Looking for input on a fuel gauge/sensor design

Posted: Tue Jul 16, 2019 3:45 pm

After running out of gas twice- once because I forgot I was on reserve and once because I’m too stubborn to fill up when I still have a gallon left- I decided I could use a fuel gauge on my 900. I’m not a big fan of the float type gauges because they require a lot of work to instal and need modification of the tank typically so I decided to make my own sensor and controller unit that can just be tapped into the fuel line with a tee right under the petcock. I plan on mounting a small led indicator on the top of my triple tree to see the fuel level but I know different people might want a different gauge or mount. So my question for everyone is if you could add a fuel gauge to your bike, what kind of gauge would you prefer and where would you want it? I’m thinking about making a number of these and selling them because most options out there are pretty expensive or not that great. I’ve attached a few pictures of examples gauges. I’ve seen them mounted next to the primary gauge, in the tank, on the handlebars and sometimes down by the petcock before. Just not sure what the majority of people would prefer.

TL;DR- I’m making hydrostatic type fuel sensors and looking for input on what kind of gauges/mounts people would prefer on their custom builds.

Posted: Tue Jul 16, 2019 3:50 pm

I personally don't need a fuel gauge since I fill up anywhere between 50-80 miles into the tank; I pass by gas stations all the time, so it's easy. But more important, we have an electronics expert here in the forum that's currently working on a non-invasive fuel sender unit/gauge design. If it works out, it will be a fairly easy installation. So rather than reinventing what's presently being invented, you might want to hang in there a bit longer. I'll ping him and ask that he provide an update so we can all know where it stands.

Tom

Posted: Wed Jul 17, 2019 3:54 am

Posted: Wed Jul 17, 2019 3:46 pm

Yeah I realize it isn’t something that everyone needs but for me going on long rides out in the country it could really save my ass. Definitely would like to hear how the other project is going.

Posted: Wed Jul 17, 2019 5:47 pm

Isn't running out of gas like a right of passage? It teaches you to monitor your trip meter and take frequent breaks while riding. That and remembering to turn off the petcock when you park it. Get burned a couple times and you learn, forever. I have 2 bikes with fuel gauges, and they're accurate enough but i don't really use them. Learning to ride on the CBF it has become ingrained in me to reset the trip at each fuel stop, and keep an eye on my distance.

These are life lessons.

That being said, I'm not trying to discourage anyone from learning something new or designing something to help make our lives easier.

Posted: Thu Jul 18, 2019 3:59 am

I agree that the NEED for a fuel gauge may be vanishingly small, but it sure would be nifty. How about when you fill up the tank and then forget to turn the petcock off reserve? It can bite you in the ass a few days or weeks later.

So yeah, the challenge of a fuel gauge is that owing to the irregular shape of the tank, the height of the fuel in the tank does not bear a simple relationship to the volume of fuel in the tank. In order to get a sensor to work with an analog indicator like an actual gauge, you need to discover that relationship and build a complex, non-linear algorithm into whatever you are using to tell the indicator what reading to display. If you try to drive an analog instrument directly with a fuel level-based sensor your readings will be wildly inaccurate. This applies not only to sensors that would go inside the tank and have a float mechanism, but also to any sensor that is trying to read the pressure of the fuel caused by gravity. That pressure is also a function of the HEIGHT of the fuel above the sensor, and is not directly related to the VOLUME of the fuel that appears to be giving rise to the pressure. I may be missing something, but I think that's what the OP's design is doing. And if so, then I don't see how it would be accurate with any sort of analog gauge. (And if I am wrong here, PLEASE tell me!)

The sensor I have been working on uses a load cell contained in a replacement for the front rubber tank mounts. That sensor is effectively weighing the tank and its fuel. If it worked, it would be able to sense and report a full analog range of fuel volumes. On a recent ride, I actually had the thing working pretty well. It is not ready for primetime, however. It is extremely sensitive to jostling of the tank, it takes a certain level of care to install (although installation is really very simple), you have to be very careful not to let the throttle and clutch cables touch the tank and throw off the readings, and its reproducibility has much room for improvement. It really has a long way to go, but I think with suitable programming of the microcontroller, I would be able to get it to play.

However, it may well be the case that you don't really need a smooth curve describing at every moment how much fuel is in the tank. In fact, since my display of choice is a 10-segment bar-graph LED, that means I have already given up on a full analog display, and that means that all I really need is to collect 10 data points relating various fuel levels to volume. And if that's the case, then we don't need to weigh anything. We can simply take sample readings from a level-based sensor at ten points from full to empty and turn on the appropriate LEDs.

Even that might be overkill. We could use a tiny RGB LED to show 4 fuel levels as Blue, Green, Yellow and Red (like my voltmeter and oil temp displays do). So then you would need only 4 data points to map into your microcontroller and you would be able to see 1/4 tank intervals.

In short, once you have abandoned the need for full analog behavior, it becomes much more manageable to relate a fuel pressure or a fuel level to some representation of fuel volume.

