Fuel Gauge Sending Units compared

[rich006]

I posted this on slantsix.org but I thought I should share it here too. There have been several threads here about the aftermarket sending units not working quite right with the factory gauge. In particular, it reads full when full, empty when empty, but at mid-ranges the aftermarket unit reads much lower than it should. In other words, when you have 3/4 tank, your gauge reads 1/4. These aftermarket units are made by Spectra Premium. I called them to complain about it, and surprisingly a person answered the phone. He told me it's designed to put out 80 ohms empty and 10 ohms full, same as the factory sending unit. I explained it's the mid-range where it's off, and he said he'd report that to the tech team.

I compared the response of the OEM sending unit to that of the Spectra Premium. First I measured the resistance of each at several float angles. Then I did a little geometry and math to convert float angle to fuel level. Here are the results. As you can see, the Spectra unit matches the factory unit reasonably well at the end points (as the Spectra rep told me), but in the mid-range its resistance is much too high, leading to a too-low gauge reading. For more details on my measurement process, read below the picture.

Geometry: I measured the float angle relative to horizontal. I reasoned that the float height (fuel level) is proportional to the sine of the float angle, plus a constant.

Electrical measurements: On the OEM unit I couldn't get steady resistance measurements, so I hooked it up to a 6V flashlight battery and measured the current. From there I used R=V/i to get the resistance. Those current measurements were steady, but my multimeter only reads to the nearest 10 mA of current, so the higher-resistance measurements are a bit uncertain. On the Spectra unit I read the ohms directly with the multimeter.

 

[rich006]

Here's why the Spectra unit doesn't work right. In the photos below, I've opened up the box containing the sliding contact and the resistive wire. The variable resistance comes from the total length of wire between the sliding contact (that moves with the float) and the case, which is electrical ground. Each time the contact touches a new turn of the wire, an additional length of wire is added to the circuit and so the resistance increases. You can see that all turns of the resistive wire are of equal length, so as the sliding contact moves along, the resistance between the contact and ground increases by the same amount each time a new turn is contacted. Also, the turns are equally spaced, so every inch of contact movement adds the same number of turns to the resistor.

 

[rich006]

And here's what the resistor from the OEM unit looks like. Sorry I don't have a pic of the whole unit for context. You can see as the contact moves across the wires, each length of wire added to the circuit is longer than the last length added.

 

[abodyjoe]

i've seen something similar done in the past.

personally i think too much is made of it. i couldn't care less if its "accurate" throughout the sweep as long as it shows empty before its actually empty.

 

[rich006]

This OEM sending unit came from a 1969 Dart in a salvage yard. The resistor's phenolic board had warped, so that the contact would dig under the nichrome wire. In the picture you can see where that happened. The warping also caused the resistor not to make good contact with ground at the narrow end. To stop the contact from digging under the wires, I had my son solder a brass nail head into the worn hole at the end of the contact, to make a nice rounded contact surface. To make a good connection at the ground end of the resistor, we soldered that contact, then epoxied both ends of the resistor to hold it in place. I don't know how long it will last but it's working well now.

 

[rich006]

personally i think too much is made of it. i couldn't care less if its "accurate" throughout the sweep as long as it shows empty before its actually empty.

I hear you. I drove 11k miles with the Spectra unit and I got used to looking carefully at which part of the E the needle was on. But I like it when things work the way they should, and I enjoyed tinkering with it. This post is really for the scientists and engineers out there, or anyone who is curious why the new ones do what they do.

 

[Demonic]

Great work! Email those pics to Spectra, along with your observations. A slim chance of improvement is better than none.

 

[Mike69cuda]

I used a meter match to solve this problem on mine. Works nicely.

 

[rich006]

Great work! Email those pics to Spectra, along with your observations. A slim chance of improvement is better than none.

I did. We'll see if anything comes of it. At least I had fun with the project.

 

[RedFish]

What the aftermarkets people dont understand is how these thermal instruments work. Inside the early instrument 13 inches of wire that measures approx' 1.5 ohms per inch at 68 degrees. Total resistance is approx' 20 ohms. Later instrument had 13 inches of approx' 1 ohm per inch wire for total resistance of approx' 13 ohms. Both function exactly the same. The later runs a little cooler thus lives longer.
Heat causes the bimetal to bow moving the gauge pointer. The big issue.. as that wire heats up its per inch resistance changes. Plus there is some geometry involved in the arc/bowing of the bimetal.
So from 80 ohms to 73 ohms, 7 ohms difference will lift the pointer from home to empty. Thats equal to about 1/8 of range spacing. 73 ohms to 34 ohms, about 40 ohms difference, will move the pointer to 1/4 range. 34 to 23, 11 ohms difference, moves the pointer from 1/4 to 1/2 range. 23 ohms to 14 ohms, 9 ohms difference moves the pointer to 3/4 range. 14 ohms to 10 ohms, 4 ohms difference moves the pointer from 3/4 range to full. In a nut shell , As the wire heat up, fewer ohms resistance required to generate the same amount of pointer movement. All that is why a linear sender simply will not work with a thermal instrument. Half way between 80 and 10 is what? The middle of a linear senders ohm signal. The graph plotting of OEM sender is pretty close to accurate. The plotting of aftermarket sender isn't exactly accurate but close enough to prove the fault. Maybe send this info to Specta/whoever.

