nyemi

Hi bluejets
I called Jake's attention.
The bounce-free input.
This Software can be solved.
The hardware is not bad.
But not enough.
Find a good solution.
link:www.labbookpages.co.uk/electronics/debounce.html

I checked, Jake Code.
I'm curious. Jake, that solved the problem.
I compiled ASM.
I see everything.

bluejets

Nyemi,
this is the change I sent to John to try out. It may need changing the new values to smaller sizes to change the pulse time though. in principle it should work.

nyemi

Hi bluejets
Yes, the circuit is good.
Part solves the problem.

But this is not causing the problem.
The software does not turn off, the input pin (GP2) of the interrupt.
When the pin (GP2) of the signal is displayed.
I do not know. Jake, why do not you use IOC register (INTERRUPT-ON-CHANGE GPIO REGISTER).
The opportunity exists to separate interrupt handling pins.

I'm sorry. I now see GP4-5 pin interrupt is not used.
It seems to me that in this case the register INTCON, INTE bit is enough to delete it.
The task is waiting for him.
I was debugging and testing.

COM

Jake this is why the pic needs to fire at start up( at least) on the low to high transition.After 500 rpm then you can have the pic trigger at the high to low transition. Then if the magnet stops on the HES it won't continuously fire the spark plug. No extra parts are needed for the HV board. the 680 resistor, .1 cap and the 220 resistor the way they are setup are designed are for a one- shot signal to the gate of the scr. the issue is in the software.

nyemi

There are two commands in the code.
INTCONbits.INTE = 0;
INTCONbits.INTE = 1;
The problem is solved.

COM

Ok,

So Nyemi how do you implement that into the latest spreadsheet?

gr8flyer55

I edited the command line to reflect Nyemi's proposed fix, recompiled and am programming the chip now. We shall see how it works shortly! Charlie, have you tried it yet?

John

COM

Hi John,

I have not tried it yet.

gr8flyer55

Charlie, well it stopped the buzzing for sure, but it still burns up my transistor!!

gr8flyer55

Looked thru the code after it overheated and destroyed another transistor, and there is definitely something wrong in the section where it is testing to see if the engine is running yet or not. So I tried a 8000 rpm tester plugged in instead of the hall sensor simulating an engine running and then the transistor was cool as a cucumber. So, it is only a problem in the non-running engine loop that needs to be fixed. Don't waste your time, let Jake look at it and fix it. I've now ruined the 3rd transistor and am not sacrificing any more to this code error.

Plugged in another chip with my previously working hex file and it worked perfectly, no heat, no buzzing spark before starting, just good performance.

John

Gompy

Burnout the transistor is a problem from using the flightbacksystem.
Normaly the ocsilator will stopped if the HV is 2 or 300 Volt.
At low speed this Voltgae will there be sooner then when the engine is running fast.
The transistor will be shorted for a long time at low speed.
A CDI don't use this, he must be open untill the capacitor is loaded, not longer.
I want to try the schematic below if my brain is working back OK.

iura

Gompy, I told you about fly back converter some years ago. I whant to say again: in our case this is the best scheme for HV converter. But this one is too difficult, look for this:
http://www.dos4ever.com/flyback/flyback.html

Gompy

All the flybacks I build the last years have problems with the transistor (at low rpm)
If I'm standing befor a traficlight, I can't run my engine up to 2500 rpm........they will ask me if I'm grazy.
I know for models they use or thry to use as less as psossible component, every component cost money and is weight.
Thats why I want to use a differend kind of HV-board for my bike, not one who need by eveery traficlight a new transistor.
Not one of the HV-boards we have build is using a "shutoff" of "slowdown" if the capacitor is loaded.
Maybe just Jake do right now , a fresh look to the software, I think we need a fresh look to the hardware too.

http://users.triera.net/zupanbra/pdf/MC34063.pdf

I think you have to ask John, Charlie and all the others how many transistors they have blow the last copuple of years.

iura

This only one problem in fly back converter: transformer. In this scheme we can use the stand by transformer from old ATX power module(EE-19N) It’s need 8v input and can provide output about 200v at 33mgh frequency.
http://webx.dk/oz2cpu/radios/psu-pc400.htm
It’s depends only on they experience.

COM

Well Actually I have never burned up a transistor in my HV boards. If The transformer is would correctly, at idle or WOT the total amp draw from the HV board is minimal, and thus no burnt up transistors.

COM

I tried to reproduce the problem and now everything works as it should. I am not getting multiple sparks. Even if I stop the magnet at the hall sensor the ignition fires once and just sits there. This is testing on my engine sim though.

gr8flyer55

Charlie, it seems like a random latchup problem caused by the software though.. I disconnected the timer board completly, eliminating the software glitch, plugged a sensor directly into the HV board and it works perfectly every time.

