Air and Coolant temp sensor scaling

[Sprocket]

I am tryig to configure the coolant and air temp sensors, and again im finding it difficult to understand the process. Can anyone help please.

The coolant sensor measures aproximately 2050 ohms at 25c and the air temp also measures roughly 2050 at 25c. These are standard Rover sensors.

What do i need to do? I know the coolant temp is not reading right

[rob@vems.co.uk]

Have you tried tempsensor_upload?

If not :-
In a command window in the root directory of the the VEMS firmware type:

download-config
tempsensor_upload 2252 2252 256

Then take a look at how your values are looking.

[Sprocket]

I'll give that a go

[Sprocket]

I have as well as uploading the 2252 airX and 2252 clt hex files, i have upgraded to 1.0.73 firmware from 1.0.71.

However, the temperature still reads out the same amount, so, it would appear that the 2252 hex files were already in use.

After reading the Easytherm http://www.vems.hu/wiki/index.php?page=EasyTherm pages, (still a little confusing) i found this line near the bottom of the page

'For the classic 2057 Ohm (@T=+25C) NTC type sensor (often closer to 2063 Ohm in practice though difference is negligable), 2063/263 (or 256 or 268) is best match'

This is what i need!! the sensors read 2050 at 25c. Only problem is that the 2063 hex files do not exist on my computer, i have searched, i have also searched the Wiki, and cant find it anywhere.

Note also on the Easytherm sensor table page http://www.vems.hu/wiki/index.php?page=EasyTherm%2FSensorTable , that there is no link to a list as suggested see this line 'Here are a list of temp calibration files for different cars'

Can some one  help

[Sprocket]

Scratch that, I found it here http://www.vems.hu/files/Firmware/release/ntc_ptc_hexpatch.zip


I suspect im only interested in the 256 files

I'll give it a go shortly

[Sprocket]

Ok so i uploaded the 2063_256 sensor factor tables and everything read right at room temp, I had the CLT MAT, EGT1 and my calibrated digi thermometer read 22

Upon imersing in a cup of boiling water and allowing the temps to settle out, the coolant temp sensor reads out low by 5 degrees at 75c. It would apear that my sensor is FUBAR'd. I have ordered some new ones, and shall repete this test.

Looks like im finally getting to the bottom of this

[Sprocket]

Problems problems

I have new sensors, and they read pretty much the same as the ones i had. I have spent hours with boiling water watching the temperatures, i uploaded new sensor scales to read correct at above 60c, that caused the lower temps to slowly read out, eventualy by 5 degrees, I uploaded sensor scales to read correct at 40 and the sensors read out at both high and low temp, I uploaded scales to read correct at 20 and the high temps read out by 5c.

I think i need some new hex files specific to these sensors, Easytherm? whats required. I found this on the Wiki, so there is some truth in this. http://www.vems.hu/wiki/index.php?page=MembersPage%2FRichardBarrington%2FRangeRoverClassic%2FConfig

I think this is screwing up the speed density algorythms, may be why when the engine gets warmer it starts to run rough and the EGC starts to run away??

[rob@vems.co.uk]

You can make and patch using EasyTherm, dnb has done a load of work on this subject, you can see his write-up here:
http://www.vems.hu/wiki/index.php?page=MembersPage%2FDavidBlades%2FSenTest
"I recorded the temperature reported by the ECU for a selection of known resistances with the original GM curves compiled into the firmware. This let me know what A to D bin each resistance fitted in, so I could then build correction tables using an excel workbook for my sensors and compiled these into the firmware."

So it might be worth giving him a nudge for the spreadsheets.

Rob

[dnb]

EasyTherm can be made to produce good curves.  It can also produce some extremely silly ones...  The Steinhart-Hart algorithm it uses is not ideal for thermisters that are used over a wide operating range.

The trouble is that you can't tell the difference a good and bad curve by simply looking...

Coolant temperature isn't all that critical.  I'd be happy with 5 degrees error when cold and say 2 or 3 when hot.  Intake would be better if it were 2 to 3 degrees all the time, but a constant(ish) offset from reality won't hurt - you'll more than compensate for this in the VE table, lambda table & closed loop control.

What have you got the "fake airden low limit" parameter set to in the config?  This caused me no end of problems.

