Idiot lights bother me. Gauges disguised as idiot lights bother me more. Mazda did this with the oil pressure gauge in the Miata in 1995 -- instead of a real pressure gauge, it's an on/off switch with a ton of hysteresis. According to the stories, they did this because dealerships were getting too many calls from customers concerned that the gauge freaking out on their dash was indicative of a problem.

The 991 has a similar problem with the coolant gauge -- both analogue and digital. Many people have noticed that it goes up to 194f (90c) and stays there until it overheats, then spikes to to whatever crazy overheating temperature.

I don't think a "driver's car" at any price point should ever do this kind of stuff. The GT cars do not have this problem, so Porsche obviously sees the non-GT cars as targeting people who do not care or understand this information. Admittedly, this information is entirely useless for street driving, but it still doesn't sit well with me to mask information from a user.

After asking in the above linked thread, as well as reading in the APR 991 tuning thread that the boost gauge is adjustable, I figured it just had to be a programming thing. So I started digging around and found a lot of cool stuff -- most notably, I changed my temperature graphs to display 38-135 instead of 60-155, and lowered the 'red' area on it. Other things like the timeout for the gauges to turn off when you lock the car (default 240 seconds), various menu options like the largely useless torque/power display, MPG readout, G-force gauge automatic reset, and much more. I even found settings around DRLs and rear axle steering chains, so that was neat.

You should be able to perform this operation yourself with any tool that can program your car like a PIWIS -- I used an Autel MS906BT, though any other Autel MaxiSys should be able to do it, as well as the Snap-On equivalents to those Autel units. According to several people, the ThinkDiag+ dongle and phone can do this coding, too!

What we're going to focus on here is the Coding menu for the Instrument Cluster -- on my Autel, this is accessed by Coding (or Coding Without Rules, or Online Coding, depending on where you enter diagnostics) and choosing Instrument Cluster.
There is an option called Kuhlwassertemperatur kennlinie which translates to Coolant Water Temperature Curve. This appears to be an input/output map for water temp display.

Within this coding menu you have 16 parameters: X1 through X8, and Y1 through Y8. From what I can tell, the X axis is the input value, and the Y axis is the output value.

On my 2017 C4S, the default values are:
X1: 36.00 Y1: 36.00
X2: 50.30 Y2: 56.00
X3: 60.00 Y3: 68.00
X4: 80.30 Y4: 90.00
X5: 110.30 Y5: 90.00
X6: 114.80 Y6: 105.00
X7: 118.50 Y7: 115.00
X8: 125.30 Y8: 125.00
(all values in celsius)

You should be able to immediately see the problem here. Values between 80.3c and 110.3c are mapped to display 90c. From 110.3c to 114.8c the gauge will jump 15 degrees from 90c to 105c. The next four degrees jump 10 degrees from 105c to 115c.
To illustrate this, have a chart:


The screen looks like this on my Autel unit -- pardon the cellphone picture of an LCD, it's just too annoying to screencap and upload from that device.



Every story I've read about the coolant gauge during overheating has been "it said 90/194 until the overheating error popped up, at which point it was suddenly pegged!" -- I'm guessing the sharp, sharp rise was missed by the driver (who is staring at their gauges, come on) so by the time the overheating temp was hit it was already far above the 110c dead zone.

What's important is that the car's overheating temperature alter does not appear to be dictated by the display temperature. I say this because while I was messing around with the map, I set all the Y values from Y4 to Y8 to be 90, which resulted in the temperature needle and display gauge to max out immediately... but there were no overheating warnings.

I made the probably incorrect assumption that the X axis is a calibration of the sensor, and the Y axis is the gauge read-out of the sensor, but at this point I'm pretty sure it's input/output and these values have no bearing whatsoever on the coolant temperature on the CANBUS. Even so, I have not and likely will not be touching the X values. Only the Y values.

To get an accurate coolant temperature gauge, all you need to do is set the Y values to their corresponding X values. That means Y2 through Y8 in my case were modified to their corresponding X2 through X8 values -- in the graph above, this would stack the line.

Without even restarting the car, this changed the readout on the cluster. I verified it with a separate OBD2 reader giving live data for the coolant temperature, and it's accurate within about a degree. The OBD2 reader seemed to be just a tick higher than the cluster, but there is still a tiny bit of hysteresis in the cluster (which you can adjust elsewhere these menus) -- for all intents and purposes, my cluster is accurate from 74C to a little over 100C where I monitored it during testing.


