Gas gauge calibration unit

-
Thank you all for the resistance test of the gauges.

Del, I share your thoughts the thermal gauge resistance will change, but 20 ohms is a good starting point. It equates to about 167mA maximim current. The 3V for a bit under full scale helps too. We are in the ball park, 5V * 20/30 is about 3.3V. The duty cycle for that will be a PWM value near 170 out of 255. The calibration will adjust it as necessary, the resolution will be good, much better than our ability to discern needle position .

The LM317 is adjustable, with a resistor value so we can set to 5V, or whatever works. The system will be very compliant to various senders resistances, and gauges requirements.

I could step up to a slightly larger micro controller with more pins, and accommodate temperature and oil pressure calibrations too. However without changing the scales to provide ticks and values, not much would be gained. The gas gauge in my modern car has ticks that serve as good estimates, a tick gives me about 40 miles.
 
If I get time I'll do a little more bench "riggin" with some gauges. I believe I have 4 gauges.
 
Pawned said:
At this point I am not going to disassemble the cluster to see what voltage that is.
Click to expand...

Of course each car is likely to "want" different, and that's the whole point. All I was saying is we probably don't really know "what they started with."

Even though the original pulsing IVR was crude, it HAD to have had a design target, IE duty cycle equating to some RMS

By the way, guys, Ford, and I don't know who else also used these pulsing "regulators". I can still remember, "one fine day" starting up my '60 Ferd Falcoon and having the gauges "both" peg. The mighty, 144 CID 85 hp six. With the integrated head/ intake. A glass bowl on the "huge" single barrel Holley. And a GENERATOR, gasp.

(And a heater that constantly plugged, making the trip into and back from Spokane one Jan day an ordeal in survival to take my Navy entrance physical. Jan?? 1967!!!!)
 
KitCarlson said:
Pawned,

While you price was very good, it came from China, and just the chip for the switch mode regulator costs more here. It makes me wonder if parts that do not meet spec, or other issues, end up being sold, in a gray market.

The solution I propose should work for many, it is not a single point calibration. It costs a few dollars more, but much less than some IVR being sold now.
Click to expand...


I know I am sounding argumentative, but my question is where do you think your switch mode regulator or for that matter any electronic parts you plan to use are made. Grand Rapids, Minnesota???

There is no gray market involved. I ordered two boards, I screwed up the first board because my arthritic hands do not solder like they use to.
The board is so nice I ordered 5 more. In case someone local needs one. I will probably give them a board. For $1.15 I can be a spender
 
Jeep used the same design (fuel gauge ivr) I have been tempted to buy a jeep gauge because they "look" close, if not spot on, in our design in form, fit and function.
The Jeep gauge is also repoped and pretty cheap in comparison and also retain the gauge internal IVR. It has been a couple years but also think the internals are updated with a lot more adjustment.

I have found when the gauge is outside of that +/- 20 ohm it becomes non functional. Another factor to consider about gauge calibration is the extent of the bi-metal deterioration. The supplied voltage and insulated resistance wire heats up the bi-metal and that is what actually makes the gauge needle move.



67Dart273 said:
Even though the original pulsing IVR was crude, it HAD to have had a design target, IE duty cycle equating to some RMS

By the way, guys, Ford, and I don't know who else also used these pulsing "regulators".
Click to expand...
 
Rocky_JS said:
I have found when the gauge is outside of that +/- 20 ohm it becomes non functional. Another factor to consider about gauge calibration is the extent of the bi-metal deterioration. The supplied voltage and insulated resistance wire heats up the bi-metal and that is what actually makes the gauge needle move.
Click to expand...

I know what you mean about that. I had a 57 Studebaker pickup. When I first got it, the fuel gauge was lazy. It was thermal too, but had adjustments. I made some tweaks and had it working correctly. I then drove it for another 33 years and the gauge got a tired with age, just like me.

This fix will not fix a gauge that fails to register empty and full with an applied voltage ranging from 0 to 5V. A bad gauge would read above empty with zero voltage, or not able to swing to full with 5V, or fails to be repeatable. With some crude estimates, and Del's tests, I think the OEM gauge reads "Empty" near 1V and "Full" near 3V.

I am in the process of working out the math for the calibration calculations using a spread sheet. I will try post later today.
 
Here are some calculations I made in a calc sheet. I do things like this prior to writing code.

The current source measurement of sender is linear compared to the gauge resistance in conjunction with sender. Not sure the linearity of the sender and gauge. I have always noticed that fuel gauges seem to be slow to move off full, and fast near empty. If using the current source is not an improvement, the use of a 20 Ohm resistor to feed sensor, and a few changes in measurement will fix that.
 

Attachments

  • calcs (653x608).jpg
    102.8 KB · Views: 477
Kit, you lost me. Or more likely "baffled me with brilliance"
I bet you know the second part of that couplet!

BC
 
Thank you for kind words. I am not so smart. I am horrible with subtraction, addition, multiplication and division. That is why I use a spreadsheet for calculations. The formulas only involves those. I have attached the cell formulas.
 

