Code: EGR and CTS conversions

[gregs78cam]

Thanks 1P2M, I spent a couple more hours on it, but couldn't get it a lot closer, I got it a little closer, but I gave up. I didn't write down the closer one before I closed the page though. It's hard to get both ends and the middle close, It was almost one or the other, and that upper part of the curve from 80C on up a PITA. I will try again and write down the end product.

EDIT: I will get the center of the curve as accuate as I can and to worry too much about the extreme ends i.e. above 150C and below 0C.

OK, this one gets it within about 5C all the way from -30 to around 95C...I THINK! And it is the format TP 'should' recognize?

((-.055x+6.5)(-.026x+4.6)(-.024x+4.0))-((.049x+1)(.028x-2.3))+.04x+53

The upper half of the curve is just so steep it's very hard to get both ends correct.

I don't know if this page will come up with the graph and points in it already, but go here:

http://fooplot.com/#W3sidHlwZSI6MCwi...CIsIjIxMCJdfV0

It is nowhere near close above 100C, but I think if you ever get above 120C coolant temp (250F), you will have bigger problems than the dash display wrong.

[gregs78cam]

OK, this is it, I give up. This is +/-5 across the whole curve except 150C and up, with an average error of 2.72. I think the sensor itself has more accuracy issues than that.

(-.0315x+4.5)^3+(-.00045x^2)-.135x+73.

I don't know how to put it in TPv5 to make it work, that is up to you guys..?


http://fooplot.com/#W3sidHlwZSI6MCwi...AiLCIyMTAiXX1d


Moving on......

[EagleMark]

Thanks for putting you excellent math skills to good work!

[gregs78cam]

I think excellent is a bit of a stretch, buy you're very welcome,...... if it works that is.

[1project2many]

I had to fight with TP to make the formula work but I got it going. Then I remembered that I had the original installation CD for Matlab and re-installed that software and promptly realized I forgot how to use it. Then the Easter events started so that was that for the day.

I'll post the TP version of Greg's latest formula here tonight. It's closer than the original and it will probably work well enough for most. If it's not close enough the user will simply have to consult the lookup table. I probably won't get a chance to see if Matlab can generate anything closer for a while.

Again, thanks, Greg.

[EagleMark]

I have to thank all you guys for this!

Would be another advanced math class for me, while still learning the reading end of disassembly let alone the disassembly itself which has not worked out so far.

My time has been spent learning this forum software, my cgi, php and js courses of 12 years ago are really lacking in making changes needed for search functions.

Then some good weather made me go outside and get some sunshine, fix the lawn mower, cut the grass, wash my truck, do some work on my truck and if another day of good weather... polish my wheels because they started using salt on highways up here at 20f and below... my aluminum wheels have black spots!

[gregs78cam]

You know I was thinking about this issue the other day and realized that this is also why the E-Fan patch doesn't turn on/off the fan at the exact temp when viewed in the Dashboard. Because we see the temp converted by the simple (linear) equation, when the PCM is seeing the true temp generated from the non-linear look up table.

..... they started using salt on highways up here at 20f and below...

ERRRR! When did this start?

[EagleMark]

..... they started using salt on highways up here at 20f and below...

ERRRR! When did this start?

All this winter and half of last. State Highways only, no city streets.

[1project2many]

Mark, everyone here plays a part in what gets done. You've done a helluva lot of organizing and cleaning up and imo that's worth a bunch. And you're not doing too badly with the code stuff so don't worry. I spent many years on it.

[gregs78cam]

Mark, everyone here plays a part in what gets done. You've done a helluva lot of organizing and cleaning up and imo that's worth a bunch. And you're not doing too badly with the code stuff so don't worry.

I agree. Thank YOU Mark.

[EagleMark]

Thanks guys! It's always been a great team here!

[1project2many]

Just got back to this.

Here's a version for entry in TP:
(-.0315X + 4.5)(-.0315X + 4.5)(-.0315X + 4.5)+(X*X*-1*.00045)-.135*X+73.00000

You might have to type it in by hand. TP didn't like it when I copy / pasted it from another program.

FWIW (not to take anything away from Greg's work) I've found the *real* equation for resistance vs temperature. It's called the Steinhart - Hart equation and you can find plenty of information online starting with a Wikipedia entry sure to confuse anyone who doesn't think big equations are fun. Someone involved with Megasquirt worked out A, B, and C values (see online if curious) for the GM temperature sensors and with a slight correction it produces correct temperature values for a given sensor resistance value. I'm not sure it can be made to work in TP though. It uses logarithms which may not be supported by TP. The first step is to determine a relationship between A/D counts and resistance and modify the formula so I can produce correct results in Excel.

As an aside, this equation and the non-linear nature of the resistance vs temp chart is the reason why Megasquirt guys developed the small piece of software which can generate a lookup table for a temp sensor given three resistance values. They also chose to skip trying to force the ecm to do this somewhat tricky math and opted instead for simple lookup values. Very interesting.

Mark, you seem to have a way of finding the right question to start a long quest for answers. Keep it up.

[1project2many]

Found interesting info related to the "two step" temperature measurement strategy mentioned earlier:
wps.prenhall.com/wps/media/objects/6568/6725731/ppts/Chapter14.ppt

Slide 9 shows a circuit using a 1K pullup resistor.

[EagleMark]

Just got back to this.

Here's a version for entry in TP:
(-.0315X + 4.5)(-.0315X + 4.5)(-.0315X + 4.5)+(X*X*-1*.00045)-.135*X+73.00000

I gave this a shot today and choose this entry from disassembly.

 LD2C8   FCB  218    ; 10c, Cool TEMP THRSH FOR COLD START 
                    ; (A/D 1K PU)

Verified hex at $DA and decimal 218 which is 10C on AD count chart. But temp in TunerPro was 35.59c ?

[gregs78cam]

Nevermind, graph doesn't like it either.

The equation 1P2M came up with plots properly in the graph. Don't know what TP problem is.

(-.0315x + 4.5)(-.0315x + 4.5)(-.0315x + 4.5)+(x*x*-1*.00045)-.135x+73.00000