«Calibration of the Temperature Gauge Ed Sowell June 18, 2007 Background Whenever the XJ-S driver sees the dash temperature gauge creeping up higher ...»
Calibration of the Temperature Gauge
June 18, 2007
Whenever the XJ-S driver sees the dash temperature gauge creeping up higher than he/she is
comfortable with the first question is generally “Is the gauge right?” One way to answer this is to
make some kind of independent measurement. Some have gone so far as to fit additional sensors
to the engine and mount a separate gauge or gauges in the car. More commonly, an infrared
radiation (IR) “gun” is used to measure temperature of various surfaces on the engine, e.g., the top radiator hoses or thermostat housings. Presented here is another method that’s not very difficult or expensive.
Note that the gauge discussed here is the “barrel style” gauge used in the XJ-S from 1975 through 1990.
The Approach The approach taken here employs a two stage calibration process. First, the sender for the gauge (part number DAC 2583) is removed from the car and tested on the stove top, producing a curve of resistance vs. coolant temperature. Then the gauge itself is calibrated in the car by substituting an adjustable resistor (a potentiometer, or pot for short) for the sender, producing a table of gauge position vs. resistance. These two calibrations can be used together to give a table of coolant temperature vs. gauge position, or the reverse.
I carried out these calibrations for two different senders and two different gauges and used the results to get curve fits of the components individually and for the sender and gauge working together. If your sender and gauge are in good working order the results can be used directly since the differences between the components I tested were relatively small. Or, you can use the methods I describe to test and calibrate your components specifically. If you do some tests of your own, carefully following the procedures I describe here, I would be grateful if you would send them to me.
Results The final results are presented in Figure 1. Temperature is in Fahrenheit and gauge position is in “needle widths,” or NW. The center of the gauge was defined to be 0, and the top and bottom “tick marks” on the barrel are +11 and -11 NW respectively. This puts the C, N and H marks on the instrument cluster at -10, 0, and +10 NW respectively.
Gauge position vs temperature G a u g e p o s itio n (N W )
Discussion of Results If you are really paranoid, you might print it out the above table and tape it to the back side of your visor or something. Most will be happy with some simple rules of thumb. Here are some
points on the scale that are easy to remember:
• Halfway between N and H is about 267 F (130 C). Post-shutdown coolant boil-off is likely.
• 2 NW above N is about 238 F( 114 C)
• Overlining N (bottom of needle at top of N) about 234 F (112 C)
• N is about 220F (104 C)
• Underlining N(top of needle at bottom of N) about 210 F (99 C) 2 NW below N is about 207 F( 97 C). 88C thermostats probably fully open.1 • • 4 NW below N is about 194 F (90 C), 82 C thermostats almost fully open.
• Halfway between C and N is about 188 F (87 C).
To put these temperatures in perspective, though, you need to keep a couple facts in mind. One is that you have (or should have) a 15-16 PSI pressure cap on the remote header filler neck, meaning that the cooling system operates at about 30 PSI absolute (about 2 atmospheres). The I’m assuming the thermostats go from closed to open over a 10 degree F range, beginning at the rating point.
other fact is that pure water at 30 PSI absolute doesn’t boil until it gets to 250 degrees F (121 C).
Thus if your gauge reading 3-4 NW above N and you have pure water in the system (which you shouldn’t) you will be at the boiling. But, since you probably have a 50/50 mix of water and coolant, the boiling point is somewhat higher. A UK Jaguar fan, Bob Egerton, has spoken with coolant manufactures in the UK and was told that “…a 50:50 solution of their product with H2O at atmospheric pressure would boil at 108 C and at 2 atmospheres at 135.8 C (281 F)”. (See http://www.far-out.demon.co.uk/cardiy/moreinfo.htm). Putting all this together, if your gauge and sender are good, you have a good pressure cap, and have a fresh 50/50 coolant mix, you are probably not going to be boiling until you get about 6-7 NW above N.
I say “probably not” because there were a lot of “ifs” in that sentence. Here are some other things
to think about:
• Post-shutdown heat soak-back. I have no specific data to offer, but I can say that at times when my cooling system was not up to snuff I have had “total loss of coolant events” in the garage when there was no sign of boiling at shutdown.
• Yet another reason for not treating the +3 to +7 NW area as free headroom is the possibility of hotspots in the engine. That is, places where the circulation perhaps isn’t as good as it should be and/or near the combustion chamber can be significantly hotter than what the sender sees.
• Even an 88 C (190 F) thermostat is fully open when the needle is about 2 NW below N.
What this means is that when the needle is on N or above the engine is well into the “free float” range, with respect to temperature. Any increase in outside temperature will result in a degree-for-degree increase in engine temperature, and any increase in engine load will result in further coolant temperature increases.
• And, finally, as Kirby Palm frequently points out, the left bank may be hotter than the right where the gauge sender is. So, it would be prudent to follow the conventional wisdom and get a bit worried when you get more than a couple NW above N.
A Simple Test A simple test of the gauge as installed requires only a 50 Ohm resistor and a pair of jumper wires. First, disconnect the sender. Use one jumper to connect one end of the resistor to ground.
Using the second jumper, connect the other end of the resistor to the disconnected sender wire.
Turn on the ignition and look at the gauge. If the needle goes to the center of N your gauge is working fine. If it doesn't, take out the instrument cluster and clean up the connectors. This will probably correct the problem because the gauge and sender units seem to be fairly robust.
Effect of Poor Ground Numerous XJ-S owners have reported erratic instruments having been traced to poor grounding of the instrument cluster or the ground connection to the gauge in question. I can show that there is a theoretical effect of resistance (such as would be caused poor grounding) between the temperature gauge terminal that is supposed to be grounded and engine ground (where the sender is solidly grounded since its screwed into the engine). However, since any such resistance would be in series with one of the windings in the gauge that happens to be about 64 ohms it would have to be quite a bad connection before a significant effect would be seen at the gauge.