The spreadsheets Rob is refering to are indeed available, but I'd need to comment them so they make sense to someone seeing it "cold".  Not a problem, but it'll take me a couple of days to do. 

[Sprocket]

Interesting,

still dont get the easyterm bit, lol.

The only factual information i can get hold of on the coolant sensor is that its a Lucas jobbie and 2000 ohm at 25c and 182 ohm at 100c, so if you transpose those points onto the 2252_256 curve it shows that it reads high at 25 and low at 100c, the curve is not as steep.

what i was getting at when commenting on the speed density algorythms, is, when the engine idles, the temperature rises, as does the under bonet temperature. As this happens, you can watch the ECG start to correct and the engine does start to run a little rough. I dont think just letting EGC sort everything out is the right way to go about it   You can tune the VE table for normal running temp and warm up enrichments for cold running but if the engine temp rises the algorythm kicks out the wrong response and then EGC corrects its mistake, how will this affect acceleration enrichments where EGC is not switched on?

One thing that is odd, is that since fitting VEMS, the engine temp at idle seems to rise quite quickly and the electric fan kicks in, this is a thermostatic switch, not the fan control on Vems. when driving the coolant temp reads 78c yet there is an 88c stat fitted. Temp gauge is now controled via a standard temp sender where before it was controlled by the ECU, nothing tallies up hence why i am talking here.

Everything will be back on the car and new, including the thermostat, gauge temp sender was replaced about 8 weeks ago, as that deffo was reading out, the gauge never read above 1/4, lol.

I cant easily fit any other sensors as the holes are all the wrong sizes and there just isnt enough space to fit them anywhere else, and im not taking the head off just to drill a hole for a sensor

I'll have a look at this easythrm thing again tonight, other than that i'll have to use the 2252_256 hex files for now as i have been doing since the begining, even though its not right, i just dont have faith in it

Just want to get to the bottom of it as its a bit daunting not knowing the true engine temperature.

[dnb]

Factual information is over rated...

Boil the temperature in a pot of water on the hob and measure the resistance of the sensor and the temperature of the water.  Fit a best fit curve to this data in excel and use this as the input to the ECU. 

Simple

As for the idle thing, I sugest that either the temp sensor is heatsoaking, or you have configured the "fake airden low limit" parameter incorrectly.

[Sprocket]

LOL dont know what the fake airden low limit is or even heard of it till you mentioned it.

Please help with that one

Im off to jave a look at some config files

[Sprocket]

Is this the line i need to be looking at?

airden_ignore=00

[dnb]

Don't know - it's called Fake airden low limit in Megatune.  Not sure what it is in the tables list.

The trick for setting it is to turn it off and see if the car wants to go lean or rich when it's sat idling.  If it starts to go lean (EGO correction will be large +ve reading), then the IAT sensor is getting heatsoaked by the engine and you should set the Fake aiden limit thingy to 80 or something around there.  Then the maximium leaning caused by temperature will be 80%.  Gradually increase this towards 100% to stop the car leaning out.

The other option is to simply put the IAT sensor somewhere it can't get heatsoaked.  Probably as difficult in a mini as it is in a TVR...

As I keep saying, absolute accuracy isn't at all that important to the ECU, but it is to us who read gauges...

[Sprocket]

I would think in my situation sensor heat soak is at a minimum, the sensor is mounted in the K&N in its rubber base, id expect the air to be the same temp thus the air density to be correct with the measured temp reading?

By the looks of it I have the Fake Airden disabled.

I recon the EGC changes at idle are down to the CLT and MAT sensors reading 5 degrees different on a different angle curve resulting in fueling changes that are not representative of ctual engine conditions.

Still not satisfied that letting EGC correct it is good for all conditions, the speed density algorythm should take care of the changes in engine temperatures. Remember that you say you need a well tuned VE map for acceleration enrichments to function correctly, how can you tune this if the speed density algorythm isnt working as it should with varying temperatures, engine and ambient. Maybe the problem im suffering with AE as it seems fine one day and the next its not. LOL its a learning cure, Woo

I sat down with the chip fryer last night and set up a lab to measure the resistance/ temperatures of the CLT sensor, i downloaded easytherm and generated INC files, not sure where to go from here though, but it is deffo clear that the 2063_256 patch is a steeper curve, yet coincidently at 25c the resistance is 2063 ohms ( I see what you mean about factual information)