ECT Modification Help

[Dominic]

This is a modification I designed approx 1-2 weeks ago trying to figure out how to make a 160° thermostat work in my 2002. (For those who don't know a 160° Stat causes problems in 2002 models :mad [img]smile.gif[/img] [Thanks EPA!].

After doing some simple math, and with the help of strange_trp sending me a sheet showing the resistor values as they correlate with temperature values, I discovered that a 1500ohm resistor in parallel with the ECT Thermistor would/should lower the resistance to make 160degree water temps look like 180degree water temps.

My hypothosis was that by making 160degrees look like 180 to the PCM, I would be able to run a 160 stat, and the added benefit is the engine fans would come on at 195degF actual water temp (The PCM would be reading approx 215-220).

Well this weekend I attempted such an install. Current flowed through the resistor, and I even got a temperature on the dash - but it didnt work. The fans came on as usual when the temps became BLISTERINGLY hot (215degrees or so) at idle, and the gauge on the dash wouldn't budge over 210!

In anger and disgust I threw a 1K Ohm resistor in parallel instead, thus making a gross change in the overall resistance of the circuit - and to my disbelief everything stayed exactly the same as though I had done nothing at all!

:(

I do not understand what I did wrong - I tried putting the 1500Ohm and 1000Ohm resistances in parallel many different ways, and nothing changed. Removing it completely from the circuit yielded the same results. I do know I used the right sensor however, since unplugging it made the gauge drop to 0 as it should.

Furthermore I took a spare ECT that was at room temperature and measured its resistance, and it was right on target with the chart strange_trp sent to me. The hot ECT on my engine was also right in line resistance wise with the chart...

Has anyone else tried such a thing or does anyone have any suggestions? Without autotap I really don't know what the PCM was reading, all I know is the gauge never budged and the fans didn't do anything different - so my only conclusion is that changing the resistance of the circuit somehow, I dont know why though, DIDN'T affect the PCM's thought process concerning fan temps.

[John_D.]

This may be oversimplistic, but hey, that's how my mind works...

If the resistor you're adding in parallel, has a higher resistance than the existing wire, then wouldn't the current just follow the path of least resistance, and take the wire and skip the resistor?

Seems the resistor would have to be put in series to get the result you're looking for. Maybe...? [img]smile.gif[/img]

Edit- never mind... I see you are trying to lower the resistance, not raise it....

[ July 03, 2002: Message edited by: John_D. ]

[ July 03, 2002: Message edited by: John_D. ]</p>

[BillyCobin]

Would you mind either posting the chart or emailing it to me? It would be nice to have.

[Dominic]

I will post the chart in about 20minutes, let me dig it up.

I will also post my xls spreadsheet including all the math I did and my equations for reference.

[nova]

<blockquote>quote:</font><hr>Originally posted by John_D.:
This may be oversimplistic, but hey, that's how my mind works...

If the resistor you're adding in parallel, has a higher resistance than the existing wire, then wouldn't the current just follow the path of least resistance, and take the wire and skip the resistor?

Seems the resistor would have to be put in series to get the result you're looking for. Maybe...? [img]smile.gif[/img]

Edit- never mind... I see you are trying to lower the resistance, not raise it....

[ July 03, 2002: Message edited by: John_D. ]

[ July 03, 2002: Message edited by: John_D. ]<hr></blockquote>

Thats entirely correct.

[remembering from high school electronics]

The computer that you are trying to affect is reading voltage. The voltage splits into two circuits when you add resistor in parallel.

It goes something like this. If you have a wire then down the line split it into two more equal wires and then recombine into one and apply 10v, the voltage will be 10 in both singles and will be 5v in each of the double.


You ARE affecting the voltage along that parallel circuit that you have created with the resistor. Since it is only half of the original voltage value to begin with though and it then recombines with the unaffected half you are not affecting the signal as much as you thought.

I think the 1500 ohm resistor would have to be spliced in in series in order to have the desired affect.

[ends memory trip]

Ok, since I just pulled all of this from the top of my head it might not be entirely accurate.

