Darwood said:Ok I've now spent the past couple days at work refreshing up on Thevenin equivalents and I think I have the math working right. Yay I have my freshman understanding of electronics back.![]()
Congratulations , I never got Thevenin equivalents to work right and gave up after a day or so, I completely abandoned trying to figure out the needle position besides center. And used the test bed, trying of pin down the center point was hard enough and that was for the much easier to follow math of a balanced whetstone bridge,
Thanks again RavenTai for your research and numbers. Also thanks for the gif showing the stock gauge.
I have a about 900 MB of pictures and charts that you may find useful also, maybe smaller if I compress them, If you want I can send you copies, not sure how would be the best way to get them to you though. Too large for E-mail.
I did some math on the 110 degree resistor with a wire replacing the zener diode (assuming that L1 = 95.3 ohms, L2 = 81.5 ohms, and L3 = 46.0 ohms) and I got .9140 volts across L1 at 226 degree F. This is using a resistor value of 38.4 ohms for R2. This should put the needle at above the red zone.
Ok with the 110 R1 and shunt replacing diode I got pegged hot with 226, in-fact about 224 or 225 or so the needle stopped pegged. According to the stock needle sweep (the sweep we have L1 voltages for) .9140 should have been near the top of red not pegged full hot.
I've read the actual results you obtained and it seems like this is about 10 degrees off (not unlike the value for the 110 ohm resistor). In reality .9140 volts across L1 with R1 = 110 ohms and a short for the diode is about 217 degree F.
That is true, every attempt I have made to go from paper to the gauge have been off, I would have to go back through this thread any my notes to see if there was a consistent amount they are off by, also the stock needle sweep was early, I cannot remember it that was at a good source voltage or not making the needle position per voltage across L1 suspect.
A coworker of mine mentioned that since the needle is pegged cold with a spring of some sort that this could be why the temps are lower than we expect given the resistances. The spring is biasing the sweep a bit. This may or may not be true. I think to figure out if it is we would have to divulge into the realm of differential equations (which I'm currently not up to).
I am not either, there is something in the needle movement that brings the gauge to the cold mark, It does so slowly with to me means it is fairly weak and that the needle is dampened somehow. It could be a spring or also could be a small magnet placed in the air core motor case.
I'm going to be playing with resistor values to replace the diode today to see what I get and to see how closely the math mimics reality.
RavenTai as a request, if you do any more tests can you always measure the voltage across L1? That would be awsome. To me that value is the best way to see if the math on paper matches reality. Source voltage is nice to know but seeing that it doesn't change much it isn't that valuable.
![]()
I was monitoring source voltage early on because I was working with a battery and battery charger. Had to constantly turn the batty charger on and off. To keep close to the desired voltage. When I built the DC power supply that took care of that but I was still afraid to monitor the L1 Voltage as the currents in the L1 coil are very small and I was afraid that the current the meter draws to measure the voltage (albeit even smaller) would skew the results. Also the DC power supply wanders a bit (0.1 volt or so) especially as it warmed up, I monitored and adjusted the pot as necessary to keep it close to the goal of 14.50 volts.
Darwood said:The more I do the math the more I would like more readings in regards to the voltage over L1 and the temperature. Right now I'm not trusting my calculations to calibrate the thermistor graph due to not being 100 percent sure on how to handle the diode in the circuit when it comes to Thevenin equivalents. As noticed in real life the temperatures and the resistances measured from the thermistor aren't exact when it comes to the circuit. The 110 ohm resistor should have created a center of 200 degrees but it seems like it was off 9 to 10 degrees. It looks like the curve is a bit off when the circuit is involved.
I would agree that the thermistor curve is off.
I did some calculations on one of the gif images and for a voltage of .200 volts across L1 I calculated that the resistor R2 should be 31.9 ohms (or 30.8 ohms depending on how I calculate it). This corresponds to 228 degrees Fahrenheit based on the gif.
If you look at the chart 31 ohms should be between 239 and 240 degrees.
I also have some calculations on the current circuit with the 110 ohm R1 resistor and various diode replacements.
These are the numbers I get for the R2 resistance when the voltage across L1 is .740 volts which should put the needle at the bottom of the red zone (i.e. the bottom of the needle is touching the bottom edge of the red zone):
D1: 100 ohms R2: 32.9 ohms
D1: 75 ohms R2: 34.5 ohms
D1: 50 ohms R2: 36.9 ohms
D1: 90 ohms R2: 33.0 ohms
Ok taking the 50 ohm D1, R2= 36.9 and crossing to the thermistor curve that should be approximately 229F but in reality bottom of red is 213 (using top of needle sighting) so there is defiantly something amiss
I also thought some about a replacement zener diode but the smallest break over I could find in the Mouser catalogue was 2.4 volts which is larger than the 2 volt break over of the stock component. I'm guessing a 1 volt break over would shrink the dead zone to only 20 degrees but I haven't calculated that nor is there much point if we can't find a diode to meet that requirement. Also with a diode the resolution would be harder to modify.
Correct me if I am wrong ( I very well may be) but I think a 2.4 ZD would actually give 3.1 volts total dead time, as there is .7 v (or .5) on the foreword bias side that is added to the reverse bias break over when working with both foreword and reverse bias as we are here.
I do not think any dead spot is desirable no matter how small, it leads to surprises.
I would love for someone to work on the theoretical model side of this, I have given it about all the metal energy I could muster at the time. It needs fresh eyes.
So do you think the thermistor curve is the major source of flaws in the model? It very well may be and is actually probable that the thermistor reacts differently when under load, the thermistor curves I have are direct unloaded meter readings.
so
Is it the load that is causing the change?
What is the best way to get very accurate loaded thermistor curve?
First thought would be put a load on it, might as well be the same load it experiences, 110 ohm resistor in series with R2, apply 14.5 volts across both. Wile you cannot directly check the resistance of R2 wile it is electrified you can check the voltage between the two resistors, since source and R1 are known you should be able to calculate for the resistance of R2 based on the voltage between them.
But what of the current To/From L1? Is it better to measure R2 wile experiencing this current (ie measure with in the entire circuit) or is that double dipping when you drop these numbers into the Thevenin equivalents? I have an very early cook that did something similar to this