Guest ratchethack Posted March 2, 2009 Posted March 2, 2009 Does this mean that you can pack the holder with sugar and have the same effect as copper paste? Just when I thought I was out, they drag me back in. Good one, Dan. Don't let the raucous ruckus from the peanut gallery turn you away, my friend. Your expertise is greatly appreciated here.
GuzziMoto Posted March 2, 2009 Posted March 2, 2009 Sorry but you have it wrong. If you've ever measured one, the readings are steady and repeatable in the tight range and vary substantially when you get into the range out side of calibration. Edit: A sensor is more accurate if 10 degrees is represented by only 50 ohms rather than 500. Remember, the ECU is reading voltage after the sensor. Small ohm changes mean small voltage changes or accuracy. What you are suggesting is the engineers are having these sensors be in their most accurate range at -40C rather than in normal operating operating temperatures. Why? That is not how the thermistors I work with work. You may be confusing the range the thermistor is most accurate in with the range the thermistor is least sensitive in. They are not one and the same. Yes, it may be easier for you to read a value when the reading slows down, but that does not mean the computer is in the same boat. The range a thermistor is most accurate (which is what I assume you meant when you said "sensitive") is determined by the material it is made from and it is usually somewhere in the middle of it's range, not on the far end of it's range. The air temp sensors I work with are usually most accurate in the 77 degree F area, where 5 degrees is worth about 1,000 ohms.
Dan M Posted March 2, 2009 Posted March 2, 2009 That is not how the thermistors I work with work. You may be confusing the range the thermistor is most accurate in with the range the thermistor is least sensitive in. They are not one and the same. Yes, it may be easier for you to read a value when the reading slows down, but that does not mean the computer is in the same boat. The range a thermistor is most accurate (which is what I assume you meant when you said "sensitive") is determined by the material it is made from and it is usually somewhere in the middle of it's range, not on the far end of it's range. The air temp sensors I work with are usually most accurate in the 77 degree F area, where 5 degrees is worth about 1,000 ohms. I agree wholeheartedly. I never said that 90-110 was at the end of the range. I said it was the accurate range. I'm sure the thermistor will continue to read at temps far higher than on the chart but there is simply no reason to publish temp / ohm numbers much above 130C or 140C because an engine is going to be toast by then. The engineers are not going to map fuel trims for an overheated engine. When a coolant temp sensor is shorted the scanned value is typically beyond 300F. Likewise when it is open it shows -40F - I'm guessing that is the approximate range of the spec'd thermistor. If you look at the chart that I linked to in a previous post. While it is rather crude you can see that when the graph makes a nice arc is is easy to get a steady voltage change. When things cool the line straightens out and ohms change in chunks. I'm sure if the graph went beyond operating temp (again, why would they?) the line would level out. However you want to term it, accurate, sensitive, least sensitive, The target range in this case with a typical coolant temp thermistor is 90C to 110C. Edit - I just have to add that if thermistors are can be made to function accurately in various temperature ranges, why would automotive (and motorcycle) engineers almost universally choose sensors calibrated to the same range and many times the exact same sensor if its sweet spot was not in the area of normal operating temperature? I mean really, to suggest that all these guys the world over would spec a sensor that was most accurate in the area where the engine didn't operate is quite absurd.
Guest ratchethack Posted March 2, 2009 Posted March 2, 2009 What type of insulator material could be used to sleeve or wrap around the brass temp sensor holder and withstand the heat involved ? Cheers Bruce Bruce, I'm no materials engineer, but lacking a source for replacement of the OE plastic holder, seems to me any good machine shop that you describe your requirement to should be able to source a readily available composite, or even a phenolic sleeve as an insulator, and fit it to either a cut-off brass holder, or to the base of a broken plastic holder, that would work as well as the original OE plastic holder. Good luck.
