Everyone the code for the new fix

[comsoft]

Hello everyone.

we've decided to release the code for the new fix which alber35 and I worked on - for free.

Hope one of the chinees compenies will make a pcb of it, since in my country is very expensive.

since i couldn't upload HEX file, i added ".doc" extension. please remove this extension before burning it to pic.

good luck
enjoy

[loy]

Nice hope theres a newz post on this

[comsoft]

Everyone feel free to combine the fix with any combination and post reports.
Even though i think the combination with the capacitors is great one.

good luck

[loy]

http://www.console.ombouw.com/v12/protectivecircuit.asp is the diagram for their hex

[comsoft]

Thanks for the help , loy

[cakcetin]

Thank you very much Comsoft and Albert35 for this kind share....
I will must try it and report here.

[comsoft]

No problem.
We are open for sujjestions for forther improvments.

afther reading the "ripper dead beef" I was thinking my self for deploying the code for a chip with higher clock rate to make it even more reliable, but it will cost abit more.

but i'll do that only if i have too, meanwile i know the pic in current state working great agains mechanics contoller lockups.

I'm opened to any remarks on that fix.

thnx

[gee-man]

hey comsoft, I couldn't help but disassemble your code and see how it works - i'm a hardware hacker by nature

Is there any reason you're not using the PIC's timer (T0CKI pin) to count interrupts? it seems you're only using BTFSx instructions to poll for activity on the interrupt line, which is very likely to miss an edge. The PIC's TOCKI input can ideally detect a ~500ns pulse, and the ripper gang's V12_DEAD_BEEF document mentions that the interrupt lines are 5us in length.

Overall, it seems like a more reliable way to go.

[comsoft]

gee-man: now u can confirm that my code is nothing the same as matrix code and it's not a copy of it

Yea, u have a point there, but the MECHACON INT line is always active so if i miss a 5us pulse it's not that critical, since the code take a 200 ms of NON ACTIVITY on the INT line as a MECHACON crash!!!

the code is based on the idea of matrix team that there is a loop that constantly monitoring the input line and it check it and recheck it for every 1ms.

basicly at start, we didn't think of changing the idea of how it works, just improving it (by changing the code and optimizing it) and also add the extra fancunallity.

Now for u'r idea.
if u connect the MECHACON INT line to internal COUNTER then u'll have to replace the current watchdog loop with a loop that count 200ms and will check at each 200ms if the counter is ZERO which means NON ACTIVE LINE. and at each start of the counter loop there is a need to RESET the counter.

if u think that will work better, i'm willing to improve and change my code and release a 3rd verstion of it

but i think it would be better of using a better microchip with INTERRUPS support. for example a timer interrupt and external input interrup will help in this case. so we need to find PIC with fanctionallity as the 16f84 but in 8 pin case.

and if we find one witch has internal clock of 20MHZ insted of 4MHZ then we can run the code faster and improve it. (we don't wan't the installers to add resonator to the fix do we )

changing the pic will make it cost more then 1$ each, but i think it will be worth it.

[R0Bzombie]

Hi ,
I just designed a pcb for this fix. I'm waiting for a quote from the guy that will make my pcbs.

If it's cheap enough, I will sell them with no profit.
It should be around 2-3$ with the components on it.