| Project Purpose:
To develop an outdoor robot power system which can deliver many more amp-hours of energy than batteries, using a gas engine and a modified automobile alternator. The initial purpose was not to build a robot, but to prototype a Hybrid Power Plant that I could use on an over-the-snow robot. A battery would merely support surges in current demand, like a big capacitor. It wouldn't be there to provide power for anything more than half of a minute. This could lighten my load and extend operating time, particularly for a GPS autonomous robotic project.
After building the Hybrid Power Plant and running some tests on it with real loads I decided to put some motors on it and test it out on an R/C robot requiring much more current. This additional step turned out to be a big eye-opener, and gave me another purpose to the project - to test the system on a robot.
Mechanical Design:
Coupling an alternator to a small gas engine requires that the shafts be perfectly aligned. Small "spider couplings" provide for up to 1 degree of angular misalignment, and 0.01" of horizontal offset. To align the engine and alternator I decided to make this a machining project (I had tried other ways before, and failed!). Using belts and pulleys would be out of the question due to the introduction of more moving parts, increased system slop and noise, and danger of moving parts.
The Hybrid Power Plant would was built by mounting the engine and alternator on their own plates, using a 3/4" aluminum rod as an additional alignment (and to double as handles). Although the alternator had two mounting holes, I opted not to use these since they were offset at different z-axis places. Instead, I machined the front of the alternator flat and drilled/tapped four holes in known locations. These holes allowed me to attach it to its mounting plate and perfectly identify the center shaft. The Honda mini 4-stroke engine (GX35) really needed a clutch - the power curve is way up in the RPMs and it generates little torque until 3,500 RPM. Staton-Inc makes a $60 clutch that has perfect mounting holes, enabling it to be mounted to a plate. I also attached the back and bottom of the engine to four more mounting locations on the base plate. The thought was that by mounting the engine to some sizeable chunks of aluminum I would smooth out vibration by closely coupling it to a mass.
The parts are all 3/8" 6061 aluminum, anodized clear. If you want to copy this design, you can download my drawings below (I have provided them in DXF format). You will need to scroll down to find them since I'm only given five (5) attachments per post.
The Hybrid Power Plant now looked like this:
Electronic Design:
Going back to my purpose, the initial idea was that I'd only build and test a portable Hybrid Power Plant - I didn't plan on actually using this sytem on a robot. Therefore, my electronics intiaially consisted of the following:
- Alternator current/voltage regulation Although I chose a "regulated 1-wire output" alternator from a 1992 Toyota pickup, I determined the regulator caused some problems for this little engine. It was difficult to overcome initial resistance to get the alternator RPMs up 4-6,000 where it produces a lot of power. The alternator was opened up and the voltage regulator was bypassed, bringing the field current outside for my own limiting. This eliminated all resistance and allowed it to turn really freely. The voltage regulator design we came up with allows me to limit field current with external resistors, and charge the battery with a shut-off feature if 14.5V is attained (or other voltages, as set by a potentiometer). This provided a regulated voltage for motors and other loads, a battery charger, and 15-20 amps of current should I need it. A schematic is below:
- Throttle control / RPM measurement The amount of current generated relates closely to engine RPM. The higher the engine throttle, the more power! The initial system uses an SX28AC/SS (on an SX28 Proto Board) to count magneto pulses, and a servo to manage the throttle. Having measured current at different RPMs, the goal was to set the engine at known RPMs and known current output. Controlling these variables would ensure that I'm not overcharging the battery (max charge is 5A) and generating more power than I need. Through this process I destroyed many pants and shirts with acid from overcharging batteries. If I was to generate lots of current, I'd have to use what I couldn't put in the battery. With lots of help from a key friend, the following circuit was developed:
The following SX/B code is an open-loop, potentiometer-controlled servo throttle for the above schematic. This example doesn't count magneto pulses at the same time. I've got both pieces of code working independently, but have yet to integrate them using the SX's RTCC/ISR features.