Unfortunately, I have had to suspend all work on my project because the engine has come up lame, and I need to tend to that first. But I intend to return to it as the first project once as my main ride is back on his feet.

I would love to hear more about the in-line fuel sensor, CBNooby. Can you share? I'd be more than happy to go into excruciating detail with you about what I've learned so far at my end.

Posted: Thu Jul 18, 2019 4:15 pm

https://www.lskelectronics.com/fuelgaugeprouni

I'm probably going to be using this when the time comes however I won't have stock gauges.

Posted: Thu Jul 18, 2019 6:07 pm

nathanhouse wrote:

I'm probably going to be using this when the time comes however I won't have stock gauges.

As near as I can tell from looking at the Calibration procedure they give in their user manual, they will have you calibrate a setting at full and again at empty, and then they will interpolate all other readings between full and empty. And my guess is that they use a simple linear interpolation. I can't imagine how that could give you accurate readings using an analog gauge on an odd-shaped tank.

To avoid disappointment, I think you should question them closely before relying on it or committing to the 100 euro price tag. I'd love to hear what they say.

Posted: Thu Jul 18, 2019 6:17 pm

Looks similar to this: https://www.ebay.com/itm/FuelGaugePro1-universal-Float-free-motorcycle-fuel-gauge-meter-pressure-sensor/222736304812?hash=item33dc1e4aac:g:zqUAAOSwMQ1b5X7j

Description suggests individual LED's can be level calibrated.

Fancier version: https://www.ebay.com/itm/FuelGaugePro2-universal-Float-free-motorcycle-fuel-gauge-meter-pressure-sensor-/223015647854?_trksid=p2047675.m43663.l10137&nordt=true&rt=nc&orig_cvip=true

Posted: Thu Jul 18, 2019 10:48 pm

PGSmick wrote:

nathanhouse wrote:

I'm probably going to be using this when the time comes however I won't have stock gauges.

As near as I can tell from looking at the Calibration procedure they give in their user manual, they will have you calibrate a setting at full and again at empty, and then they will interpolate all other readings between full and empty. And my guess is that they use a simple linear interpolation. I can't imagine how that could give you accurate readings using an analog gauge on an odd-shaped tank.

To avoid disappointment, I think you should question them closely before relying on it or committing to the 100 euro price tag. I'd love to hear what they say.

It would be on a digital aftermarket gauge. It seems to me that every car I've ever driven for any length of time has a wierd drop somewhere on the gauge sweep. Stays full for a week then drops to empty in the next three days or something like that. . Im not super concerned with the accuracy throughout the entire range bc no matter what you end up doing you'll have to get used to how it operates.
A regular float doesn't know if the tank has 1 or 2 gallons in it just that it is in a particular part of it full travel, same as this sensor. I could be wrong. I still have a ways to go before I'm there so we'll see what happens.

Posted: Fri Jul 19, 2019 12:24 am

What I’m going after is much like lsk’s fuel gauge pro. I just couldn’t understand the price tag. It took me $30 and about 4 hours to make a sensor that would calibrate to my tank better than it seems theirs could. Basically it reads the fuel height in the tank through a pressure probe.

Surprisingly A digital sensor can work with an analog gauge use a digital potentiometer to mimic a float type gauge. That said a digital gauge would probably be a better option. I know a couple of the pictures I posted are analog gauges, I just used them as examples of what might be a good design for the gauge. The more I’ve thought about it though, the more it makes sense to me to find a way to incorporate an led bar graph or a few led’s in a small arc somewhere around the handlebars or triple tree. That way the dash doesn’t get cluttered with another large gauge.

As for the calibration, I’ve got a working version of my device that I’ve hard coded to read the level in the tank by pouring in a quart of fuel at a time and taking readings then interpolating. I’d say it’s probably accurate down to half a quart of fuel. Of course if you stay leaned into a corner for a decent amount of time it will throw off the reading but that will happen on any modern bike with a fuel gauge as well.

Posted: Fri Jul 19, 2019 1:52 am

I now have completely abandoned my idea of weighing the tank. I can see that you have gotten around the odd-shaped tank problem using a fuel level sensor. If I understand you correctly, with your interpolation between calibrations, you must also have essentially an analog output. It would seem like the world is your oyster.

SO if I were in your position, I would be looking at the LSK setup as your design goal. Incorporate the features you think are worthwhile and then build it. They have some pretty slick features you might want to incorporate. Their means of programming using a magnet to activate and manipulate the menu and using the 10 segment display as the programming interface seems sheer genius to me. I'm not sure you need quite as many programmable features as the LSK unit and you might be able to manage programming with just a few external jumpers instead. But I would definitely include a light sensor to automatically dim the display at night. I also think that the compactness of the 10-segment LED display gives a lot of mounting flexibility. For even more flexibility, I would separate the display from the microprocessor package.

If you can do that, and make yourself whole for less than 114 Euros, I'm happy to be your first customer!

