Alternator conversion to BLDC motor with externally excited

[Opus the Poet]

This is what I have so far. I have been prowling the local junkyards for used good alternators to convert, I have chosen a uController chip to run the motor, I have set out a control methodology to be used, and I'm starting to define the program parameters in the flow chart.

I'm looking at import alternators because they are smaller and lighter, and have the potential to make the power I need at the speed I want (100W @ 300 RPM) with the voltage I wish to use (12V)

I'm using the Zilog Z8 MC chip that was designed as a dedicated motor controller (everything but the FET driver and some discrete components) because I'm familiar with programming the Z8 from when I designed toys back in the early 1990s. I love the Z8 uController.

I'm setting the program to run the motor as a sensorless BLDC motor for mechanical and electrical simplicity, as the controller will support this and it only requires inserting a few bytes of code into the program. I'm also setting the program to limit RPM so as to not overspeed my legs The program will act as a cruise control as long as a switch on the handlebars is active (deadman controller), and power will be displayed by a light on the handlebars. Green is system armed up to about 50% duty cycle (maybe less, it depends on the amount of power available with that level of duty cycle). Yellow will light somewhere between 5% and 25% depending again on the amount of power available from the motor and stay lit until 95% duty cycle. Red will light at 90% duty cycle and stay steady until 99% duty cycle, and flash at 2 per second above 99% duty cycle.

Duty cycle of the PWM will depend on what it takes to maintain 300 RPM. If the motor is turning faster than 300 RPM then no assist will be applied, but below 300 more or less assist will be applied to maintain the set point. One benefit is this will develop a steady cadence of about 100 RPM at the cranks. I'm still in the flowchart phase of writing the program as I decide everything I need to get done as well as everything I want to be done, also certain things about the program can't be decided on until I know the number of poles in the stator and the rotor.

One thing I have decided on was to have the duty cycle for the rotor and the stator the same. It's very unlikely that the no load RPM of the motor will be anywhere near 300 RPM with a full power applied to the rotor, so the torque levels can be adjusted by changing the rotor duty cycle with the duty cycle on the stator, so that as the motor moves further below 300 RPM the controller makes each cycle from the stator make that much more torque by putting the same duty cycle on the rotor, so that at max power the controller is just switching unmodulated DC from phase to phase while running max current through the rotor. I realize this might make a bit more than 100W, but then 100W was just a ballpark figure.

An interesting thing about the Z8 MC is it has current monitoring capability and the ability to send that data to another device as well as monitoring voltage. It could behave as a bare bones version of the Cycle Analyst along with running the motor.

[jmygann]

I hope you get it figured out. Its too much for me. but looking for a BLDC motor for a recumbent 2 passenger quadracycle.....

[Opus the Poet]

Still in the flowchart part of the programming process. I have been going over the inputs and outputs and it seems that I will need to use 4 H-bridges to control the motor, 3 for the phases of the motor plus an additional one for the rotor, which is complicating software design. I had been planning on using 3 of the 6 PWM outputs for the phases and one for the rotor, but it looks like I need to use all 6 for the H-bridges controlling the phases so that the back EMF can be used to sense rotor position, plus sending PWM to the rotor via one or two of the multipurpose I/O lines on the Z8. I'm still trying to figure out if that is possible, if the PWM logic level can be sent to any of the other I/O lines. I have been looking at the data sheet (all 700 or so pages of it) and I can't see that PWM signals can be sent to multipurpose I/O lines, but I'm sure they can be somehow with some programming trick. I just have to figure out that programming trick. The actual hardware design has been fairly simple so far with 5 MOSFETs, a few resistors and a honking big capacity diode to flywheel the back current from the rotor to simplify the H-bridge for that. Then a few driver chips for the MOSFETs, a voltage regulator to run the Z8, and the interface to the dashboard readout for power consumed and estimated power remaining.

The easiest thing so far in the flowchart was limiting the RPM. There is a counter that can be used as a speed control, for a hard control that just shuts down the motor above a certain RPM just figure the time between phases at the desired RPM and shut down the phase power until the time for the signal between the phases is greater than the minimum time, but I wanted a softer control so there is a range that below the beginning of the soft part of the control the PWM is reduced by a set amount that varies as the ratio between the actual RPM, the maximum RPM and the RPM at the bottom of the soft speed control. It actually took more words to express that than lines of code. And while posting about it I had a sudden flash that will let me use even fewer bytes of code to make a soft control using the difference between the current speed and the upper limit speed up to a certain maximum difference where the RPM control isn't limiting power...

[Opus the Poet]

OK now I'm having trouble with the lower RPM setting. What I was thinking about was having the counter for the power control just increment after the counter for the RPM limit went to zero when the period for the phase was longer than the period that would be needed to turn 300 RPM, but to get full power with that method the RPM would have to drop to 62. I'm looking for full power below 290 RPM, or maybe 270. HMMM

I should do this every time I hit a coding block, I always seem to come up with an answer when I post about it here

[Opus the Poet]

OK it has been a while. Here is what has been going on.

