Page 2 of 3

Re: ok,, New Project time,,,

Posted: Wed Jan 30, 2019 1:42 pm
by Janitor Tzap
dacflyer wrote: Tue Jan 29, 2019 3:44 pmI'd like something like this, but Damn the price.. how do they justify the price
https://fuelminder.biz/lpg%20gauge/lpg% ... 0info.html
Looks like most of the cost goes in to the 50 feet of wire, and the "ZENER BARRIER" surge protector.
It makes sense, you don't want a stray lightning strike to hit the wire and travel back to the tank and cause an EXPLOSION!

But, yeah....
Compared to the Fuel Oil Monitoring Systems.
The Propane Monitoring Systems are twice to three times as much. :(

I know the Temp Circuit isn't right, but I see it has a 10K "Gain Adjustment" POT on pin 13 of IC1.
Maybe try a different value POT to get more sensitive readings.


Signed: Janitor Tzap

Re: ok,, New Project time,,,

Posted: Wed Jan 30, 2019 2:32 pm
by Lenp
Use a micro, and A/D conversion with a look-up table to correlate the digital values to the tank contents. As a lower tech approach, chose several pot adjustable set points to get an LED output for any tank levels desired. As for an analog design, you should be able to use an op amp and send the low voltage from the sensor output into the op-amp and it could output a linear output nearly up to the op amp supply voltage.

Re: ok,, New Project time,,,

Posted: Thu Jan 31, 2019 11:10 am
by dacflyer
i am about at my wits end on this.. Grrrrr
i was doing some more testing,, the output maxes out at 4.08v at about 78% of full. anything beyond that is the same.
there is no way to tweak the sensor as it snaps onto the dial face ( it's isolated from the tank. )
i think that will be my main obstacle to over come.
i want to keep this as analog as possible,, i don't wanna use a micro. if at all possible.. (last resort )
i was thinking to try a analog meter..have to figure out that with a OP amp i am sure.
i was also thinking about using a 5 or 10 led display. LM3014 or such. but if i did that, i'd want to figure out a way to make 2 or 3 leds flash when the fuel got low. then steady when the other leds turned back on.
i did something like this with a temperature meter i built.. LM3914 and a 555.. in dot mode. the led would blink after the temperature got past a certain part. just have to think out how to trigger it..
i need ideas.... I'm not giving up on this yet..

Re: ok,, New Project time,,,

Posted: Thu Jan 31, 2019 12:02 pm
by haklesup
I think using pressure alone to determine the Volume of the tank is always going to have limitations since the gas is probably more elastic at low pressure than at high pressure, the entire function is probably non linear but linear enough in the 5 to 75% range that it works good enough for a BBQ tank.

If you want precision, you need to sense something else like level like the adhesive thermometer strip gauges do but that don't work well in the cold. Most other solutions will be more complicated that what you have tried. Maybe you can find something ultrasonic

This looks interesting and reasonably priced. Maybe you can sniff out the bluetooth signal and your project can be a wall mounted monitor instead.
https://www.pplmotorhomes.com/parts/ele ... or_06.2191

No info on accuracy or range but there is a video link though I lost confidence when they showed the readout wildly swinging between 2% and 80% and read the comments. This seems to be sensitive to tank construction and the sender does appear to be ultrasonic so could have full range if done right.

Re: ok,, New Project time

Posted: Thu Jan 31, 2019 1:43 pm
by Lenp
Is it possible to disassemble the standard tank gauge and fit a potentiometer to the shaft, or make a dial that has either a reflective path or or slots that are proportional to the gauge indication. If so an output signal correlated to the gauge indication could be obtained.
Yes, tampering with the propane gauge may be a safety issue but if the gauge were sealed elsewhere than at the gauge face it may be quite safe.
Not knowing the tank size, since propane is a liquid, would measuring the tank weight work, since that's how they are filled?

Re: ok,, New Project time,,,

Posted: Thu Jan 31, 2019 7:00 pm
by Janitor Tzap
Hmmm..................

