I/O accuracy

Hardware inside and outside of the ECU
Post Reply
Mad Max
Posts: 187
Joined: Wed Nov 13, 2013 12:34 pm
Location: Ukraine, Kharkov

I/O accuracy

Post by Mad Max »

I want to discuss about different ranges accuracies and limits for input and output ECU signals.
Because this is very important in future project.

So. There is my vision based on Subaru, EVO and some other non stock ECU:
1. Fuel injector pulse width 1- 40 ms 0.1 ms step
2. Ignition total timing -15- +55 deg 1 deg step
3. Coil charging time 3-10 ms step 0.1 ms
4. Manifold relative pressure 0- 3 Bar (1-4 bar absolute)step 0.05 bar
6. Maf 0-5 V step 0.02V
7. TPS 0-100% step 1%
8. Coolant temperature -40(Siberia) to +120 C step 1deg C
9. Intake air temp -40 +100 step 2deg C
Skype- max.mad486
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: I/O accuracy

Post by kb1gtt »

I tend to think a bit more in terms of crank angle, with a desire to hold .1 degrees of crank angle. I also tend to want to see one edge of fuel and spark events being synced with that .1 crank angle. So I tend to look for 9kRPM /60S/M = 150R/S --> 1/150R/S = 6.66mS/R. Then 6.66mS/R /3600deg/R = .0018mS/.1 deg. So I tend to want one edge of the pulse to be synced around .0018mS. After that the pulse lengths are far less critical.

-- Hmmm, at high rev's with 4 stroke, a pulse longer than 13mS will be 100% duty. 40mS might be a bit long in the tooth for the spec. I would look for a pulse between 1mS and 13mS for the injectors. A 1mS pulse often doesn't flow fuel. Also for 2 cycle I would look for 1mS to 6mS.
-- For ignition, a charge time of longer than 6mS also seems a bit long in the tooth to me.

-- Coolant temperatures aren't really 1C accurate. The sensor may appear to be 1C, however they are typically a fair bit off. This doesn't matter that much as the adjustments due to 1C even 10C are minimal. We should aim for 1C accuracty, but even 10C generally won't change your fuel delivery all that much.

-- Air temp sensors also lag the real world, so 1C is probably tighter than required. What the sensor tells you and what is reality are significantly different.

A very small thermocouple will have one of the lowest reaction times for sensing the real temperature. From this graph you can see that's this tiny thing is still only 63% accurate after 2 seconds. To get 99% accurate you have to elapse 5 time constants, so 10 seconds before you have a reasonably accurate reading. If a more accurate temperature sensor is needed, I'd suggest one that doesn't have a thermal mass, like optical. If you have a large brass slug like the typical engine thermistor sensors, the reaction time is going to be much slower. So I would like to caution, that even if you think you have 1C accuaty, you probably are not that accurate.

http://www.omega.com/temperature/Z/ThermocoupleResponseTime.html
Welcome to the friendlier side of internet crazy :)
Mad Max
Posts: 187
Joined: Wed Nov 13, 2013 12:34 pm
Location: Ukraine, Kharkov

Re: I/O accuracy

Post by Mad Max »

we need more than 13 ms for cold start.
I don't remenber exactly but at -20 C I have on my Subaru pulse width about 30 ms to start engine normal.

1 deg of coolant step we need only for radiator fan controll. (98C on- 96C off).

Air intake temp - really we need step not less that 5-10 deg C.
But I think subprogtamms will be same for coolant and air temp calculations. So there is no reason make 2 different routine.
Any way this is programmers choice.
Skype- max.mad486
Post Reply