dlaing Posted March 2, 2009 Posted March 2, 2009 Look at where the numbers tighten up: -40 100950 -30 53100 -20 29120 -10 16600 0 9750 10 5970 20 3750 25 3000 30 2420 40 1600 50 1080 60 750 70 525 80 380 90 275 100 205 110 155 125 100 In the range of 90C to 110C the variation is only 120 ohms. Pretty easy to pin point a temp. By comparison, 0C to 20C it is 6000ohms. Far greater margin for error. Make sense? I think the ECU is measuring voltage (0-5V), not Ohms. What would that chart look like in Volts? As I understand it it will follow evenly, so for known accuracy we are left with how the ECU will interpret the Voltage. I know my volt meter's resolution increases at lower voltages, which would be at colder temps. The PDF I have specs 5% accuracy between -40C and +125C, which seems accurate enough. While I agree it would be nice to see greater accuracy at 100C, I don't see the chart above as proof of that. The percentage change in Ohms is greater at temperature changes other than 100-110. But percentage change might not imply accuracy. We may be getting 5% accuracy at -40C and 1% accuracy at 100C, but this chart is not evidence of that, and it only seems to be evidence that the ECU will get a more accurate reading at lower temps, unless I am missing something. I'll go back and re-read your posts, but I am not getting it.
Dan M Posted March 2, 2009 Posted March 2, 2009 I think the ECU is measuring voltage (0-5V), not Ohms.What would that chart look like in Volts? As I understand it it will follow evenly, so for known accuracy we are left with how the ECU will interpret the Voltage. I know my volt meter's resolution increases at lower voltages, which would be at colder temps. The PDF I have specs 5% accuracy between -40C and +125C, which seems accurate enough. While I agree it would be nice to see greater accuracy at 100C, I don't see the chart above as proof of that. The percentage change in Ohms is greater at temperature changes other than 100-110. But percentage change might not imply accuracy. We may be getting 5% accuracy at -40C and 1% accuracy at 100C, but this chart is not evidence of that, and it only seems to be evidence that the ECU will get a more accurate reading at lower temps, unless I am missing something. I'll go back and re-read your posts, but I am not getting it. Right Dave, the ECU sends out a 5 volt reference to the sensor. The resistance of the sensor knocks down the voltage and that is what the computer sees. When it is cold the signal does not change steadily. You can hook up a meter to read either the ohms across the sensor or the voltage going back to the ecu (the better way) on the signal wire. What you will find is when the sensor is at low temps the temperature can change a few degrees and the ohms will not change then the ohms will change a great deal and stay there over the span of a few more degrees then make a big move again. (look at the stair steps in the graph linked in the earlier post) As the sensor gets into the range we've been speaking of, the ohms / volts are moving very steady (small even steps) with temp. I've never tested one at extremely high temps like 290F but I'd assume the opposite would be true, the changes (stair steps if you will) would be too small and difficult to measure, and the line in that graph would go almost horizontal.
Guzzi2Go Posted March 2, 2009 Posted March 2, 2009 ...Edit - I just have to add that if thermistors are can be made to function accurately in various temperature ranges, why would automotive (and motorcycle) engineers almost universally choose sensors calibrated to the same range and many times the exact same sensor if its sweet spot was not in the area of normal operating temperature? I mean really, to suggest that all these guys the world over would spec a sensor that was most accurate in the area where the engine didn't operate is quite absurd. We have a situation where same thermistor type is used for measuring both air and motor temperature. Obviously, this will be in a different temp/resistance hence accuracy ranges. If accuracy was of concern for both applications, wouldn't different sensor types be specified? I think that on a liquid cooled engine, the task of air temp sensor would be to fine tune EFI (I think it matters if it is -10°C or +40°C outside), whereas the task of a cooling liquid temp sensor on a warmed up engine is to switch fans on or off. Hence the same sensor can be used for both applications, since the requirements on the accuracy are quite different. Earlier I wrote that emission tests are done at engine temperature between 60-80°C. I interpret this as freedom to "freeze the map" at that point, rather then continue to lean out. So I don't see the point of fine tuning past 80°C point as the engine (at least ours) would benefit from running richer than it is running. That also explains why the same sensor can be used for oil temp measurement. It simply does not matter. The map should be constant at high temperatures. Sure an attempt to "optimize" was made in the early map, but this can be attributed to an error "on the specer's part".
Greg Field Posted March 2, 2009 Posted March 2, 2009 What type of insulator material could be used to sleeve or wrap around the brass temp sensor holder and withstand the heat involved ? Cheers Bruce I use duct tape. It's been on there 18 months and hasn't gone anywhere.
Greg Field Posted March 2, 2009 Posted March 2, 2009 Does this mean that you can pack the holder with sugar and have the same effect as copper paste? Yes, but only at temps of 185 C or higher, the melting point of sugar.