Here are a few pictures with the analogue gauge slightly in view -- both the digital and the analogue gauge use this map, so making this change will make both accurate.

The 192F picture is switching to sport mode from normal mode. It actually undershoots the 90/194F set point by a degree or two before it settles back up to 90/194. This is capturing that.
The 198F picture is either switching to sport mode from normal or the other way around. This is in middle of the sweep down to the lower (or higher) temperature -- something that is normally hidden from the driver. The OBD reader showed the same temperature the whole time.
The 209F picture is back in normal mode and relatively settled. It was pretty cold out so it didn't get much warmer than this. Again, it matched the OBD reader's display.

 

You had me at “hysteresis”. Nice writeup.

DaveGee

 

Thank you. I'm going to re-read your post a few hundred times until I understand it.
My 2013 C2 accurately reported engine temperature until it received an "update" at the dealership...I' asked them to return it to the prior state to no avail.

I am hopeful that I can figure out to to do this ...thanks again.

 

Auto manufacturers like to play this game with fuel gauges too. Your average driver doesn't want to see the fuel gauge dip below full accurately after filling up so they programmed the gauge to be dimed out at "FULL" a bit longer.

Had an old Pontiac that was really annoying for this. The fuel gauge would be stay at full for quite a while then sudden jump down to less than 3/4 full. Always annoying at the start of a long road trip thinking you are departing with a full tank only to see you 'lose' 1/4 tank of gas minutes after entering the freeway.

 

Wow great post. I had no idea the 3rd party tools had that much level of control. I want to do this asap. Anyone know a shop in the Tri-State with an Autel?

 

Very thorough post. Thank you. Personally, I watch oil and not water. I wonder if oil is the same? I've noticed that both track closely - in fact too closely in my view. I'd used to water going up much faster.

I'm also pretty sure I understood this correctly. This temperature mapping affects both the gauge and the digital display, correct?

Thank you. Happy New Year!

 

Great to see more experiences with coding for the 991.

The extent of coding with BMWs is huge, and nice to see inroads with the 911.

 

In my experiences this is actually a hardware limitation, not a gauge being dumb. Fuel level senders typically have a float on an arm, and if the fuel tank has bad geometry (perhaps like your Pontiac, definitely like my girlfriend's Golf) it physically cannot measure above a certain fullness. Or it could be sticking.


My water, even before this change, heats up a lot faster than my oil. I have a .2 4S with a PDK though, so the water cools the turbos and the PDK which could definitely cause some disparity versus something like a 991.1 7MT.

I did notice that the oil temperature gauge didn't match OBD data -- the gauge was about 10 degrees off, and continued to be as the car cooled down. Not sure if the sensor being read is different or if there's some futzing going on with the instrument cluster like with water temperature. I have never noticed a dead spot with the oil temperature readout, it's always fluctuating, so I imagine the OBD reader is reading from a different sensor or there's some kind of calibration/offset for the gauge. I'll be looking into this later.

 

Great information in your post, and thanks for digging into this. I have a .2 GTS with 7MT and also notice the water heats up a lot faster than the oil (as reported by the gauges). My car also reports a flat coolant temperature reading in normal street driving.

 

Does the gauge and digital readout both track to the same conversion table? I.e. they move exactly together?

 

This is what is also observed with multiple cars that I have owned that had aftermarket water and oil temp gauges (with separate sensors). A separate debate is whether you monitor oil temperature or pressure.

Water temperature rose much more quickly than oil temperature.

Oil temperatures also varied more under WOT/ load than water temps which definitely lagged behind both in rising and dropping.

 

Yes, both are fed the output from that table.

 

Nice write up! My 2014 Cayman coolant temp seems to be more active than my previous 997.2 911's coolant gauge. This is an interesting modification! thanks!

 

Ok since we're on the gauge topic can or could someone explain the temperature differences between the Ambient temp, Oil and Water temperatures.
The car has been a been parked in the garage for two days, why are all three temperatures off the by 3-6 degrees ?
If the ambient Temp is 66* shouldn't the Oil & Water, be closer or the same ?

 

I would say that oil reading does not have an artificially shelf or flat spot. I have seen my oil all over the place when driving is sport and pushing the car.