Building your own generator

[dillwhacker]

Bugger...
...

Jimmy - PM me tomorrow - I have several small digital scopes you can borrow if it will help.
The spark signal will always be noisy and can probably be filtered in a simple analog way to achieve what you need.
I will read closer tomorrow but this cannot be too hard.
Cheers
T

[bpgoa]

Building your own generator

So. How is this going.

[JoeBailey]

Building your own generator

Have you thought of using a stator gator on the coil, we use them on gassers for governor pickup and it converts ignition pulses to a Futaba signal. Might be of some use to you.

[jimmyhorns]

Have you thought of using a stator gator on the coil, we use them on gassers for governor pickup and it converts ignition pulses to a Futaba signal. Might be of some use to you.

I hadnt thought about that... didnt know about them

will look into it


Dillwacker has kindly offered to drop a scope round for a couple of days next week, so things should be moving on soon

Porscheboy has also fixed up the exhaust

[jimmyhorns]

So, time for a quick update:

Firstly the new modified exhaust, kindly fabricated by Paul (Porcheboy), has now been installed..... We didnt quite guess the angle right so I have had to take one of the frame sides off for the time being, but its a very simple fix....

Unfortunately it has done nothing to stop it sounding quite so loud, so we are still going to need to find a cars back box or similar to use as an expansion chamber and 'muffler'....

Dillwhacker kindly stopped by on his way back from relatives on Friday, took a look over the beast and thought it was a bit mental..... He scribbled down the basis of a low pass filter, and walked me through it, and we used his scope to look at the raw signal coming off somewhere from the magneto coil....

There are two wires, that when shorted together cause the engine to stop running, so I guess these are either side of the magento coil, and the hope was we could pick up a signal from it..... well that we managed..

between 100 and 200V and a relatively clean signal, which was nice....

Anyway, a bit of tinkering with the low pass filter and a zener diode and I had a clean enough signal to be using for the arduino.

A super quick test program (that just counts up from zero and outputs the counter to the serial monitor) shows that the arduino is happy with this signal, although im not 100% sure how many times its being triggered per revolution. It should be once per revolution, and twice per cycle, though when turning it over on the pull start, im getting more counts than double the number of exhaust 'pops' so this will need checking with a normal rpm counter that runs of the HT lead (I have one of these on a timing light)....

Thanks again go to Dillwhacker, who was very trusting given we hadnt met in real life before....

Well here it is:

[Boggy]

I've seen paint dry quicker than that James

[dillwhacker]

Hi James,
Good to meet you on Friday - it was fun & I didn't think your generator was 'mental' at all ... just a splendid concoction of mechanics and electrics and things that go up-and-down and round-and-round and I loved it!!
I don't understand your scope display at 2:02 - that should be a square ish signal going from plus 3.3V (or whatever your zener is) to minus 0.7V... Yours looks like plus 100 V to minus 30 V ...
Maybe scope connected at wrong point or channel one is x1 not x10? - I don't know - but if the micro controller is happy then it is OK, I guess.
If you set the scope trigger to AUTO instead of NORMAL, then it will run continuously, and you will be able to monitor how many pulses per rev you get by cranking it by hand. (watch your fingers, disconnect the plug..).
I'll be back soon so good luck & call me anytime.
Trev

[jimmyhorns]

Hi James,
Good to meet you on Friday - it was fun & I didn't think your generator was 'mental' at all ... just a splendid concoction of mechanics and electrics and things that go up-and-down and round-and-round and I loved it!!
I don't understand your scope display at 2:02 - that should be a square ish signal going from plus 3.3V (or whatever your zener is) to minus 0.7V... Yours looks like plus 100 V to minus 30 V ...
Maybe scope connected at wrong point or channel one is x1 not x10? - I don't know - but if the micro controller is happy then it is OK, I guess.
If you set the scope trigger to AUTO instead of NORMAL, then it will run continuously, and you will be able to monitor how many pulses per rev you get by cranking it by hand. (watch your fingers, disconnect the plug..).
I'll be back soon so good luck & call me anytime.
Trev

Its a 4.7V zener, but interestingly the voltage never gets up that high...

I was reading between 2.7 and 3.5V from the scope, you cant quite see the gain/range on channel 1 due to my poor videoing skills, its not easy to crank the engine and hold an iphone still....., but the gain was set to 1.0V / division

[dillwhacker]

Its a 4.7V zener, but interestingly the voltage never gets up that high...

I was reading between 2.7 and 3.5V from the scope, you cant quite see the gain/range on channel 1 due to my poor videoing skills, its not easy to crank the engine and hold an iphone still....., but the gain was set to 1.0V / division

Silly me, can't read my own scope.....What a dillwhacker I am...
The voltage never gets to 4.7 because of the rise-time of your low pass filter compared to the width of the input signal. It looks fine - no double edges and a decent clean pulse, but maybe you should reduce the cap value a bit, because if some pulses only reach 2.7V, is that enough to guarantee a logic '1' (high) to the Arduino?
See what it is like when running - pulses will probably get bigger with higher RPM.

[jimmyhorns]

the harder the pull, the higher the spikes, so hopefully will be ok when up and running... I will get to test that out tomorrow.

