DIYEFI.org Forum

XGATE isn't a GP CPU, it's purely interrupt driven, and in order for the CORE functions of our system to work their best, the core will be restricted to interrupts with very short execution time. This is most definitely an extra function, not a core function, and thus it is unacceptable for it to severely compromise core functionality, and therefore, given it can't be done well if it takes a back seat to normal EMS functions, it belongs on its own dedicated core, external to the EMS. Ditto transmission control, wideband control, etc etc.

hi fred
have you had a look at the maths involved in finding ppp from the paper i mentiond it is a moving window non linear least squars problem working backwards from the end of signal aqusition. i have not looked at how you guys have set out your system as i am not looking at folowing your system but willing to shair any information i come accros as i messaround with my own stuff when i get board. i was thinking along the line of using spark duration detection to detect end of spark event signaling start of aquisition of ion current signal then sensing wether signal has droped below a threshhold value
stoping aqusition. the xgate wich i am looking at the xe wich is a risc processor operating at 100mhz with some clever programing and a reduced sampling rate again reducing calc time should be able to achive this i dont know what you have got the xgate doing i have not looked at the rest of this site i know it is good for timing but i dont know what its timing load idle time i understand it is a interupt driven but it will lend itself to a limited dsp (frescale even mentions this ) fillter all you are looking for is the location of ppp there is no bellcurv fitting no re interation trying to make it fit no huge amounts of dat yes this way will consume more memory.
i do have diferent goals than you guys im just sharing my mind and what i have seen and done (most of the stuff i have worked on is machine control) involving gps accelorometers rtos and upto 110 ton excavators

if i cant fit it on to this style of system i will go to a larger 32 bit system
as i want to build a compact system i am actualy looking at diesel systems
but considering what i have stuffed into slower and smaler systems and kept it all in real time looking at this system and i have options about how i set up my system. although a 32bit system would allow me to fully expliot the information that is containd in ion sens data eg lambada values witch would alow me to do any mods to the engine and have it auto tune

you can still do all of that with 8bits but, I do understand how luxurious a 32bit platform can be .. especially when the prices are dropping like a rock.

how do you derive lambda from ion sense data ? could you point me to a paper ?

well the pressure is a result of the tempriture of the gas from this information an a math moddel of your engine you can calc the amount of energy relesed during combustion compair this to the amount of fuel that was injected as we know the calorific value of the fuel and we can see if all the fuel has been burnt or not (this is a over simplifcation) but you get the idea and there is lots more data in the ion sens signal including combustion turbulance

too experimental for me .. I'll just measure it and be done with it. Sounds like I'd have to measure it to come up with an empirical model anyways.

yes it can be complicated it is intended to have a cylinder by cylinder tuning aproche and reduce lag in the system as current sensors sens all cylinders
so you could have a lean cylinder and a rich one but the sensor would see it as ok becaus the sum of all may be fine a lot of these processes were created during the late 90s as part of the saarb sponserd tests there is even one paper out there about using ion sens to detect leaks in the exaust system!
it is based on how well your cylinder fills

the system i eventualy want to develope is very diferent from a alpha-n system or mass flow system like i said i am looking at diesel but for now will mess around with petrol looking at multi layerd head gasket for ion sensor (diesel without glow plugs eg cummins isb 5.9 ) have used isoleted injector but is big hedach lots of live components looked at isolated valve but there will be mec interferance looked at inductive voltage on the valve stem but the engine i was had did not leve much room it was 4 cyl ford mon eng

i did get the head gasket to last and got some information from it untill it started to get carbon biuldup and shorted the sensor new des when i get heps of time will include a heater strip out of nicrome or make it out of some catalitic material to combust the carbon biuldup perhaps but i see problems geting it to the right temp

Here are some papers I found courtesy of IEEE.