C270CDI limp mode - coolant temp sensor???

[JoeHorner]

Trying to get to grips with the limp mode / no limp mode on the C270 without finding someone with Star diagnostics around here (apart from main dealer....)

After the "cured / not cured / repeat" this afternoon I decided to get out and record some data thus evening.

Sure enough, it was out of limp mode again when I went out to it. Engine was cool / cold to touch but coolant sensor was reading 62 degrees C (seems a bit high after 3-4 hours standing!)

Drove for about 10-12 mins in a "spirited" way, then suddenly limp again.

The only thing that stands out on the recordings is that temp reading never went above 65 degrees and, just as limp came back it dropped 5 degrees in 45 seconds. Then rose 7 degrees in 90 secs then dropped again in about 2 mins. That seems like unfeasibly fast variations to me and, even with a stuck thermostat hovering around 62 degrees driving hard in 20 degrees ambient seems unlikely.

So, it got me thinking. Does anyone know if the CDI system is smart enough to realise it's getting unreliable temp data and limit power as a precaution?

 

[onefortheroad]

I've owned cars with totally knackered thermostats and none have ever caused limp mode.

 

[JoeHorner]

I've owned cars with totally knackered thermostats and none have ever caused limp mode.

Yes, although the ECU would have trouble detecting that because it could just be running cold. Whereas implausible swings in recorded temp would be easy to spot if the ecu looks for it.

Nearly 11 litres of coolant losing 5 degrees in under a minute, in an engine that's still running and producing hear, is firmly into "cannae change the laws of physics" territory! So a reading like that either means faulty sensor or "you've just lost all your water & this is now air temp".

Sudden drop in temp gauge was always a useful warning on old cars for exactly that reason, whether the ECU designers were aware of it I don't know.

 

[mersum1es]

I would suspect some voltage/grounding problem...

 

[Droverunner]

Our 220CDi coolant sensor in the stat housing area failed causing full limp mode. Didn't know it was the issue for certain until the following day with an outside temp of 26degC before starting the engine the diagnostic showed a coolant temp of 102.6degC. £15 for a new sensor from main dealer and the car was back to normal. On ours you could see the dash temp gauge bouncing.

 

[supernoodle]

If you have no fault code you don't have limp, but if the temp measured gets too high it will start derating the engine. That's an attempt to try and reduce any further temp rise.
Of course there comes a point at which it would flag a range error, maybe something like 140deg, which it knows wouldn't be possible.
Be wary with live data, especially that from a generic tool that uses OBD mode $01. The temp reported there is often filtered and if a fault is detected could be the replacement value.
To know what's going on you really need a tool that supports manufacturer specific data and would use mode $22. That should have the sensor voltages too and maybe even the raw temp before any filleting. Volts are the best thing to look at, they are the rawest form. You may not know what the translation of V to degC is but you can see if the voltage changes are plausible. As you mentioned above, temps can't change that quick.
If you see any spikes that definately isn't right.
There is usually a code that would flag if there are implausible voltage changes to catch dodgy wiring, but don't know if MB use it here.

If you're feeling adventurous you can test the wiring side by making your own test temperature sensor. All the temp sensors are effectively just variable resistors. You can buy one for a couple of pounds, solder two wires, unplug sensor, poke the wires into the harness and sweep from one end to the other whilst watching the live data.

 

[Droverunner]

Ours had no fault code but it restricted engine revs and power plus the operation of the gearbox... that was near enough to "limp mode" for me whatever you might call it.

 

[ajlsl600]

If you have no fault code you don't have limp, but if the temp measured gets too high it will start derating the engine. That's an attempt to try and reduce any further temp rise.
Of course there comes a point at which it would flag a range error, maybe something like 140deg, which it knows wouldn't be possible.
Be wary with live data, especially that from a generic tool that uses OBD mode $01. The temp reported there is often filtered and if a fault is detected could be the replacement value.
To know what's going on you really need a tool that supports manufacturer specific data and would use mode $22. That should have the sensor voltages too and maybe even the raw temp before any filleting. Volts are the best thing to look at, they are the rawest form. You may not know what the translation of V to degC is but you can see if the voltage changes are plausible. As you mentioned above, temps can't change that quick.
If you see any spikes that definately isn't right.
There is usually a code that would flag if there are implausible voltage changes to catch dodgy wiring, but don't know if MB use it here.

If you're feeling adventurous you can test the wiring side by making your own test temperature sensor. All the temp sensors are effectively just variable resistors. You can buy one for a couple of pounds, solder two wires, unplug sensor, poke the wires into the harness and sweep from one end to the other whilst watching the live data.

I do something similar with resistors to trans temp sensors harnesses that while showing faulty sensor prevent me from doing clutch fill calibrations. I just solder in until I get one that kids the diagnostics that oil temp is in range

 

[JoeHorner]

Yep, I know it's not "official" limp mode but, as Droverunner says, zero boost and "stick in highest possible gear" (will allow manual override but won't change down itself) is as close as it gets in practice!

Was in that state cold this morning (coolant & intake tems agreed at 20 deg). Then, by the time I'd gone (in the 5 series) to get some contact cleaner, it was back to full boost & kickdown.

Cleaned every connector I could get to (none looked bad) & it ran great for near enough exactly 20 mins again, then back into no boost at the same junction as yesterday.

Obviously need a better scanner but that can wait while the rest is done.

 

[ajlsl600]

I have mb xentry. Currently with lost kiwi and icarsoft mb v2 which recently gave me cause to be impressed with. I get back in car in 2 weeks. We see if it still does.

 

[JoeHorner]

I have mb xentry. Currently with lost kiwi and icarsoft mb v2 which recently gave me cause to be impressed with. I get back in car in 2 weeks. We see if it still does.

Having taken ages to set up an old Dell laptop (with proper serial port) to talk to my '93 5 series I'm not in a great hurry to start with more dealer software. I'm not scared by tech but, as I get older, I prefer it to just work out of the box where possible!

Have been looking at the icarsoft kit with interest but can't find anyone who's used it on the 270cdi to confirm how effective it is.

 

[Droverunner]

For some years I owned a 5-series and had both the dealer type diagnostic on the laptop and an iCarsoft handheld. The laptop software was always glitchy and went deep into areas that were no interest to me. Same with Ford software on a laptop. The iCarsoft did the job every time though and was easy to use. So when Mrs D bought her 2007 CLK 220CDi some 5yrs ago I bought the Mercedes specific iCarsoft i980 and that's been brilliant and met every need for the advanced DIY user. When I bought a Viano a year ago I bought an iCarsoft V2... in truth it's been no better than the i980 for the jobs I've used it on other than a useful increase in its size.

 

[supernoodle]

I do something similar with resistors to trans temp sensors harnesses that while showing faulty sensor prevent me from doing clutch fill calibrations. I just solder in until I get one that kids the diagnostics that oil temp is in range

You can even simplify even further and just feed a voltage into the sense pin of a temp sensor. 2500mV would be mid range and should be valid.
In fact you can do the same with pressure sensors too.

 

[JoeHorner]

No time to play the past few days so have had to content myself with trying to think through the logi of what's happening.

The best I can come up with, given the "reliable" timing if 10 mins from cold is a sticking / faulty EGR valve.

Lots of posts saying that can disable the turbo and the timing seems about right if, as normal, these don't try to activate EGR until the engine's warm.

So Friday's jobs (other stuff permitting) are to change the aux belt, replace the disks and pads, and clean out the EGR & pipework.