I can hold the magnet at the sensor for 15 minutes and nothing happens anywhere till I move the magnet off the sensor, then one spark, and it works as it was designed. But plugging in the timer board again with the bad glitch software reproduces the same problem, latchup and the transistor is destroyed in less than 5 seconds.

So there is my proof, software problem, not hardware problem. Changing the line in the software might have fixed the buzz, but it didn't fix the original problem of the latchup. In fact, if the buzz problem hadn't shown up being constantly sparking as the magnet was held at the sensor, this problem would probably never have been found until someone burned up a transistor and started bad mouthing the hardware! I know how I test things here, very methodical with numerous boards involved so it isn't just one suspect board or batch of transistors that can't take the heat. When it shows up on the last 10 boards, which previously worked perfectly for over a year until this software glitch, I know it's a software problem.... Just to prove my point further, I tried an RCexl ignition without the timer board, and a CH ignition without timer board, plugged the bad software timer board into them and POOF! Now bad transistors on both of them.

Same timer board, unplug Jake's chip, plug in Nyemi's chip, no glitch, no heating, no problem.. I have 5 HV boards and timer boards hooked up this way and all work exactly the same, soon as I put Jake's chip in and power up, swipe the magnet past the sensor, poof! No more transistor.

I also tried an earlier version of Jake's software, the one posted on the site at one time as being tested and working, and that version doesn't have the problem at all. Seems to be the last edited version that is the culprit.

Testing on an engine SIM is different than a live circuit. Can your engine SIM show you the amount of current draw on the transistor's junction? If it can, check it out. If it can't and still doesn't throw up a red flag somewhere indicating that the transistor has stopped oscillating, then you need to do some live circuit testing! I only have 3 transistors left out of my last 50 ordered until Monday. Can't afford to waste any now. I'm done testing till Jake fixes this.

John

Gompy

I've do the same John have done with our version(s) we have build the last 8 years, but with or withgout timerboard we have the same problem.
It looklike to me a self ocsilerend transistor, start fast and fatser and after a split second the transistor is opend compltete and distroyed by high current.
This only happend with no or slow pulses (rpm) and for some stranges reason not at high rpm (8 a 9,000 rpm).
We use the same ring core like John do / have done or e-cores, the result was allwas the same.
If I take a look on the internet for professional CDI schematic, they never use flybacks like we do.
They use setups like Nyemi have into the ZVS DC CDI, with differend kind of ocsilators but no one use free run flyback ocsilators.
They also use all the "Voltage safety switch", at 300 Volt they switch off / controle the ocsilator.
Thats also I want todo, it saves current and I think it's also better to contol the HV.
I want the best and I think thats wat we all want, but I'm not the software guy.
No complains to John, he is doing his very best to find the "bug", same Jake do, but I don't think it's only the software.
Both boards have to work without problems, we don't have to use both boards to eliminate the problems.

jakestew

Wow, seems like some people are having problems and some aren't.

What makes it hard for me is that I haven't seen this problem in my testing. I've never burned out a transistor, nor seen any random sparks except for one at power up (sometimes) and one at power down (occasionally).

I'm not sure how the software could make this spark continuously. A falling or rising edge generates an interrupt, the interrupt service routine then reads the pin to make sure the signal is actually high or low (whatever it should be), and only then loads a spark delay into the compare module of the timer. Only when the compare interrupt is generated should it fire a spark.

To see if the engine is running the software only checks to see if the timer has overflowed a certain number of times since the last pulse.


Some people are having Q2 (2N4401) burn out? Am I understanding the circuit right? Q2 should always be high, except to trigger the spark right?


For the continuous sparking... Has anyone checked that there is not excessive ripple in the hall signal? Maybe the magnet is only halfway triggering the hall and that is causing it to oscillate between high and low?


I will definitely be working on this to solve the issues. If I had encountered these problems I would have let everybody know. If anyone can explain exactly how and when the problems are happening that would help me replicate and solve them. I'll check to see if I can get the continuous spark problem and poke around to see what might cause things to be left in the wrong state.


-Jake

COM

Well not exactly. When I test it on my sim I am using a igniton system with sparkplug and the timer board. It acts just like it would if it was on a engine. only difference is that I can watch the signals(input, output) on my scope and read rpm. so The timer board does not know the difference.

COM

IF you are talking about Q2 on the timer board then no it should not be always high. its a npn transistor. meaning that a logic 1 turns it on which will cause the signal to the hv board to go low.