[MidLifeSpeed]

Basic resistor math ...

Two resistors in parallel ...

Total resistance = product of two resistor values divided by sum of two resistor values.

If both resistors are connected to the same two nodes, the voltage across the either resistor is the same, but the total current divides between the two in an amount depending on the ratio of the values of the two resistors.

To lower the ECT resistance, add a resistor in parallel ... the voltage that the pcm sees will depend on the current through the two resistors, and if the current is constant (likely in this setup) the voltage will be smaller.

I can't address the problem that you were seeing, but my guess is that there are other factors that the pcm uses to determine fan turn on/off, etc.

[DominusEtDeusNatus]

Adding a resistor in parallel does not lower the voltage at that node.

[MidLifeSpeed]

Whoops ... you're absolutely correct Dom ... if the two nodes are connected directly to a voltage supply.

However, the way most ADC circuits work when measuring resistor type sensors is to measure the voltage across a fixed resistor in series with a variable resistor (sensor, thermistor, etc.) or vice versa.

For example, if one of the nodes of the parallel combination is at a constant supply voltage and the other node is connected to a fixed resistor, with it's other end at ground, then the voltage across the parallel combination will vary depending on the value of the resistor added.

This is actually what I meant above by "...voltage will be smaller ...", but after a long day my brain has to work just to keep me awake ... :rolleyes:

My Chilton manual shows the ECT sensor with only two wires connected, and one side to ground. In this case, there is likely a current supplied to the ECT sensor through another fixed resistor in the pcm (although who really knows?), and the varying voltage across the ECT sensor is measured at the "non-ground" node. A resistor added in parallel across this circuit should result in a smaller voltage across the combination (if the sensor supply is as mentioned).

[chas]

a general form of the parallel resistor problem is: Rfinal = 1/(1/R1 + 1/R2 + 1/R3 + ... + 1/RN). this possibly a little easier to use.

anyway, perhaps there are two sensors? did you get a check engine light when you disconnected the ect? the pcm should be fairly annoyed by this condition.

anyway, you want to lower a resistor by 10% (and you can derive this from the above) put in a parallel resistor that is 9*R1. so if you have 1000, and want it to be 900, put a 9000 in parallel.

i am also curious about the voltage you see across the ect sensor. is it constant or vary with temperature? they usually put temp sensors in bridges to linearize their output (since they are notoriously non linear).

[ July 04, 2002: Message edited by: chas ]</p>

[Dominic]

The resistor in parallel saw constant voltage as it should.

The lower resistance from placing the resistor in parallel with the ECT Thermistor was meant to RAISE the temperature reading to the PCM.

Yes, there could be more than one sensor, and this could be my whole problem. If you know of more sensors than the one by the throttle body on the front of the engine LET ME KNOW!

The computer readout Im trying to affect is reading, or should be reading, resistance - NOT VOLTAGE, unless it is using a thermocouple instead of a thermistor. In which case I will need to go about this modification in an entirely different fashion.

For your viewing pleasure here are the images necessary to understand what I was attempting to do:


Graph showing what a 1500-ohm resistor will do to the temperature readout when placed in parallel with the ECT-thermistor

The thermistor ohm values as they correlate with temperature readings by the PCM (thanks strange_trp)

[chas]

after looking at the graph, i believe your problem is the non linear ect sensor. its very non linear.

the parallel resistor only going to fix this curve at a single point. here is the result of grafting on a resistor (1500):

t 212 Rect 177 Rnew 158
t 194 Rect 241 Rnew 207
t 176 Rect 332 Rnew 271
t 158 Rect 467 Rnew 356
t 140 Rect 667 Rnew 461
t 122 Rect 973 Rnew 590
t 113 Rect 1188 Rnew 662
t 104 Rect 1459 Rnew 739
t 95 Rect 1802 Rnew 818
t 86 Rect 2238 Rnew 898
t 77 Rect 2796 Rnew 976

when your engine is at 158 (pretty close to 160) yes, Rnew says something like 170. but if your ECT goes to 194, Rnew is showing something like 200 (oh no, turn on the fans!).

i will have to give this matter a little though. i dont know how they are trying to measure the ect. usually you want to try to limit the current to prevent self heating, and that would mean a variable voltage across the ect.


what is the voltage you see across the ect (without the extra resistor)? is the voltage the same when the ect is cold?

when you disconnect the ect do you see a voltage present on the ect input on the pcm?

