Extended Drive

Today, we headed out on a extended drive.  Not knowing the real efficiency of the car (miles/charge), it is going to take some testing to become comfortable with the nominal range.  The Amp-Hour meter that is currently connected does not retain its values after a power cycle, so it is not a great fuel gauge.  More on that later.....

It was a sunny day here, about 65 degrees, so a perfect day for test drives.  Earlier in the week, I did a 4 mile run around the flat part of the neighborhood and the batteries only required about 30 minutes to recharge fully.  Today, we drove for 8 miles with some hills.  The batteries still read about 85V but since the drop off near the end of the charge is fast, I am hesitant to let them get too low.  Well after a 7.5 mile drive, they took about an hour with the 16A charger to return to 91V (the current target charge level).  This was the previous charge level so it should represent a reasonable replacement energy measurement.

Based on these estimates, the car averaged about 180 Watt-Hours/mile.  Right in the expected range of 150-200.  With a conservative 80% usable capacity, this should equate to 22 miles of range per charge.  A GPS was used to track distance, but a more accurate charge timer is needed to estimate the actual Amp-Hours that are being put into the batteries, or taken out with the a better amp-hour meter.

Here is the boot with the batteries in position.  Not a lot of extra room  The brown and blue wires are the battery monitors which make sure that none of the cells deviate too much.  The next step will be to route these up to the dashboard so the driver can monitor the battery condition.  The large blue connector is the charger connection.  The charger will be mounted to the left of the pack.  The large (red) 12 battery is powering all of the electronics in the car (lights, radio, controller, fans)

Cooling

The motor coil temps, as reported over the CAN bus from the Kelly controller is running between 80C-90C after a couple of miles.  This is well below the limit of 145C but still on the high side.  The internal fan on the motor has two issues:

1. It does move a lot of air - @2,000 RPM, I could barely feel any air leaving the motor.
2. It does not cool when the motor is not spinning. 

A supplemental fan is needed and I found a 12V marine exhaust fan that moves a lot of air.  The challenge is how to duct the fan into the motor in the tiny Mini engine compartment.  I tried a Rube-Goldberg solution today using 4" flexible ducts but there is just not enough room at the end of the motor (where the internal fan draws air) to route the ducts.

The Plan:  Bring in 1" hoses (probably two) into the fan housing on the motor, from the top to push more air through the motor.    The challenge: How to plumb the 4" fan down to a couple of 1" hoses through a manifold.  The air is moved from one end of the motor to the other, horizontally.  There are large vents in the housing on the shaft end, so if enough air can be pushed into the fan end, it should push on through, providing much better cooling.

First test of the Full Battery Pack

We headed out around the block with the new 84V 60AH battery pack.  Starting out in 2nd gear to keep things gentle, Jane climbed our hill with little hesitation.  Next we headed for the short flat stretch of street around the corner and in third gear, she got right up to speed.  Here is a video of us returning home.   You will see the pack mounted in the boot along with the battery monitor leads.

Even under the full hill climb, the pack only sagged to 80V, still close to 3.1V per cell.

Next - Clean up.  Lots of cables and mountings to sort through.   However, it is a great motivator to be able to see it all coming together.  Also, a quick run down the to the non-hilly part of the neighborhood for a real speed test.

Battery Time

The batteries arrived and this opens an entirely new chapter in design and learning.  Not only is there the mechanical design elements of assembling and installing the battery pack, there are the safety challenges. Compared to gasoline, I believe batteries are safer and easier to work with, but society's false sense of security with gasoline causes some initial concerns.

The batteries of choice for Jane are the CALB CA60FI.  These were introduced last year and the reviews are all very positive.  These were sourced from Shift-EV in Albany Oregon.  Kirk is a true expert on EV conversions.