' Potentiometer Throttle Control
DEVICE SX28, OSCXT2, TURBO, STACKX, OPTIONX
FREQ 4_000_000
' -------------------------------------------------------------------------
' I/O Pins
' -------------------------------------------------------------------------
Servo PIN RA.1 OUTPUT
RCCircuit PIN RC.3 INPUT
' -------------------------------------------------------------------------
' Constants
' -------------------------------------------------------------------------
' -------------------------------------------------------------------------
' Variables
' -------------------------------------------------------------------------
ServoVal VAR WORD
analog VAR WORD
' =========================================================================
PROGRAM Start
' =========================================================================
Start:
LOW RC.3
PAUSE 5
RCTIME RC.3, 0, analog
ServoVal = analog/2
ServoVal = ServoVal+130
PULSOUT Servo, ServoVal ' PAUSE 20
PAUSE 20
GOTO Start
- Car Headlights / HB-25s and 12V Motors for a Load To test out the system I needed a big load. I used two 50W high-beam headlights, two massive 12V Groschopp Planetary Gear motors, and two HB-25s to control the motors. The HB-25s were connected to the SX28CA/SS Proto Board for control (nothing important here - just a PULSOUT pin, value). I tested the system out with some big loads, and it worked great. It ran for an hour, keeping the battery charged and headlights on high-beam. I quickly learned that an entire power management system would have control over engine RPM and alternator field resistance. I probably wouldn't care much about the battery since it would always be charged. I'd only generate current if I was using the power. I was using about 15-20A at 12V. This is what my load looked like:
The Oversized Boe-Bot
I was satisfied that the Hybrid Power Plant would perform really well, and I was ready to shelf it so I could design a robot around the concept. However, the same friend who helped me with the electronics pointed out that I could further prove out the concept if I simply mounted wheels to the Hybrid Power Plant. It made sense considering I hadn't been able to test the motors with a load. Then I wound up with this:
I switched the HB-25s straight over to R/C function and added a transmitter. I set the engine throttle with a potentimeter/servo (SX/B code below) and didn't bother counting magneto pulses. Away we went! Videos are posted in the next message. Under R/C control the whole system works great. I'm certain the Hybrid Power Plant could be used on an outdoor robot. The videos show the first pilot run when I had 16 Ohms of resistance in the alternator field, but now there's only 8 Ohms and it runs much faster. It'll lay down scratch, slide around corners and reverse at full throttle (I completely destroyed an HB-25 doing this, by the way).
The oversized Boe-Bot runs better backwards due to weight distribution. Since the Hybrid Power Plant was never supposed to roll, this is something I'll fix on the next project. Moving the battery around lets it run better in the Boe-Bot's "normal" direction with the motors up front.
Safety First
I nearly had several accidents with this whole project. I think they're worth sharing so we can all keep our hobbies for the rest of our lives. First, I had to develop and prototype the magneto RPM circuit with the engine running on my desk, inside the office. It was necessary to attach an SX-Key, a PC and let the system run to develop the code and to tweak the signal conditioning circuit from the magneto. This really shouldn't have been done indoors due to CO poisoning (and noise). I kept doors and windows open, but it was a bad idea and I knew it at the time. Nothing bad happened to me, but I wouldn't do it that way again. Other bad things happen from running an engine on my desk. Once my laptop mysteriously shut down due to vibration. Next time I'll run the engine outside and use several serial cables (this is the same way we used to test BASIC Stamp projects on R/C airplanes, such as the dual-engine synchronizers).
Another time I was a bit unaware of the power I was generating. I had connected the battery to the alternator and increased the throttle to a high RPM. I then disconnected the battery and easily dumped 100A into the voltage/current regulator circuit. The electronics truly exploded and caught fire - I didn't know if I should have reached for the fire extinguisher or tried to cut the engine first. I decided to cut the engine first. Meanwhile I exploded capacitors, LEDs and a 25 Ohm/1 W aluminum housing resistor. The trajectory of all components was away from me. Could have been avoided by keeping the battery connected.
I also experimented with some smaller spider couplings to connect the alternator to the engine. Though they were rated for much higher RPMs than I was using, a small misalignment sent some parts into the air. I wasn't standing along the axial line of rotation, so I didn't get hit. Before any schoolchildren get to see this I'll need to build a shield around the coupling.
I was prepared for most of the accidents, but didn't quite anticipate the fire on my desk.
Credits
My friend John from Tigerbotics provided tons of help on the whole project, especially the electronics. Without his assistance I'd still be using a hacksaw to cut aluminum, trying to understand the "how" instead of "why" and be somewhat frustrated. Since I can't help him with his projects the same way, my intent with posting the whole project is that others could benefit the same way. Post Edited (Ken Gracey (Parallax)) : 10/26/2006 9:48:51 PM GMT |