The VDO 323-050 is a 1/8-27 NPT 400F oil temperature sender. The resistance is higher at colder temperatures, lower at hotter temperatures. At 400F, the resistance is approximately 12 ohms.
The problem I'm having is two fold: I'm not sure how to properly interface it, and after two runs in the oven, the data isn't consistent. I also have not had any luck finding any useful information on the sender on the web. I had made a web-form request for to VDO for the profile of this part, but never got a response.
The 323-050 is also known as a 323-801-013-001D. I never found the actual chart, but I did find the 323-801-018-001D in the VDO Catalog And Technical Information Guide (document page 26). The characteristics are identical, but the 013 has a 8.2mm x 22mm stem to extend into the oil, whereas the 018 has no stem. The VDO Datasheet has an entry for the 323-801-018-001D (document page 4), and shows the senders characteristics in table 12.
I drilled and tapped a small aluminum box and mounted the sender. A hole was drilled in the cover to fill it with oil, and to insert a thermocouple. The paint was scraped from the area under the cover to insure a good connection to the body of the sender. Checking with an ohm meter showed about 2 ohms between the case screw and the body of the sender. The box was filled to the 3/4 mark with Castrol 10W30 oil for thermal mass. The sender was fully submerged.
For the first pass, I heated the oven to 160F. Once the oil temperature (as read by the thermocouple) passed 150F, I shut down the oven. The oil temperature then coasted up to 170F or so. I left the oven door closed the whole time for the next part. As the oil cooled, each time it passed a 10 degree mark, I recorded the temperature and the resistance.
I repeated the test a second time, heating the oven to about 420F, and recording readings during the cooling period.
For some reason, the data between the 1st and 2nd pass do not correlate well at all, even though the test was performed identically.
The processor I want to interface this to, an NXP LPC2148, has 8 10-bit A/Ds. The Vref is 3.3V, and the A/D input is 0-3.3V. The nominal input impedance is 40K ohms, according to the data sheet. The whole system is 3.3V, but if necessary I can put an additional regulator out there if 5V or some other value < 12V is useful. I would like to avoid any negative voltages. Using 1% parts or a high-end op-amp is not a problem. There isn't much cost sensitivity in this project.
|Temp (C)||Temp (F)||Pass 1 (ohms)||Pass 2 (ohms)|
|Temp (C)||R (Ohms)||Tol (%)||Tol (Ohms)|
To give you the most detail at the high temperatures, I have proposed to have the sensor as the bottom element of a divider network. On the schematic R-TEMP is the sensor and R1 is the other element in this divider network. C2 is a filter cap, I have it at 0.1uf it could be larger. The OpAmp is to buffer the sensor and match the high temperature range to A/D converter. I have the gain at 4 in this example (R2 – R3) , and from the spread sheet this give ~4 to ~8 mv per degree in the range of 200 to 250 degrees. The final R4/C1 at the output of the OpAmp is the input filter for the A/D. We need to look at the data sheet for the processor for there input range, filtering requirements,…
Info from the spread sheet:
On the spread sheet under R—R1 column is the raw output values from the divider network. The next two columns give the output value at a gain of 2 and 4. The last column gives the mv/degree at the 4x gain. The other interesting value is the max current at the bottom of column “D”. If you change the value of R1, you will see current change and the mv/degree. As R1 is make smaller there is more detail at the high temperatures and the current rises.
The LPC2148 is a 10 bits A/D with external reference. You will need to supply the Vref pin with a reference voltage.
I have enclosed a new drawing with the reference - the inductor is a bead, I will find you a part number.
So on to the A/D 10 = 1024 values 3.3V/1024 => 3.2mV/bit. With 3.3V gain = 4 you will have ~ 115 degrees to 250+ degrees with 1 degree accuracy (more or less)The LPC2148 is a 10 bits A/D with external reference. You will need to supply the Vref pin with a reference voltage. I have enclosed a new drawing with the reference - the inductor is a bead, I will find you a part number.
So on to the A/D 10 = 1024 values 3.3V/1024 => 3.2mV/bit. With 3.3V gain = 4 you will have ~ 115 degrees to 250+ degrees with 1 degree accuracy (more or less).
I would have software digital filter in the system, this will increase you accuracy and decrease the noise.