|
January 9, 2018 I'll leave the old stuff below. I
went to my friend's place this morning with an upgraded
firmware on a chip. The verdict is in. As long as the
calculations for the next firing is done with the
measurement from the previous cycle, the engine will run
flawlessly. The first version was working on numbers one
cycle back from that and it just was not solid. It was
just the way the software evolved, I knew I needed to be
as close as possible, and now that is the case. So the
stuff below about two magnets is out, unnecessary. The objective was to start the engine
on TDC and advance the spark as the engine sped up. This
is the best way to go as the battery doesn't have to
work against an advanced spark on start up. The only thing I can think of adding
from here is a pot on the box so the spark advance can
be trimmed with a timing light. And this won't be
necessary, it is just that it is pretty simple to do. If
the installer just wants to go with good enough, leave
the pot centered. The other thing I would do, and again
because it is simple. Put a couple of LEDs on the
package that would indicate when the engine is running
close to 1800 rpm. Maybe use a three color LED that
would: color 1: less than
59 Hz. The MSP430 sports a watch crystal so
measurements would be spot on. So, it was a good day. David lives
off grid and it has been cloudy. He has been running a
back up generator on gasoline. But he has a natural gas
fissure on his property. Last fall I created version
three of the natural gas compressor. (I'll post
about that later.) Back in the 60s there was enough
pressure from the cap that it could be used directly.
Not so much today, maybe the drought?. He runs his water
heater from the gas and there is a leg of the line up to
the energy shack. This generator was fitted for propane.
But the mixer reaches the 2.5" of gas and the heat
content of the gas so that the generator runs just fine.
We ran it on and off for an hour using ~2500 watts to
charge the batteries. Next couple of months we are looking
to get good water from down the canyon. Here is the start of that project. December 29, 2017 ( continued) The first thing was to
bread-board a driver for this coil. From what I've read
these things like a 3 millisecond charge time to build a
magnetic field for a good spark. That works, it created
a wicked spark, over an inch and would light paper on
fire.
Of course a bread board
won't cut it as the generator has quite a bit of
vibration. So once I was satisfied with the circuit I
drew up a schematic so I could soldered it up on a
prototype board.  Close up of the Hall sensor. I mounted it on a stiff piece of PC board as it has to be positioned over the flywheel.  Here is the schematic. Hover over for a full size look.  Simply beautiful. Here is the Hall Sensor in place on the engine. Notice the magnet on the flywheel above the board. This is one of those really strong rare earth magnets, it will not move. I'll guess it takes many hundreds of Gs to move that magnet and those kind of forces won't exist on the flywheel. I did put a timing light on the igniter box. The light will go off when the magnet is under the Hall sensor so turning the flywheel to the right just until it goes on is where the sparks will fly, three milliseconds later. Well, that is a problem. It will still run very retarded at full speed. Version 2 will have a micro processor, something like an MPS430G2211. All that is really needed is one timer. The game so far is proof of concept, the engine runs beautify even if it is wrong.  December 31, 2017 Last night I started programing an MPS430G2211 for version 2.0. Everything else will stay in place, it is just a matter of swapping out the dumb electronics with a brainy controller.. A true ECU, at least, for the spark.  noon... I wasted a couple of hours chasing a goofy interrupt. Thinking, "What I'm I doing wrong!?" It has been a while since I worked with TIMER_A, especially with capture mode. I shut off the interrupt on TC0 and watched memory. The interrupts were still being set! I checked the erratasheet and: TAB22 TIMER_A/TIMER_B Module Function Timer_A/Timer_B register modification after Watchdog Timer PUC Description Unwanted modification of the Timer_A/Timer_B registers TACTL/TBCTL and TAIV/TBIV can occur when a PUC is generated by the Watchdog Timer(WDT) in Watchdog mode and any Timer_A/Timer_B counter register TACCRx/TBCCRx is incremented/decremented (Timer_A/Timer_B does not need to be running). Well, that makes this chip about worthless. I replaced it with an MSP430G2553, problem gone. Lesson, always completely read the errata on a chip. Back to work.... January 5, 2018  For having a pretty lame tach this thing works rather well! Here is the code and the result of the the proxy motor running at about 1000 rpm. I've run it on the actual motor and watched the spark with a timing light.  The objective is to keep the infrastructure of this system as simple as possible. By using two magnets on the flywheel, one north facing and the other south facing up, it will be possible to measure the motor speed almost in real time.  And then two Hall sensors on the same board mounted with sense at opposite magnetic polarity. Copyright
© 1996 - 2018, http://www.lakeweb.net/Onan/index.html |