I have bench tested this theory on my spare gauge. I put a 47 Ohm resistor (that I happened to have handy) in place of the sender and this brought the gauge to about 1 NW above the N when the ground terminal was connected to battery ground. I then placed 2 Ohm resistor between the ground terminal and battery ground, which raised the needle a tiny bit. I then put a 10 Ohm resister in the ground path, raising the needle by about 1 NW. I conclude that a bad ground will indeed increase gauge readings, but it has to be a really bad ground to affect it significantly.
Voltage Effects The question of battery voltage effect on temperature gauge readings occasionally comes up. As shown in the Appendix, there is a theoretical linear relationship between the gauge position (or at least the currents upon which it depends) and the voltage difference between the gauge 12 volt (nominal) supply and ground. Joe Bialy (Jaguar Joe on the Jag Lovers list) has done some bench
tests on his spare instrument cluster using a variable voltage source. Here are his findings:
“Attaching a 49.8 ohm resistor to the temperature gauge input and varying the gauge's
supply voltage produced the following results:
9.00 volts supply = needle at bottom of the N
13.2 volts supply = needle at center of the N
17.0 volts supply = needle at the top of the N Further, Joes says “Voltage between 12 and 14, which is pretty much where these cars run at, provided almost imperceptible changes in needle movement. “ So, there you have it. Not to worry, because if the engine will start the gauge will be pretty good over the expected voltage swings, provided the gauge is well grounded and doesn’t have too much resistance in the supply voltage.
Also, you might note that Joe has confirmed that a 50 Ohm resistor in place of the sender should give a gauge reading of N. This is the 3rd confirmation of this I’ve had from Jag Lovers list members.
Method Used and Basis of the Results You don’t need to read this rest of this unless you want to know more about how I came to the above results, or you want to do the tests on your own components.
Sender Tests I had three senders on hand, Figure 2. The one at the left is the one that came with the car, part number GTR108. The parts book shows an inline resistor in to go with it, but I don’t recall ever having seen one on my car. (This sender is no longer available so I shouldn’t even bother you with it.) The other two are both the current part (DAC2583) listed for my car as well as later models, probably up until the 6.0 L engine. The rightmost one is brand new, and the one in the middle is the one I put in several years ago. I was going to replace it but decided not to for two reasons. For one thing, the test showed no difference significant between the new and old parts.
Second, the new one would not easily screw into the hole, while the old one does. Rather than investigating, I decided to reinstall the old one after the tests. Note that the probe is longer on the new one. That could be to ensure it projects well into the coolant stream.
Figure 2 Senders tested The test setup is shown in Figure 3 and Figure 4. I formed a jig out of #6 ground wire (any hardware store has it) to hold two senders in the pot at once because I wanted to compare them under identical conditions. (That turned out not to be much of an issue because the data is very repeatable.) The jig has to grip tightly around threads of the senders so it can serve as a ground path for resistance measurements.
I used two DMMs, one fitted with a thermocouple temperature probe adapter (www.tequipment.net, TPI A301) and the other to measure resistance. The negative probe of the resistance DMM is clipped to the mount jig, and the positive probe is shifted between the two senders at each measurement point.
I also tried using my digital kitchen thermometer, partially seen in the lower left of the photo.
The readings agreed pretty well, so it would be a less expensive alternative compared to the thermocouple adapter. The advantage of the thermocouple is it can be wound around the mounting jig to reliably hold it in the water near the senders, whereas the kitchen thermometer has to be held in the water for a few minutes to get an accurate reading. That plus switching the resistance measuring DMM between the two senders under test makes it difficult to record data for each sender at exactly the same temperature.
The technique I developed starts with the water cold, straight from the tap, and the burner set to the lowest possible level. A higher flame results in water temperature changing so fast that it’s difficult to accurately read and record the data. As the water warms up the flame has to be turned up a bit, but not too high. It will take the better part of an hour to complete the test.
Since I used water the highest temperature I could get was about 212 F. At some future time I may do a test using cooking oil so I can get some higher data points. However, as you will see below the trend line is well behaved so extrapolation beyond 212 F shouldn’t be a problem.
Figure 3 Test setup
Figure 4 Close-up of sender test.
Correlation of the Sender Data With the setup and procedures described above I recorded many data points, i.e., pairs of water temperature and corresponding sender resistance. In addition, I had some data published by Bob Egerton mentioned previously. I put all of this data into a program called CurveExpert (http://curveexpert.webhop.biz/). After trying various forms from the CurveExpert menu I found a very good fit to a formula of the form Eq. 1 Rs=exp(a+b/T+cln(T)) where Rs is sender resistance and T is temperature in degrees F. The values of the coefficients found by CurveExpert are a = 26.61777 b = -185.966 c =-4.04296 I realize this is more mathematical than many readers will appreciate, the important thing is the fit is very, good. That is, all of my data plus Bob Egerton’s fall very close to the fitted curve, as can be seen in Figure 1.2
Figure 5 Correlation of Sender Test Results So, it appears that these senders are quite predictable across time and geographic distance. I also believe these senders are reliable since even the old GTR8 part (which was in service for 25 years) was still giving a signal that, while a little different from the DAC2583, was not wildly out of the ballpark. I am moved to say that unless your sender has suffered a fire incident or some other catastrophe, you may well assume it’s OK and performs as indicated by the formula.
Nonetheless, if you want to be absolutely sure yours is OK, just make three careful measurements: It should read close to 240-250 ohms at 125 F, about 70 ohms at 200 F, and 59 or so at 212 F. If it passes that test, it’s alright. Change your focus to the gauge.