I have not checked, but seeing that the whole setup is powered from a battery pack I don't think there is a whole lot of ripple. I will continue to try and reproduce the problem. I only noticed the issue after I had to make changes to the spreadsheet.

nyemi

Hi Jake
You do not deal with the Hall signal length.
If the signal is too long. The interrupt is generated.
Because of the long signal, electronic smog is present.
To overcome this problem.
If the input signal for sure. (GP2 control, after)
Clear INTCON register, INTE bits. (INTCONbits.INTE = 0
The output delay after. (GP0, GP1 default)
Turn on, again. (INTCONbits.INTE = 1
With this problem solved.


Hungarian
Ön nem foglalkozik, Hall jelnek hosszával.
Ha túl hosszú a jel. Az megszakítást generál.
Mert a hosszú jelben, elektronikus szmog van jelen.
A probléma megoldása.
Ha a bemeneti jel biztos.(GP2 ellenĊ‘rzés, után)
Törölje INTCON regiszter, INTE bit. (INTCONbits.INTE = 0
A kimenet késleltetés, után. (GP0-GP1 alap állapot )
Kapcsolja csak, újra be.(INTCONbits.INTE = 1
Evvel a probléma megoldva.

bluejets

Com,
The way that Q2 operates is as Jake said. It is in the "on" state and goes off when the spark is required.
The output works much the same as the suggested mod I posted for the hall switch. i.e. R7,C4 junction is held "low"

As soon as Q2 turns "off", C4 charges via R7,R8 and puts a positive pulse on the SCR gate, turning the SCR "on".

If Q2 is off longer than the required time for the gate to fire the SCR then this does not matter as the capacitor charging is now controlling the gate "on" time. The gate gets one positive pulse on each transition of Q2 from "on" state to "off" no matter what.

This however cannot control the off state of the SCR because the SCR requires commutation to turn it off once it is set on by the Q2 circuit. Commutation is either where the SCR is forced to turn off by making the cathode and anode voltage approx. the same by discharging the complete charge in the HV or by shorting out the SCR anode to cathode or by forcing it off using things like static dc circuit breaker.

Jake,
As far as I am aware, the transistor that John is burning up is the oscillator transistor Q3. I have had only one of these ever get "too hot" and it was because the HV transformer must have had a shorted turn somewhere. Difficult to know where because you destroy it getting it apart again. Not too sure if a shorted turns tester would detect it but do not have one anyhow. Put a newly wound one in it's place and everything back to artic conditions.

As far as I can see, this oscillator would work harder the more load was palced on it. If previously the oscillator runs fine, then it sort of points to self-oscillation where the oscillator is out of control, evident by the array of sparks John is experiencing.


John,
In itself, the hall effect switch works fine. The oscillator for most part works fine and as you said many times before, you have been using this as a HV source for years without any problems. It is compact which suits the application which all points to a reliable unit to me at least.

You said the oscillator keeps running if you hold the magnet over the hall switch, i.e. the software, for whatever reason, is seeing the input to the micro held in a "low" state. Seems to me that if you change the arrangement of the hall switch to what I had suggested, i.e. one pulse per high to low transition, you would only ever get the one pulse you require.

With the current problem, if you look at it from a logical point of view, the situation you are facing is the micro looking at a continuous low input and running the oscillator flat out as a result and trying to fix it with software. I cannot understand why the hall switch arrangement is not changed to make it just the one pulse like the micro and the software is expecting.

There may well be an interrupt on this input and I can see Nyemi's approach with the change-of-state register. It probably is a software bug but my point is, it that if you gave the micro only what it was supposed to see then the problem would go away. Everyone says it works fine except for this.

Jumping back a bit to Transistor Assist Ignition, it was a problem we used to face before micro's etc. where , if the ignition was left on and the magnet was sitting on the hall sw, it would supply the coil continuously and give the output transistor a real workout and over-heat. So we used a pulse arrangement there.

Cheers...Jorgo

gr8flyer55

Jake, it's not the transistor on the timer board at all, but the MJE521 on the ignition board being destroyed by the fact that at less than an idle speed or engine not yet running, the signal to the SCR gate is not turning it off. As it is doing this, it acts like a dead short to the circuit, which in turn makes the MJE521 turn on too hard heating it up while it is trying to pump more and more current into the transformer primary winding. It is a thermal runaway situation, where the transistor gets so hot it can't turn off. I registered over 5 full amps of current draw from the transistor which is past the rating of 4 amps.
This problem only popped up after the last version change. Take a look at the "engine running?" lines to find what is different. I'm sure you'll see the problem in that area.

John

bluejets

John,
Bit of confusion there sorry. Forgot you were using a different HV board but the same applies anyhow.

Saying the SCR is not turning off I'm assuming you mean it is running at exceptionally high frequency.

SCR can only turn on once because of cap in the gate lead and then off again when it dumps the HV into the coil.

If running at high hz then yes, getting a high hz signal.

You did say it is only when magnet is over the hall switch, is this still the case, then and only then...??