Posted: Fri Jul 19, 2019 2:43 pm

Upon further thought I’ve come up with two different ideas for displaying the fuel level that are the least obtrusive I could think of and would work on a number of bikes regardless of the modifications done. Both rely on the use of leds which are individually controlled and dimmable.
1- 3D print a small clamp ring to go around 1” or 7/8” handlebars that will house a column of 5-8 led’s. Each led can be turned off as the level drops. And different color schemes are possible.
2- Use a single multicolor led built into the neck cap/hole at the top of the neck of the triple. This led will cycle through a range of preset colors as the fuel level decreases until it gets to red to show minimum fuel, maybe around half a gallon.

The sensor and control unit/sender package will be separate from the indicator they can be mounted hidden away behind the petcock. As for programming/calibration I haven’t yet decided if I want to go with a simple push button or another scheme. Just need a way to keep things simple but still water tight.

Posted: Sat Jul 20, 2019 6:37 am

I would be most interested in the sensor portion with no need for a display. Best case I could poll the microprocessor for a digital reading of fuel level, worst case I suppose reading an analog value would suffice.

Hawk Joined: Jul 14, 2012 Posts: 389 Location: New Zealand

Didn't early kawasaki's have a digital fuel gauge with a bar graph indicator ?

CB1100F Joined: Feb 17, 2009 Posts: 2111 Location: Oz

PGSmick wrote:

nathanhouse wrote:

I'm probably going to be using this when the time comes however I won't have stock gauges.

As near as I can tell from looking at the Calibration procedure they give in their user manual, they will have you calibrate a setting at full and again at empty, and then they will interpolate all other readings between full and empty. And my guess is that they use a simple linear interpolation. I can't imagine how that could give you accurate readings using an analog gauge on an odd-shaped tank.

It’s based on fuel weight (pressure) so I imagine it’d be quite linear.

Posted: Sat Jul 20, 2019 5:14 pm

DonR wrote:

PGSmick wrote:

nathanhouse wrote:

I'm probably going to be using this when the time comes however I won't have stock gauges.

As near as I can tell from looking at the Calibration procedure they give in their user manual, they will have you calibrate a setting at full and again at empty, and then they will interpolate all other readings between full and empty. And my guess is that they use a simple linear interpolation. I can't imagine how that could give you accurate readings using an analog gauge on an odd-shaped tank.

It’s based on fuel weight (pressure) so I imagine it’d be quite linear.

PGSmick wrote:

That pressure is also a function of the HEIGHT of the fuel above the sensor, and is not directly related to the VOLUME of the fuel that appears to be giving rise to the pressure.

Posted: Sat Jul 20, 2019 5:38 pm

PGSmick wrote:

That pressure is also a function of the HEIGHT of the fuel above the sensor, and is not directly related to the VOLUME of the fuel that appears to be giving rise to the pressure.

[/quote]

Yeah the pressure and volume are not directly related which is the reason for calibrating for a specific tank and position of the sensor.

The LSK calibration is similar to the way that I do it. You tell the controller what empty is (basically fuel in the line but only up to the petcock and then pour in fuel a bit at a time calibrating each individual light to a certain fuel height. So their way of programming is entirely user defined. Not necessarily even linear. You could set the lights to all go off at the exact same pressure if you wanted. Either way this would be a quantized calibration with arbitrary points. Their linear gauges instead use linear interpolation but only allow inputting empty and full positions so it doesn’t compensate for tank shape.
My calibration instead uses multi-point linear interpolation by pouring in a specific amount of fuel and calibrating each time more is poured in. This gives the possibility of having a full 10 bit range of outputs that can be used to accurately drive an analog gauge or a digital gauge with any number of lights/bars.

Posted: Sun Jul 28, 2019 5:45 pm

An update on my progress- I’ve gone with a cap that fits in the top of my bike’s neck. A single LED changes color as the fuel level changes. Wiring is simple because the light only needs one data wire. I’ve got auto dimming working but still have yet to put the sensor in the cap. There’s a full menu to choose different coloring schemes with different numbers of levels, set calibration points and choose the length of time it averages over. I’m just using a push button at this point to navigate through the settings. I’ve still got some work to do on the calibration and programming side but I should have everything working together on the bike within the next few weeks. Also you’ll notice from my pics that I have a cbr600 front end so obviously I’ll have to tweak the dimensions of the cap slightly to fit the original neck or any other aftermarket necks but this should be fairly quick and painless to do in cad given the dimensions.

Posted: Sun Jul 28, 2019 5:57 pm

Looking really good! I'll be interested when you get the final product done. Is your sensor internal or external of the tank right now?

Posted: Mon Jul 29, 2019 9:58 pm

nathanhouse wrote:

Looking really good! I'll be interested when you get the final product done. Is your sensor internal or external of the tank right now?

The sensor unit is external. I have a tee between the petcock and fuel filter that goes to the sensor. There’s plenty of room under the tank to hide it away so you can’t even tell anything is there.

Posted: Sat Jun 06, 2020 2:27 pm

Is this a finished project now? What did you end up with? I'm still looking to replicate this.