The chemistry is LiFePO4 (Lithium Iron Phosphate) and these offer 60AH  at 3.2V nominal.  The pack will consist of 26 of these in series providing a nominal pack voltage of 83V.  At a 80% depth of discharge assumption, these cells will provide 48AH @ 83V or just shy of 4 KWH.  The Kelly Controller is limited to 90V input voltage, so  this provide a little margin, but should still be sufficient voltage for the motor to achieve 3,600 RPM under load.   Based on the best gear ratio calculations, in 4th gear, Jane will cruise comfortably at 55 MPH.  3rd Gear will yield 45 MPH.  More than fast enough for this car and any city driving.  There is an option final drive gear (3.44:1 --> 2.76:1) which will pull in another 13 MPH at the electric red-line.

Pack weight=26 x 4.45 lbs = 116 lbs.

If you want to see the detailed calculations, here is a spreadsheet (Called Vehicle Calculations) with the best-guesses for performance.

Charger

Another major component for the battery pack is the charger.  We selected an automatic, adjustable voltage charger from BatterySpace which provides 16A and 1,500 Watts since there is still much debate over the best charging process to maximize longevity of the battery pack.  The first charge cycle completed successfully set at 91V and the charger performed as expected, shutting off when the pack reached terminal voltage.  This is plugged into a 6 hour mechanical timer and a 220V 30A outlet in the garage.  It will get mounted in Jane's boot and eventually a simple cord is all that will have to be plugged in to "fill her up".
This charger will fully charge an 80% depleted pack in about 5 hours.  The CALB cells can take a faster charge, but there are cascading effects from increasing the charger power such as more heat, more battery wear-n-tear, larger cables and bigger power sources.

Battery Monitor

And the final critical element of the battery pack is the monitoring system.  With LiFePO4 batteries, maintaining them within the proper voltage range is critical to prevent damage.  There are many opinions on how best to do this but it is clear that the tighter the range of voltages, the less stress on the pack.  Initially, the system will be set to keep the cells between 3.0V (discharged)  and 3.5V (charging), though others say the window can be slightly larger.  Battery Management Systems range in price up to thousands of dollars, depending of the size and number of cells in the pack.  Since the battery is the most expensive element of the EV drive train, it makes sense to monitor and manage it diligently.

For Jane's initial BMS system, passive cell monitors will be used to report and alarm if a cell goes outside of the target range.  The driver will have to respond appropriately based on the situation and severity of the deviation.  Not terribly different than the check-engine or over-temperature light on a gas car.  A driver can choose to ignore it but damage may occur.


The CellLog 8M is a clever device.  It provides monitoring for up to 8 cells and alarm capability for High Voltage, Low Voltage or if there is too much variation between the cells.  It will take 4 of these to monitor the 26 cells (2x6 and 2x7).  There is an minor annoyance with these devices in that they draw less power from cells 7 and 8, causing those to discharge at a slightly slower rate than the first 6 cells.  Given Jane's configuration, this will only impact 2 cells which may need to be balanced periodically to bring them back in line with the rest of the pack.  The four CellLogs will be mounted in a cluster on the dashboard so the driver can quickly see all 26 cell voltage values if an alarm occurs.

The last component in the monitoring system is the AH meter from Elite Power (600A 90V Combo Meter).  This will act as a fuel gauge and provide a count of the Amp-Hours (AH) drawn from the batteries. Using a 600A 75mv shunt on the negative side of the battery, this meter will count the actual cumulative power (AH) fed into the controller.   With LiFePO4 batteries, voltage is not a good indicator of State-of-Charge (how much energy is still available from the cell), so using AH counting will provide the most accurate fuel gauge.  It will take some verification and experimenting to map AH to overall state-of-charge.    More on that to follow...


Elite Power Meter

Shunt

Finally, the EVALBUM page is up for Jane.  Not all of the details yet, but a great gateway to other electric vehicles.   Be sure to check it out....   http://www.evalbum.com/4768

Instrumentation

A big part of Electric Vehicles is the instrumentation that is possible with a electric motor.  There are some generic systems that can monitor basic functionality of the system and display it on a small LCD, but as with most of the elements of Jane, this too is going to be have to custom built.

The good news is there are now off-shelf electronics available that can be utilized.

Available information

The Kelly KHB72701 controller has an optional CAN bus.  This provides an industry standard bus that allows a computer to communicate with the controller.  The controller has a wide range of data available via the CAN bus.