 

[Keystone]

Total resistance is approx' 20 ohms. Later instrument had 13 inches of approx' 1 ohm per inch wire for total resistance of approx' 13 ohms. Both function exactly the same. The later runs a little cooler thus lives longer.

Can you explain how that works? It sounds backwards to me. If the voltage across the circuit is the same, than the lower resistance would have a greater current flow. Wouldn't more current equal more heat?

 

[67Dart273]

I used a meter match to solve this problem on mine. Works nicely.

You should have made more of this post. Guys the MeterMatch seems to be a pretty good and not expensive help for the problem

TechnoVersions - MeterMatch for Analog Gauge Correction

"MeterMatch is wired between your sender (sensor) and your gauge. It reads the resistance of the sender, then based on your user-defined internal calibration values, it electronically emulates the sender signal to drive the gauge to the desired reading. It treats the input and output entirely separately, so it doesn't care if the input to the gauge is scaled differently than the sender, or even if the sender signal has a reversed sender signal than the gauge - it can correct for that."

 

[Dicer]

The windings and shape of the restate is proportion to the shape and size of the fuel tank, and in our case (A Body) the size and the cut out use for the spare tire well. The Spectra sender would work perfect if we had a prefect square fuel tank.

 

[67Dart273]

The windings and shape of the restate is proportion to the shape and size of the fuel tank, and in our case (A Body) the size and the cut out use for the spare tire well. The Spectra sender would work perfect if we had a prefect square fuel tank.

I think the real question is "why can't the repoppers" make the damn thing like the OEM? But that would require actual work.

 

[Dicer]

But that would require actual work.

You answered your own question....

 

[Dana67Dart]

We got very similar results. I used the tank and 18 gallons of fuel one gallon at a time from empty to full amd measured resistance. My OEM sender was a NOS from Dodge but it was different in shape than the OEM from the 60s This was in a 67 Dodge dart.

 

[Mike69cuda]

You should have made more of this post. Guys the MeterMatch seems to be a pretty good and not expensive help for the problem

TechnoVersions - MeterMatch for Analog Gauge Correction

View attachment 1715462667

View attachment 1715462668

"MeterMatch is wired between your sender (sensor) and your gauge. It reads the resistance of the sender, then based on your user-defined internal calibration values, it electronically emulates the sender signal to drive the gauge to the desired reading. It treats the input and output entirely separately, so it doesn't care if the input to the gauge is scaled differently than the sender, or even if the sender signal has a reversed sender signal than the gauge - it can correct for that."

Yeah, guess I was still half asleep.

I made a calibration box for a meter match when I did mine. I used info I found on the forum, probably Dana67dart’s chart, to make it. Using the box you don’t have to fool around with the sender to calibrate the meter match. I chose the closest resistances that I had. it seemed to work pretty good. Refilling the tank at different gauge readings seemed to correlate about +/- a gallon or so.

If anyone wants to borrow it I can send it to you.

 

[MOPAR4Me]

I posted this on slantsix.org but I thought I should share it here too. There have been several threads here about the aftermarket sending units not working quite right with the factory gauge. In particular, it reads full when full, empty when empty, but at mid-ranges the aftermarket unit reads much lower than it should. In other words, when you have 3/4 tank, your gauge reads 1/4. These aftermarket units are made by Spectra Premium. I called them to complain about it, and surprisingly a person answered the phone. He told me it's designed to put out 80 ohms empty and 10 ohms full, same as the factory sending unit. I explained it's the mid-range where it's off, and he said he'd report that to the tech team.

I compared the response of the OEM sending unit to that of the Spectra Premium. First I measured the resistance of each at several float angles. Then I did a little geometry and math to convert float angle to fuel level. Here are the results. As you can see, the Spectra unit matches the factory unit reasonably well at the end points (as the Spectra rep told me), but in the mid-range its resistance is much too high, leading to a too-low gauge reading. For more details on my measurement process, read below the picture.

View attachment 1715462562

Geometry: I measured the float angle relative to horizontal. I reasoned that the float height (fuel level) is proportional to the sine of the float angle, plus a constant.