Ive actually doubled up on the capacitors there, there is a 200nf and a 470nf in parallel, or I was getting a double spike when I looked carefully enough.... not sure that would have been enough to upset the arduino, but no point in having two for the price of one in terms of triggers

Provided the weather forecast was wrong and its not pouring down with rain I will be getting it going, and also getting the MIG welder down from the loft to see if it still works

just need some gas for it now

[jimmyhorns]

so near and yet so far.....

running it crashes the arduino board.... and I think I know why.....

when it misfires (and it will do that from time to time, alot when starting from choke....) it seems to randomly hang the board.... all is ok after a 'hard reset',but the same thing happens again.... it will count til the cows come home when its not actually running itself...

I think here is the problem :

photo(52).jpg

When it misfires, you see the large negative deflections that are not apparent on hand cranking or when it is running well.....

So, how to clean up the input further?.... I currently have a 100k 3W resistor onto the coil output, then a 200nf capacitor to earth to make the low pass filter.... there is also a 4.7V zener across the capacitor to limit output voltage....

So.... feed it into the base of a transistor? use an op amp with only +5V and 0V across the supply pins?.... something simpler?....

[cjcj1949]

What happened to the hall effect sensor? Coupling directly to a high voltage source is unusual. You could inductively couple with just a loop around the ht. I'm not sure what voltage you have across the 100k resistor, but that needs special care to avoid tracking. If your zener fails you will have many volts going into your arduino. Do you have a reverse bias diode (anode to 0V), to clamp the input, (from going negative)?

A resistor across the zener would make it safer,(to act as a voltage divider) but an opto coupler like a 4N27 (probably out of date now as I used it in the 70's) would isolate your electronics from the high voltage. You will probably (definitely really) need hysteresis either in software or hardware to avoid multiple counting. It will need to be large. A voltage value close to the peak followed by a low value. Something like 70% and 20%. You could do it in software by using the A to D converter.

[jimmyhorns]

The hall effect sensor was not reliable.... reading off the flywheel was a bit hit and miss with the vibrations from a small old engine, and I wasnt getting a regular pulse, as the flywheel had balance holes drilled within....

The proximity required to the flywheel was also a concern, as with only a couple of mm clearance, it wont take much movement, or expansion with heat for it to no longer be reliable....

I dont have a reverse bias diode in line yet, but thats going in this morning.

You are not the first person to tell me that its not a good idea to couple directly to the HV. Because its a magneto there isnt a LV side of the coil to use. The voltage across the 100k resistor is between 100 and 200V peak, so that should be ok I would have thought.... If I double up on the zeners (run them in parallel) would that give me some protection from one of them failing (open)?....

Im going to have a look at using an optical isolator like you suggest.... as that may offer me a degree of separation from the arduino crashing as well...

Thanks

[dillwhacker]

The zener diode will act as a normal diode when reverse biased and conduct at about 0.7 Volts i.e. minus 0.7V in your circuit, and you can sort of see that on the scope trace - it would be more visible at faster timebase speeds. However it might be a bit slow and let the larger negative spikes through. An additional reverse bias diode is a good idea - use a shottky diode like a 1N5817 as this will have a lower turn on voltage (about 0.4V) and a faster response.
Double zeners or opto isolators are a good idea too TBH, to save the controller if it all goes wrong...

[cjcj1949]

The zener diode will act as a normal diode when reverse biased and conduct at about 0.7 Volts i.e. minus 0.7V in your circuit, and you can sort of see that on the scope trace - it would be more visible at faster timebase speeds. However it might be a bit slow and let the larger negative spikes through. An additional reverse bias diode is a good idea - use a shottky diode like a 1N5817 as this will have a lower turn on voltage (about 0.4V) and a faster response.
Double zeners or opto isolators are a good idea too TBH, to save the controller if it all goes wrong...

Oops. Really embarrassed on the zener diode front. A zener is just a silicon diode after all. Just designed for a certain reverse breakdown. There is something in the back of my mind about not relying on them for forward clamping. Probably not important for this application. I'm not sure offhand what the reverse volts spec on an Arduino input is. The current will be small from a 100k though. I don't understand why you have such a low voltage across the 100k I thought spark voltages were much higher, in the tens of kv.

Having lots of wires floating about with a breadboard is asking for trouble as well. A small board with the resistor and capacitor plugging into the Arduino socket would help. All tracks need to be as short as possible. A hex Scmidt trigger would help like a 7414 would help, but an op amp with some positive feedback would be better. The time constant of the CR network needs to be quite large as you are charging to 200V, but only needing 3V to register a "1". The CR value gets you to around 70%, and you are around 3% or so, so a factor of 20 is what you need. The pulse should have slow rise time if properly filtered. If you sample the pulse with the AtoD you'll be able to see what is there and come up with a robust algorithm to extract the basic frequency. You have many pulses that you can look at to ignore the odd noise spike. You can use bit sync/phase lock loop principles to ignore much of the noise as you know that the acceleration of the generator is slow. You can have a narrow window of where the next pulse is coming from. This sort of interfacing is not trivial but a trivial algorithm may be possible. The odd error may not matter much, it could be just a glitch on the throttle servo which the engine may be too slow to react to.

I was looking at O rings for my 130=X and came across anti-vibration mounts. These give you a lump of rubber between metal supports. This would be ideal for isolating your alternators from a metal frame, if you need one.

The most annoying thing about designing things yourself is that you often find something cheaper ready made. Aldi are doing a generator for £50 or so this week.