Dan M Posted March 2, 2009 Posted March 2, 2009 Yes, but only at temps of 185 C or higher, the melting point of sugar. At least the sensor would be in liquid as it was designed. Another important discovery!
Dan M Posted March 2, 2009 Posted March 2, 2009 We have a situation where same thermistor type is used for measuring both air and motor temperature. Obviously, this will be in a different temp/resistance hence accuracy ranges. If accuracy was of concern for both applications, wouldn't different sensor types be specified? I think that on a liquid cooled engine, the task of air temp sensor would be to fine tune EFI (I think it matters if it is -10°C or +40°C outside), whereas the task of a cooling liquid temp sensor on a warmed up engine is to switch fans on or off. Hence the same sensor can be used for both applications, since the requirements on the accuracy are quite different. Now you are losing me G2G. The oil temp sensor as guzzi calls it is a coolant temp sensor in any other application - brass sensor probe designed to be submerged in coolant. Part #29729461 The air temp sensor is exactly that. A plastic sensor body with an exposed thermistor probe designed to sense air temp. Part # 30729331 Two completely different sensors for two different purposes. I'm quite sure they are spec'd to be accurate in the ranges needed.
raz Posted March 2, 2009 Posted March 2, 2009 I think the ECU is measuring voltage (0-5V), not Ohms.What would that chart look like in Volts? Here are the voltages1) and, more importantly, what the ECU reads2). The ADC column is what the AD converter should end up with. It's an integer between 0 (0 volts) and 1023 (5 volts). °C ohms mV ADC -40 100950 4927 1009 -30 53100 4863 996 -20 29120 4755 974 -10 16600 4586 939 0 9750 4333 887 10 5970 3996 818 20 3750 3571 731 25 3000 3333 683 30 2420 3087 632 40 1600 2581 529 50 1080 2093 429 60 750 1667 341 70 525 1296 265 80 380 1011 207 90 275 775 159 100 205 601 123 110 155 468 96 125 100 313 64 1, 2) this is based on the assumptions the ECU use a 1K5 fixed bridge resistor and a 10 bit ADC, like MyECU. I do not know for sure if that is the case.
dlaing Posted March 2, 2009 Posted March 2, 2009 Thanks Raz! Just what I wanted to see! and Dan Thanks, that makes sense if it is jumping increments and the number increases with temperature. But then wouldn't the greater accuracy be above 110C?
raz Posted March 2, 2009 Posted March 2, 2009 Now you are losing me G2G. The oil temp sensor as guzzi calls it is a coolant temp sensor in any other application - brass sensor probe designed to be submerged in coolant. Part #29729461 The air temp sensor is exactly that. A plastic sensor body with an exposed thermistor probe designed to sense air temp. Part # 30729331 Two completely different sensors for two different purposes. I'm quite sure they are spec'd to be accurate in the ranges needed. 30729331 is used both for air and oil temp sensors on 1100 Sport, I guess that is what G2G meant. Maybe Guzzi found out it was a bad idea, I don't know. No other model I know of used that for the oil.
Dan M Posted March 3, 2009 Posted March 3, 2009 Here are the voltages1) and, more importantly, what the ECU reads2). The ADC column is what the AD converter should end up with. It's an integer between 0 (0 volts) and 1023 (5 volts). °C ohms mV ADC -40 100950 4927 1009 -30 53100 4863 996 -20 29120 4755 974 -10 16600 4586 939 0 9750 4333 887 10 5970 3996 818 20 3750 3571 731 25 3000 3333 683 30 2420 3087 632 40 1600 2581 529 50 1080 2093 429 60 750 1667 341 70 525 1296 265 80 380 1011 207 90 275 775 159 100 205 601 123 110 155 468 96 125 100 313 64 1, 2) this is based on the assumptions the ECU use a 1K5 fixed bridge resistor and a 10 bit ADC, like MyECU. I do not know for sure if that is the case. I think your assumptions are probably good Raz. Certainly beyond me to make those calculations. I was looking at specs from various auto makers and most provide ohms but not the corresponding voltages. When reading data on a scan tool you generally see temp and voltage. Ford allows an ohm variance of 15% and Toyota shows a crude chart with a fairly wide range of acceptance as well. Still 15% is less than 10 degrees on your chart until you get to the cold (less accurate ) end of it.
Recommended Posts