Building your own generator

[jimmyhorns]

The output from the sensor isnt anywhere near enough to trigger the interrupt on its own..... I tested the output voltage when it was hooked up to my AR7100R, and it was 3.5 which should have been plenty, but im only seeing 0.3V now... which must be something to do with the circuitry in the AR7100R..

It might be worth looking into this. Perhaps you have an open collector output that needs a resistor to 3.3V or 5V. There are resistors you can enable on the arduino, but they are too large. Some hysteresis on the op amp and a capacitor to 0V will help with noise immunity.

Its the standard spektrum crank sensor.... I think its a hall effect device, but I cant seem to find any more data than that

[jimmyhorns]

so, a little more work with the hardware today, and now have this prototype:

[cjcj1949]

I've not been able to find any spec for the sensor. It's designed for one rx and it switches an led. That might be direct through the sensor. You probably need a resistor connected to 3.3V as the sensor is probably just acting as a switch to 0V. Open collector logic is sometimes used without pullups and does work sort of. One system I had the misfortune to work on used small capacitors (1000pf I think) to cure the problems produced by having slow noisy edges as clock inputs. It did save on the power supply current though. Your sensor does go up to 3.3V when connected to the correct rx and so connecting a 1k pullup to 3.3V will give you a better signal and probably avoid having to use the opamp. There is a comparator on the 328 that you can probably use as well. Standard practice to get maximum noise immunity would be to set the switching threshold half way. Hysteresis is a small amount of positive feedback that modifies the threshold. A large resistor from the output to the positive input will give a higher threshold for positive going edges and a lower threshold for negative going. Lookup schmidt triggers and op amps for more detail, or the arduino forum.

It would be interesting to see what happens to the engine when you start to drag current out by connecting a charger. That will give you a feel for how rapidly and how much throttle your control system will need to give to keep the motor going. Or to put it another way, how fast your idle needs to be. If the chargers ramp up slowly it shouldn't be too bad. If they don't you could make life easier for yourself by having a boost button that ramps up the revs just before a charge is started.

[Gixxer]

Seeing as you're adding trickery, you could set the revs at a lower level when there's less/no load and raise the revs as a load is applied and add an ignition advance curve, relative to the engine RPM

[xygax]

very good article on hall sensing http://sensing.honeywell.com/index.php?ci_id=47847 it would appear that most are open collector devices. so the simplest way to reliably trigger a micro is use a pull up resistor into a comparator better still two using and upper and lower threshold that will form a solid Schmidt trigger so you end up with a really good square wave input then the micro can easily do the rpm calc for you... so allow your input to go into an interrupt driven counting register. and set up a timer to expire once every second. read the counter multiply by 60 and zero the counter... hey presto RPM... alternative apply the same logic and count the sparks you can use the same coupling method but you will need two fast diodes and a high value resistor to form a voltage limiter then a small RC network to clean the pulse up a bit and you will get RPM/2 using the same formula (4 stroke sparks once every 2 revs)

[jimmyhorns]

Well I cant mess with the timing easily (its not electronic, its magneto driven)... so that will have to stay as is.... rpm is going to hopefully be load dependent

[bpgoa]

Building your own generator

You could always drive a servo to move the plate it's sat on.

[cjcj1949]

One thing at a time. You need to establish how reliable your rpm sensing is. Just measure the time between interrupts and see how much variation there is. Then design your rpm algorithm to take account of the variation. It doesn't matter if you get the odd pulse and try to reduce the throttle once every 5 minutes. The engine probably won't notice it or will just make a noise. It is important to look and see what is going on and not be misled by what you want to happen.
I agree with Steve that the simplest way should be to use the ignition pulse, but as you have a working method I'd continue with that for now and concentrate on getting something that works. The ignition pulse could be picked up inductively with a loop around a wire or just with a piece of wire. Your signal is very slow and so will be easy to filter out any noise.

[jimmyhorns]

well back to mechanics for now.....

Beefing up the frame....

photo(49).JPG

[jimmyhorns]

Ok, well thats the mechanics pretty much sorted out now.

photo(50).JPG
image(8).jpg

Everything fits, the cover is missing so the exhaust has somewhere to go at the moment......


Paul (Porcheboy) has very kindly taken the old exhaust away, and is going to knock up a new manifold at work

There is also a slight problem with the crank pulley that is going to need to be resolved at some point in the near future... its not quite true... again Paul is going to get a new "something I cant remember" knocked up to fit on the crank shaft and then have the pulley bolted to it.....

Had it fired up, with both alternators working, and both batteries taking charge tonight. Paul tached it with his heli tach at around 2500 which is pretty much where I want to be with it, and that fits with both the rpm read out from my timing light and the calculations from my crank sensor. Im not going to be too picky when it comes to the running governor, +/- 200 rpm will be just fine, which should hopefully make the algorithm easier.


So, back to the electronics and programming, unless the new exhaust is made....

[PorscheBoy]

It will get done, maybe by the weekend......

[jimmyhorns]

My bad, meant to say, until, not unless!

of course its not made in 2 hours lol

[10x]

Hello Jimmy Just looking at you generator, mechanically I think you may need to add a idler pulley in between the 2 alternators to give the belt more grip on the alternator pulleys when under a heavy load to stop the belt slipping. At the moment you have about 90 degrees on the bottom alternator and slightly more on the top one, when mounting an alternator to a engine you always try to get as close to 180 degree belt rap around as possible, this applies to the crank as well. Then the belt does not have to be so tight and is better on the shaft bearings in the alternators.

[jimmyhorns]

Porcheboy pointed this out to me today as well.... I thought I might get away without one, but its looking unlikely......

Once the 'proof of concept' is up and running, I think I/we are going to need to build a box section frame for the engine and alternators, once I figure out how to isolate one from the other.... this would be the natural point to put the idler pulley in I guess?..

Thanks for your input

[jimmyhorns]

Bugger... looks like my idea of scoping the coil isnt going to work...

This:

photo(51).jpg


is what my (rather old) scope thinks a square wave looks like ..... I suspect that something has gone awry with the valve that controls the oscillator, but I could be wrong...

I wonder if zygax's old offer still stands ....