I got the controller breadboarded and alpha level programmed.

I built a frame and fork with remote steering for a LWB recumbent to mount a 10si or 12si Delco alternator as a mid-drive motor.

For the time being I'm using a 12V10aH SLA as power until I can get LiFePO4, but when I was testing the motor at a health club (using a trainer mount that had power measuring capacity) I used a battery charger that had jumpstart capability (40A 2 14.4V) to prevent voltage sag after prolonged high power use.

Testing, As I posted above I used a trainer that had power measuring capability at a health club to measure how much power the motor was making, and after I did some figuring I asked them to recalibrate the power unit and tell me how badly it was optimistic. After running the motor testing I ran the system myself, without assist to measure my personal input to the system before electric assist, and before calibration I managed to put out 400W for one hour. After calibration that was reduced to 280Wh what I will call my C rate. My C/8 rate is 110W, I can put out 110 W at a rate I feel I can maintain for 8 hours. I will leave calculating my Peukurt effect as a calculation for the student

The motor put out 69W before surge set in. when the system was surging the chain was being flopped up and down violently, sometimes hitting either the seat or chain stays of the frame, and the motor was consuming 13A total, with 5A being used by the rotor excitation circuit and 8A going to the stator phases. I have since learned that I could have cut back the rotor current to 3A without reducing the power output as the iron core of the rotor is magnetically saturated between 3 and 3.5A and does not generate any more magnetic flux past that point. This amp reading was the highest I have seen in on the road testing, but trying to get the system stabilized I ran it up to 20 amps total with 7A to the rotor (about half wasted), with no apparent ill effects on the motor.

I have determined that the surge was caused by my RPM limiting algorithm being too aggressive under load without having my legs to damp it out, but I don't have any issues with surge on the road, so I'm debating if I should change the program to solve a problem that might never come up in any real world situation.

The bike rides like a bike with 27 speeds, without power I could not tell any difference between it and the crank forward bike I ride to church and the grocery store aside from the better aero on the assisted bike and the lighter weight on the crank forward bike. The frame is converted from the customer bike I was making for the trolling motor people a while back that went out of business when the economy tanked. They never finished paying for the bike and I never got the LiFePO4 batteries I was supposed to get to run it either, I got the motor and controller to keep and I kept the down payment for the frame as I had already bought materials for it. And before anyone asks I can't sell either the motor or the controller, separately or together.

[Amberwolf]

If you run the motor at a much higher voltage, you can drastically reduce the current needed to produce the same power output.

When I changed my wheelchair/powerchair motor from 24V to 36V, the difference was dramatic, as the higher RPMs, geared down with a larger receiver ring into the drivetrain for the higher voltage, allowed much lower current draw for the same bike speeds.

That meant that I could use smaller Ah batteries without worrying about destroying them (or even just lowering my range). I had been using two salvaged U1 wheelchair batteries, 12V 31Ah, about 23 pounds each, for at least 46 pounds battery weight. Now I could use three salvaged 12V 17Ah UPS batteries, about 12 pounds each, for only about 36 pounds battery weight. Between the lower current draw and the slightly lighter weight, I get around the same range at the same speeds.

Now I'm considering 48V, which will add 12 pounds, but significantly extend the range, when I gear down the motor even more, so it can run faster yet take less current.

[Opus the Poet]

I understand what you are saying, but the RPM in this case is the limiting factor. Because of the design of the bike that motor can't turn any faster than 300 RPM or it will rip my legs off So to get maximum efficiency from the motor I need to run at the lowest voltage that will get me the RPM I want. Theoretically I could put a really huge PM rotor in there and get a little over 300 RPM freespin @12V or 24 36 or even 48V but that would defeat the other feature I got the alternator for, the ability to shut the motor off and get zero drag without having to use a freewheel device.

Right now I have a system that is legal even over 20 MPH (in TX), is reasonably light, and is mechanically and electrically simple. Parts not used will not break and do not impact reliability. I have left entire systems off of this bike to get the maximum reliability and lowest cost. The output shaft of the altamotor is the mid-drive for the bike, so there's no motor gearing to wear out or suck away power, and when the power is off there's no drag from a PM rotor cogging the stator. I'm redoing the controller to get less losses there, and also to reduce the power needed for the stator which doesn't need as much power as I have been putting into it, the rotor becomes magnetically saturated between 3 and 3.5a, so I just saved about 15% of my juice at full power.

If I can scare up the $$ needed I'm going with large-format LiFePO4 cells somewhere between 100 and 200 aH to get extended range for camping trips and extended battery life in normal commuting. Ideally I would also use solar cells to top off the battery when not using assist.