Available Data includes

1) Voltages: Batteries, controller supply, accelerator sensor

2) Motor Specs: RPM, Motor current and Phase Width Modulator (PWM) value

3) Temperatures: Controller and Motor

4) Switch Settings: Reverse & accelerator micro-switch

Notes: The CAN BUS is internally terminated within the Kelly Controller.

All that is needed is small computer than can read and display this.

Arduino

The Arduino platform is a great development system.  There are a number of controllers available along with accessory boards that mount directly on the controller, allowing for easy to configure HW solution.  Also, many libraries exist to help jump start development.

Useful links:  

Main Arduino Page  - We are using an Arduino Uno

CAN Bus Adapter Board (called a shield)

LCD Shield

These three boards mount together and will the main controller for Jane.

Software

Here are the project specific files for the Sketch that generates the dashboard below.  Google Drive Link

I'll have to dust off my software skills to start work on the C code for the Arduino board.  Luckily it won't take much to have it perform basic dashboard functions.  The key data I would like to display realtime is:

Motor RPM --> which can be used to calculate vehicle speed if the gear is known.

Motor and Controller Temperatures --> Need to make sure nothing is getting too hot

Motor Amps as reported by the controller

Battery Amps as measured by a current shunt

Drive/Reverse - To confirm the electronic Reverse is working

Controller Errors --> Just in case something goes wrong

Here is a picture of the prototype display 

Battery Current Measurement

The real measure of power and efficiency requires accurate battery voltage and current measurements.  The controller does not report battery current so a current is installed on the ground side of the power going to the controller.  Using the negative side helps eliminate common mode voltage since the motor battery is not connect to the 12V battery which powers the controller.  The shunt is a low resistance resistor that will generate 75mV from a current of 600A.

The current plan is to install a Amp-Hour (AH)  & Current meter like this meter from Elite 600A-90V Digital Dual Display.  It will provide instantaneous readings plus it offers some alarm capability.

Power, gears and hills

Now we are rolling along

With CV joints behaving as they should, the real testing is starting.

After dozens of runs up and down the hill outside our house, my confidence is building in Jane.  Playing it safe, I always start out heading uphill first so that if a problem occurs, gravity will get me home.

At this point the three Lead-Acid batteries easily go 2 miles.  Instrumenting the motor is a going to be key. So far there are two gauges: Motor (or Phase) current and Battery Current.   The relationship between these is dependent on motor speed and load, which has to be estimated based on the speedometer and the current gear.

I have found that there is not a single level road nearby.  This actually creates problems since any incline or decline in the street impacts the power needed to move the car.

So far, the motor current (varies dramatically based on controller settings) peaks around 420 amps (60% of the controller rating of 700A).  The Motor current meter is suspect since is jumps to 20% under the lightest load which is equivalent to 140A (20% of the controller's max amp rating).  It seems like 140A should be delivering quite a bit a power.

Battery current has only peaked at 300A which is great to see given the target battery pack is rated at 600A so this will be well below the limit, extending the overall lifetime of the batteries.

3rd Gear looks to be the best in-city gear.  During yesterday's testing, it cruised comfortably at 20 mph with the 36V battery pack.  The target pack is going to be 84V and speed scales linearly with volts, so 3rd gear should yield >40 mph.

It is a real challenge to understand how much power is actually being used.  The battery volts and amps can be multiplied to compute Watts going into the system.  The motor and controller have a combined efficiency of about 75% (conservative) so that should yield power out of the motor.  Given all of the losses in the drive train and air resistance, the predicted speed based on power should be close.  Here is one example of the power model.   From this at 30 kmh (about 19 mph) the motor should be delivering 2500 W (3.35 HP 745 Watts in a HP).  This should equate to about 3,300 W from the battery (@75% system efficiency).  At 34V, I would expect about 100A of battery current.  This comes close to what the meter reads.