How about modifying a VU meter circuit?
Image
You could take the signal from the Sensor on the tank and run it in to an OP AMP,
before sending the signal to the Input of the VU Circuit.
You could then add a POT to the output of the OP AMP to adjust the signal strength for the VU Meter Circuit.


Signed: Janitor Tzap

Re: ok,, New Project time,,,

Posted: Sun Feb 03, 2019 7:50 am
by CeaSaR
dacflyer wrote: Thu Jan 31, 2019 11:10 am i am about at my wits end on this.. Grrrrr
i was doing some more testing,, the output maxes out at 4.08v at about 78% of full. anything beyond that is the same.
there is no way to tweak the sensor as it snaps onto the dial face ( it's isolated from the tank. )
i think that will be my main obstacle to over come.
i want to keep this as analog as possible,, i don't wanna use a micro. if at all possible.. (last resort )
i was thinking to try a analog meter..have to figure out that with a OP amp i am sure.
i was also thinking about using a 5 or 10 led display. LM3014 or such. but if i did that, i'd want to figure out a way to make 2 or 3 leds flash when the fuel got low. then steady when the other leds turned back on.
i did something like this with a temperature meter i built.. LM3914 and a 555.. in dot mode. the led would blink after the temperature got past a certain part. just have to think out how to trigger it..
i need ideas.... I'm not giving up on this yet..
Let's take a look at this from a fresh perspective.

You know the limits of the sensor and the ranges it represents, so this is what you have to work with, regardless of what you think it should be. So, let's work within those confines.

Wants and Needs:
1. Want an easy indication of level in the tank in order to know when to order a fill up.
2. Want an easily discerned warning for the low level.
3. Want to use the existing sensor.

A. "Need' LED "graph" to show level. Many to choose from and many standard drivers available. Seems you have a grasp on at least one, so use what you know.
B. Flashing the LED when level is low enough. Maybe use an inverting opamp and a potentiometer to set the trip point for a 555 timer to drive a transistor or two as a switch on the GND of the LED display.
C. Might need an opamp or similar driver to boost the signal from the sensor in order to consistently drive the LED display/driver.

With this preliminary/rough outline, there should be enough fodder for the mind to come up with something that'll work. Just keep in mind that the biggest goal is to know when the tank will need attention. Therefore, the top 22% that cannot be displayed due to the sensor issue, really isn't an issue. Just know that as long as you read full, you are at least over 3/4 full and that you'll have to be okay with that.

CeaSaR

Re: ok,, New Project time,,,

Posted: Wed Feb 06, 2019 4:25 am
by dacflyer
Lenp >> no possibility to modify the sensor to a pot,, not enough strength from the magnetic coupler.

Janitor >> i did connect a analog meter to the sensor with some resistors and a trim pot, only to find out the sensor is actually not very linear at all.
go figure.. i think next I'll try the Lm3914 IC, i found a circuit that has a low set trim and a hi set trim. so i can get my hi and low set points, then there is a 3rd trim pot for calibration. the LM3914 will be a single IC compared to 2 LM324's
this might work.. really wanted a digital read out,,but i think it's not gonna happen unless i goto a micro.

CeaSaR >> you may be right,, i might have to deal with something close to what will be good enough.
i'm going to try the LM3914 circuit i found, and see how this does. otherwise i'll have to figure out a way to safely reconfigure the sensor on the dialface.
or find another sensor to attach to the dialface that will have better results.

thanks guys.. stay tuned.

Re: ok,, New Project time,,,

Posted: Fri Feb 08, 2019 1:15 am
by Lenp
I think this was mentioned previously but....
Using an inexpensive micro, with an a A/D input, you would get a value, let's say from 0 to 255, representing the full range of the sensor's output.
If the output is a voltage it may need to be scaled, or if it is a resistance, than a voltage divider may be needed. In either case, it's basic circuitry!