[Dominic]

Well duh, all Thermistors have a non linear response - glad you figured that out :D .

I wasn't trying to make it linear, and the graph shows that it still makes a valuable, or should make a valuable, difference once its above the "warm up" temperature. Below that temp things don't matter, the engine isn't in warmed up mode.

The current is extremely low, a 1/2W capable resistor is overkill for the job. The voltage remains the same AFAIK throughout the temp scale, and adding the resistor does not change the voltage - it SHOULDNT - Its in PARALLEL!

My question is still: "Why does doing this make no difference at all in the reading or fan temp kickon?"

I am going to retry the resistor again and see if it cures my 160 stat problems at a later date, currently I dont feel like fooling with it [img]smile.gif[/img] .

[chas]

i was hoping that the thermistor would be a little more linear. as it stands, a simple paralle resistor correction is not going to be a good idea. it is exagerating highs and masking the low temp readings.

why should the voltage vary? you usually just dont stick 12V at one end of sensor. for most sensor circuit (espc thermistors) you use a whetstone bridge to help correct for linearity. if the ect is in a whet stone bridge, the voltage across the ect is going to vary with its temp. if this didnt happen then you are going to have variable current flows. high current flow (i.e. when the ect is exposed to high temps) is going to heat the thermistor, which will skew readings. the ect is almost certainly in some sort of divider or bridge configuration. no, the voltage across the ect wont change when add parallel resistane. does the voltage vary with temperature? it should.

a traditional temp gauge in an older car does work by measuring current though. the ect is wired in series with an ammeter (usually just a bi-metal strip). when the ect drops in value, the current goes up and the bi-metal strip gets hotter and the needle moves. i cant imagine they preserved this scheme for modern digital clusters.

[Dominic]

I have yet to run across a thermistor that is anywhere close to linear, they always use temperature approximations and are only near valid through a given range of temperatures (usually somewhere mid scale).

The voltage might vary a little, but the reason a current is passed through the thermistor is to supply a reading through the I/O on the PCM of resistance, not current flow and not voltage. The resistance will change some if the thermistor is internally heated, but so little current flows through this circuit this is neglegable.

This is not a traditional temp gauge, hence why it uses a thermistor. I wish it used a thermocouple, if it did it would be linear and would be accurate. GM decided to save $0.20 on each car though and use a thermistor.

Masking the lows is not an issue, anything below 170degrees in a 2002 is considered "too cold" and the PCM does nothing more than try to warm up the vehicle. Over 170 the temp variation is perfect, maybe not exactly 20degrees but its close enough. The only important temperatures are when the engine is considered "warm" and when the fans need to turn on - I'm only changing those two.

I will repeat one last time though, doing this modification does not change when the fans come on - as I expected it to. I'm sorry but I'm tired of debating about thermistors and their readings being linear or not, they never are linear - thermistors suck for precision temperature readings. I displayed the chart demonstrating the difference in resistance that would occur and the read temperature that would follow, and I also displayed the resistances as they correlate with the temperature read. If you need to know the approximation formulas I have an excel spreadsheet showing my derived formula where I used an exponential approximation to the curve I can send you.

What exactly are you getting at btw?

[chas]

i am getting at that the thermistor at some point needs to be read. its seems very unlikely that you are seeing a constant voltage across the thermistor. how do you measure resistance? you cant measure it directly. if you put a constant voltage on a varying resistance you are going to get a variable current? most a/d devices look at variable voltages, not current. so it seems more likely that the ect is in series with some high value resistor (to limit total current flow) and they are watching the voltage across that resistor
. again, what is the voltage across the ect? how much does it vary?

you want to know 'why doesn't this work?' in order to answer this, it would help to try to determine how the pcm measures the ect.