Speed (km/h)	10	20	30	40	50	60
(m/s)	2.8	5.6	8.3	11.1	13.9	16.7
ForceOfAirResistance (N)	2.6	10.6	23.8	42.2	66.0	95.0
ForceOfIncline (N)	127	127	127	127	127	127
ForceOfRollingResistance (N)	115	115	115	115	115	115
TotalDrag (N)	245	253	266	284	308	337
Power to Maintain Speed (W)	756	1560	2462	3511	4755	6244

I would welcome feedback on the calculations and logic.  Next is to figure out total range based on the Watt-Hour capacity of the batteries.  The Kelly Controller has a CAN bus available and will report some useful information such as motor phase current and voltage, motor and controller temperatures, motor RPMs and battery voltage.  The current plan is to program an Arduino controller to read this data along with other system data (battery status, battery current and cell voltage).  The Arduino will act as the car's computer and display for the driver key performance data.  A bit of modern technology for a 40 year old car.

Getting to the bottom on the Thump

The Thumping diagnosed

After a great 2 hour session of trying to debug why Jane would thump loudly, lose power and stop.  It seemed like the controller was cutting power to the motor for some unexplained reason.  We ran it for 4 full minutes on jack stands while applying the brakes to simulate a reasonable load.  The motor was working as hard as it does on the local hills The brakes could apply enough to stall the motor but the thumping never occurred on the stands.  The mystery: what was different on the street versus on the stands.

Well today, first thing, I swapped out the 18V power supply that powers the Kelly KHB72701 controller.  Switched it back to a an old AC adapter with an 400W inverter, very Rube Goldberg.  First 6 inches up the drive and the thumping returned, as bad as ever.

So I opened the hood, had Carly gently apply some gas while I watched the drive shafts.  Sure enough, the driver side inner CV joint was slipping.  I could see the inner shaft rotate from the differential while the outer shaft slipped out about an 1/8" and then jumped back.  This was accompanied by the thumping sound.  Aha.

Here is a picture of what the CV joint looks like. 

When the car is elevated and the wheels drop, a few of the balls are pushed into the hub.  However, on level ground, the balls move out to the edge (like the ones on the left above).  When all six are near the edge, the slightest of force will cause all of them to slip out of their groove, only to jump back into the next slot, ready to thump again.  If the joint is at an angle, 1 or more of the balls will be deep in its groove and unable to slip.

As with anything, once you know what is it, it is easy to find information.  The following video shows another Mini having the exact same problem.

Jane was not thumping quite this bad, was was still violent.

Root Cause

Looking underneath, it became obvious the motor had slipped a bit forward.  Due to the electric, the motor only has one mount, the horizontal stabilizer and a strap holding it to the frame.  The strap allowed the  passenger side of the motor to slide forward about 1 inch, which caused the CV joint to align almost straight from the differential to the wheel hub when off the stands.  Well, it seems like having a constant angle on the CV joint actually prevents all of the balls from being on the edge of the grooves, where the slipping (groove jumping) can occur.  When the joint is perfectly straight, it is quite easy to see how the slipping could occur.  The horizontal angle was straight due to the engine misalignment and  when the car was lowered, the vertical angle would also line up near zero.  While on the stands, there was always a good angle at the CV joint, so slipping was impossible, explaining why the above test revealed nothing.  It is not logical that the controller would know whether the wheels were up and it seems like logic prevails, again.

Solution

So it is time to figure out a better mount for the motor which will prevent the movement.  With the motor sitting in the right position, the drive shafts look symmetric which is a great sign including a permanent angle on the CV joints.  Now, just to engineer a good mount.  With the motor out of position, one shaft was fairly straight (limited angle on the CV) and the other had quite a bit of angle.  Symmetry is a good thing.

If the new mount works, battery testing will resume.  A 600A ammeter is on its way.  This will allow for real current monitoring.  The Kelly Controller provides a 0%-100% current scale, but even under minimal load, it jumps to 20% and seem fairly rough at measuring phase current (Motor current).  The new ammeter will report battery current which when multiplied by the voltage will yield instantaneous power, the real indicator of how things are working.  I am intrigued to be working with such large currents, when most of my career has been spent in the milliamp range of the scale.  A 600A fuse is impressive.