So with some simple coding we might say,
if the value is...
<=1 the tank Is empty, light the 0'% LED
>1 and <=10 the tank is 10%, light the 10% LED
>10 and <=15 the tank is 15%, light the 15% LED
.
.
=255 the tank is full so light the 100% LED

If the sensor's output is reversed, just alter the reference numbers (eg. 255= empty, 1=full)

You establish the reference point numbers based on the sensor's output at various tank gauge readings. Sensor linearity is of small concern since you can correct it with the reference point values in the software. By using an LED dot display you could get dots or a bar graph, just by a coding change, or using a switch to alter the software's behavior. If you're interested, just let me know if you need some help!

Re: ok,, New Project time,,,

Posted: Fri Feb 08, 2019 5:45 am
by CeaSaR
Funny enough, I went in search of a more discrete version of a graph/VU driver and came up with the same one Janitor Tzap posted above (go to the base URL of the picture and search for "VU meter LM324") in order to customize the resistor chain to deal with the non-linear sensor.

The quick way to get the values is to use a potentiometer with a base value of the full string (between the upper and lower isolation values, ie, the 2.2K chain), then apply the sensor voltage for your required trip points, and dialing the pot into the turn on voltage of the LM324. It'll be the same value to search for each time, but since you'll be increasing your input, you'll be increasing the bottom resistance of the pot voltage divider to get there. Measure the resistance between input side of the pot to the wiper each time and then you can build the network up from as close as possible to those values.

For the blinking, my thought is that you only want them to blink when needed, yet stay solid when not needed. Build your 555 timer for the frequency you want them to flash, then take a tap from one of the yellow LEDs (between the LM324 and current limiting resistor), invert and buffer it if necessary, and tie that to the 555 reset pin. That'll keep it turned off until the sensor falls low enough. Now for the flashing of the LEDs. For those that will be affected, change their main current limiting resistors to a much higher value, something that will let them work, but at a much reduced brightness. Now take a complentary number of small signal transistors and the correct value current limiting resistors and place them across the high value resistors, drive them from the 555, and you'll flash the LEDs bright/dim. You'll retain the graph ability since you are not actually supplying power, but an additional "signal" path.

Now Lenp has the easiest way to do it AND include features, if you know how to code. Just read the input, a bunch of IF/THEN/ELSE statements that include which ones to turn on and how they are treated (solid or flashing). Tempting. Unfortunately I am only versed in BASIC, so others will have to help there.

CeaSaR

Here's another version to ponder:
http://www.next.gr/uploads/44/VUbargraph_OA%26BJTs.gif

Re: ok,, New Project time,,,

Posted: Tue Feb 12, 2019 12:46 pm
by dacflyer
Lenp >> thanks..
I'm going to try the LM3916 1st..
i really wanted a digital one,, but led will be fine enough,, just means a different housing.

Re: ok,, New Project time,,,

Posted: Tue Feb 12, 2019 2:34 pm
by Lenp
his has been said before, consider using a micro!
With the sensor going to an A/D input use that value to scale from 0 to 100% as needed. Typical digital output is 0 to 255, for min/ max voltage input. Linearity of the sensor, is not important, just the digital value for the cardinal points desired. Output can go to a dot/bar display.
Need more help, just ask!

Re: ok,, New Project time,,,

Posted: Wed Apr 17, 2019 11:40 am
by dacflyer
Update, I could not get the results I wanted with the LM3914, but later on had a :idea: Lightbulb :idea: moment.
I decided that i will use the LM324 IC, as I can program each OP amp for the exact trigger point that I want.
Since the sensor i am using isn't very linear.
Stay Tuned...

Re: ok,, New Project time,,,

Posted: Sat Apr 20, 2019 10:07 am
by CeaSaR
If, initially, you don't get good results with a direct read off the sensor, add a buffer amp to provide more current or even a higher voltage range. That's about the only caveat I can see with such a circuit, since it usually sees a much higher voltage range.

We'll be waiting on results. Good luck!

Re: ok,, New Project time,,,

Posted: Sat Apr 20, 2019 4:06 pm
by CeaSaR
Don't forget what I said earlier about the top 22% being basically disregarded. If you step back and think about it, the original dial / readout registered full at that point also. So no matter what, you still have the same resolution and accuracy of the original setup. If it was good enough for the manufacturer, then it should be good enough for (you) the end user.

CeaSaR