Electrical measurements: On the OEM unit I couldn't get steady resistance measurements, so I hooked it up to a 6V flashlight battery and measured the current. From there I used R=V/i to get the resistance. Those current measurements were steady, but my multimeter only reads to the nearest 10 mA of current, so the higher-resistance measurements are a bit uncertain. On the Spectra unit I read the ohms directly with the multimeter.

Nice write-up, thanks for sharing it with us here, helped me understand what the issue is with these gauges, and hopefully your contact with them will result in some change for the better.

I think the real question is "why can't the repoppers" make the damn thing like the OEM? But that would require actual work.

I use to ask myself the same thing all of the time, and I have come to conclude following.

Original design likely includes differences in, labor, volume of materials, quality of materials, all intentionally reduced, to maximize business profits, and to give the customers with a few exceptions (like us classic car hobbyist) what they most commonly want, least costly purchase, regardless of reduced quality.

The combination of these two market actors (businesses/common customers) dancing hand in hand, in the ball-room of the good enough disposable product society, results in the ever diminishing quality product marketplace that we live in .

 

[Mattax]

That's only half the problem.
The other half is that the arm is too short - at least for ones that are supposed to replace the senders from the 60s.
Like shown in the photos in this post by @"Dart67"

vs. original

 

[RedFish]

Can you explain how that works? It sounds backwards to me. If the voltage across the circuit is the same, than the lower resistance would have a greater current flow. Wouldn't more current equal more heat?

Yep. A hot gauge doesn't measure 20 ohms or 13 ohms, which ever type is subject. Resistance within a thermal gauge changes as the wire temperature changes.
The 13 ohm design is a lot different mechanically. The bitmetal is different. Even the pointer weight is different. For what its worth... if you have the later 13 ohm fuel gauge in your inst' panel, you will find the oil and/or temp gauge was never changed. Still 20 ohm type. Why? Oil and temp gauges rarely if ever go to the hotter end of function. So never was a early failure issue with those. I will add this, The 13 ohm fuel gauge is not as sensitive or accurate as the 20 ohm fuel gauge.
These OEM thermal instruments are basic range indicators. There are no gallon marks on the screen. Even so, a heck of a lot more accurate than what aftermarket offers.
Yes the tank shape could be considered a factor but, the vehicles stance doesn't hold the tank exactly level. The contact in the sender is touching 2 or more laps of that resistor wire at any given position. So exactly 23 ohms didn't need to be precisely in the center of the OEM senders arm swing. So is it? In the case of the rally fuel gauge for example the 23 ohm calibration target/dot is not exactly aligned with the half tank hash mark. Shape and stance of the tank is compensated right there. That's all I'm going to share here.
I don't know how many times I have briefly considered posting pics of S&W data sheets I have but.... The guy that sold this material to me warned, This stuff at exactly legal to have (stolen). 100 year copywrites have not expired. That's all I needed to hear.

 

[Cuda Al]

You should have made more of this post. Guys the MeterMatch seems to be a pretty good and not expensive help for the problem

TechnoVersions - MeterMatch for Analog Gauge Correction

View attachment 1715462667

View attachment 1715462668

"MeterMatch is wired between your sender (sensor) and your gauge. It reads the resistance of the sender, then based on your user-defined internal calibration values, it electronically emulates the sender signal to drive the gauge to the desired reading. It treats the input and output entirely separately, so it doesn't care if the input to the gauge is scaled differently than the sender, or even if the sender signal has a reversed sender signal than the gauge - it can correct for that."

Nice idea for a sender that actually doesn't bottom out BEFORE the tank is empty, useless otherwise.

Alan

 

[MileHighDart]

I have that spectra unit, and with mine it seems like it can be reading empty on the gauge, so I put like 6 gallons in it and the gauge will read full. So I'm not really sure of anything the gauge says.

 

[rich006]

as that wire heats up its per inch resistance changes. Plus there is some geometry involved in the arc/bowing of the bimetal.

That actually clears up a lot for me. Thanks!

We got very similar results. I used the tank and 18 gallons of fuel

Awesome work, thanks for sharing! With your method you can see what happens when the fuel level is above the float's maximum height. My method doesn't capture that.

 

[Dana67Dart]

One of these days I'm going to finish my resister box and sweep between 7 and 80 ohms and see what the gauge does.

 

[RedFish]

One of these days I'm going to finish my resister box and sweep between 7 and 80 ohms and see what the gauge does.

I bought a decade resistance box. Somewhere between 60 and 80 dollars, I forget. Watching a gauge respond perfectly to 1 ohm increments and factory specs doesn't help the aftermarket sender problem.