Attachments

  • calcFormulas (815x467).jpg
    94.5 KB · Views: 461
Thanks, Kit. At some point in my career as a broadcast radio engineer someone told me that I was educated beyond my understanding. I think that applies here as well!
All I really want is a fuel gauge that shows "empty" just a little before the tank truly is empty.

I can handle addition, subtraction, multiplication and division but spreadsheets and formulas baffle me. With a multimeter and a schematic and I'm pretty darn good (if I say so myself) and that satisfies me.
Getting my hands dirty and tracking down and correcting the electrical and/or mechanical failures is much more satisfying than wrestling with theory.

BC
 
The spread sheet lets you enter formulas. If you place an = in a box, you can start a formula. Then you can click on another cell, then operation * , then another cell. The cell with = is then the other two cells multiplied. In the formulas, A3 is column A, row 3. The $ locks the position, this is needed when formulas are copied by dragging, and the formula needs to reference a cell.
 
I remembered I had a tank sender from an 85 Omni. I did some tests with Ohm meter and was disappointed. The readings were random, from a 10 Ohms to meg Ohms. I could see the wire wound resistor, the slider, and hear the slider operate. I lifted the float several times, still gave flakey reading. I found my can of electrical contact cleaner, gave the sender a spray on the resistor and slider, it immediately started working. The resistance is about 7 Ohms full, 100 Ohms empty.

I located most parts to build a prototype. I plan to go to town tomorrow to pick up a few switches and LM317.
 
67Dart273, Is that fluke meter a true RMS meter? I really should not matter with DC. but if you are using a smoothed out PWM signal, it could lie to you.

Normally is should not matter when using on a car. Just something to check
 
Pawned said:
67Dart273, Is that fluke meter a true RMS meter? I really should not matter with DC. but if you are using a smoothed out PWM signal, it could lie to you.

Normally is should not matter when using with an auto. Just something to check
Click to expand...

That's a simple DC supply. It's a Fluke 79 series, "true RMS"
 
67Dart273 said:
That's a simple DC supply. It's a Fluke 79 series, "true RMS"
Click to expand...

Good, it is so difficult at times to explain to people why the readings they get are incorrect even thou the meter says otherwise.

Any frequency (not DC) above or below 60hz can be distorted by non-rms meters.
 
I was able to get to town and buy a couple parts at RadioShack. The part bins were not well stocked, but got things I can substitute in. I made a few minor changes to design.

I spent a few minutes working on a board layout, just to get an idea how to build the prototype. I have not routed, I will do that with wires on prototype. The connections you see are called nets, or ratsnest.

I think the switch buttons and LED1 may be cabled for remote mount, or removal after calibration.
 

Attachments

  • cal_layout (584x499).jpg
    94 KB · Views: 449
clear as a bell. How much is this going to set a person back, including the documentation and your royalty, of course

Mine is still at $1.15 and totally uncomplicated
 
It was a rainy day here, so I found time to build the prototype. It varies some in build due to using parts at hand. I used a few surface mount parts for resistors and capacitors. Component values are not very critical in design. The most critical component is the resistor that sets the sender current. The important part is that the sender output voltage stays within the ADC span of 2.56V provided by the micro controller, there is plenty of flexibility there. Errors are calibrated out relying on the 10 bit ADC (about 0.1% resolution), and 8-bit PWM ( 0.5%).

I have started software. I have it figured out. It a matter of writing and testing the code. I will do it in a few steps. First I will get the ADC and PWM going. Next write the function that translates sender values to PWM. Followed by the calibration routine, that interprets button selections, and stores calibration values in eeprom.
 

Attachments

  • 20140202_154940~2.jpg
    79.8 KB · Views: 400
  • 20140202_155009~2.jpg
    93.7 KB · Views: 390
What are we gonna do when RS finally quits selling components, or at least enough to "keep us busy" LOL
 
I only buy at RS when I have to. In this case a few parts saved the minimum order and postage of ordering. The RS part choice is not too good. I typically order from Digi-Key or Mouser. The service and quality and price is great when doing larger orders.

When choosing parts for a design I often located part sources using findchips.com. Just enter part number, it comes up with the vendors that have it and the prices. I only used parts if they are readily available by a few vendors.

I think all parts will be between $5 and $10 for this. A bit early to tell, it is more important to prove it works well at this point.
 
The hardware has checked out. I added a 6 pin programming connector to load code. I was thinking of doing that off board, but the connector will save time of swapping in chips.

I have the read sensor and convert to PWM going, to test reading a sender. I am using a precision pot and looking at the PWM with scope. The PWM frequency is about 250Hz, the duty cycle varies with resistance change.

I made a short video: [ame="https://www.youtube.com/watch?v=-bPTAOcXXLg"]ADC2PWM - YouTube[/ame]
 
Now THAT was a short video, LOL. I don't have anything fancy like a precision pot in a box, LOL
 
You must log in or register to reply here.
-

Back
Top