DIY Capacitive tank sensor

[FW28z]

I got a chance to work on the circuit board a bit, and am pretty close to a schematic. However, the laptop computer that I used to draw with crashed today (literally), and I have to replace the screen.

So I am using my iMAC, but unfortunately I don't have a drawing program on it.

basically it is an interesting concept. When you depress a button,
it energizes a 9.1V Zener diode that powers the lm339 and the tank sensor through one leg of the resistor ladder.

One input of three of the comparators is connected in parallel, and also connected to the resistor ladder at the measuring point. The other input of the comparators is connected to another resistor ladder as the reference. Pretty straightforward.

The output of the comparators turn on the level LEDs.

I will work on a schematic when I get the new monitor in for my other computer, or if you want to see it before then, I can draw something and scan it in.

I need to double check my work, but I think I pretty much have it finalized.

[JamieGeek]

Cool. No hurry though. I know we are getting a little impatient now that the weather appears to have switched to spring.

[JamieGeek]

I think I've settled on a final circuit:

The main difference here is that Q4 is now a MOSFET (this guy: NTE 2980 Datasheet). Due to the fact that when I attempted to use a NPN (same one as the other 3: 2N4401 I had to use too much current to turn it on which affected the sense pin). Using the MOSFET makes the wakeup circuit use 0 ma current because its switched by voltage, not current.
Here is what the prototype looks like: LOL

(Of course the wife took a look at that and said: "So is this what its going to look like stuck to the side of the kitchen cabinet??" LOL Answer: "um NO! Its going to be soldered up inside a box next to the black tank out of sight").

The green LED on the far right is 1/3. The yellow LED is 2/3 and the clear LED is Full. The black part to the left of the Full LED and its transistor is Q4 the MOSTFET. The two resistors on the far left going vertical and then a third one going to the top "rail" (+5V) is the 67k and 120k resistors representing the ones in the probe harness.

Measuring the current: when operating with a 12V power supply about 30ma when asleep 1.4ma (according to my fluke).

Next steps:

• Wait for all the snow to melt (with the weather this week we'll be on our way).
• When I can bring the RV home I'll install the foil on the side of the tank
• Use the prototype circuit to calibrate the max capacitance value
• Solder up the main circuit
• Test circuit on RV
• Mount the box
• Find spare 12V fuse on the forward fuse panel (looks like there are a few)
• Connect up circuit
• Press "Black tank" button on the panel

I have the code setup such that it blinks the next level indicator via a PWM of sorts in 10 steps (10% on/90% off for 1/3 .1 full, 20% on/80% off for 1/3 .2 full, 30% on/70% off for 1/3 .3 full for example). Thus looking at the panel you'll get a resolution of approximately 31 points (10 steps to 1/3, 10 steps to 2/3, 10 steps to full and full).

[FW28z]

Nice idea on blinking the LEDs to increase the resolution of the 3 LED display.

I'm having second thoughts... I might just want to re-build the whole thing. I like blinkies and displays, so I bought a ST7565 graphics display ($17) to try out:



There should be no issues working with an Arduino. I have used 16x2 displays with Arduino chips before, but never a graphics version.

I think it might be pretty cool to have a graphic representation of an analog dial.

See, I always make things more complex. You are a lot further ahead of me... my wife just gave me orders that "her" RV projects (that I am to do) take priority over mine.

But I told her - honey, if we overfill the black water tank, you'll wish we did my project first.

So she told me to buy those Horst sensors - which are now back in stock at www.rv-partsplus.com. If you can stand having a non-commercial package. Apparently Horst changed suppliers and they had a delay in printing the packaging, and they decided to ship them without the over-package for those that want them.

So looks like I am putting in Horst sensors at least temporarily so I can get the other projects done.

But I envision myself setting at the campground soon with a beer in one hand and a voltmeter in the other hand messing with the circuits. No reason I cannot do it whilst camping...

[JamieGeek]

I figure my final testing will be at a campground as well. I'll probably gather a crowd too as people realize what is going on LOL. As long as it doesn't end like this, though


Heck with a graphical display you could just draw the tank with a fluid level in it. As long as you are replacing the whole panel you'll want to measure the other tanks as well (and battery voltage and propane level). Going further I'd also add an inside/outside temp to the display (I'm sure you'll have the real estate! LOL).

[FW28z]

Here is the display I got in the mail. I just uploaded the sample program to see how it works. Pretty cute...

https://www.youtube.com/watch?v=u6ZIX7wcylE

That movie cracks me up. You know that Jeff Daniels is a big RV'er and lives in Chelsea, about half way between you and me. Maybe we could meet there one day and have a party...

[JamieGeek]

Yeah go see a play at the Purple Rose Theater.

If you really want to go nuts: Put a Raspberry Pi in the coach on its own WiFi network. You can use one of the TVs as a monitor and have the Pi serve up a webpage with all of the RV's stats. Then you could just check it with your phone as long as you're within WiFi distance of the coach (and the Pi is so small it could be hidden anywhere--perhaps behind the TV).

[JamieGeek]

Well I'm all the closer to installation/deployment:

On the terminal block from left to right:

• Panel Sensor wire
• N/C
• N/C
• Tank sensor/capacitor
• Ground
• 12V

The four resistors at the top right are my two voltage dividers for the input comparator. The two resistors at the lower right is the RC resistor for the tank capacitance measurement and the pullup resistor for the wakeup circuit.

The big jumble in the middle is the 3 transistors and MOSFET connected to the sensor wire. The two resistors at top left are the 67k and 120k "replacement" resistors for the tank wiring harness.

Haven't tested out the circuit on the RV yet. Have to first install the foil "sensor" and sealant over it. Was going to do that today but there was a lake at the storage lot that I didn't want to wade through.

[FW28z]

Looking good.

[JamieGeek]

"Sensor" installed:

Ultimately this will be covered with black sealant; it won't be able to be seen. It shouldn't be too important that it is irregular as its the surface area here that is important (if I really need more surface area I can cover another 2/3 of the tank to the right of the Tornado fitting there). I wanted to cover this side, though because the drain is right there at the bottom left.

[JamieGeek]

Update: On this "nice" Saturday (50s temp and mostly cloudy) I drove the camper home and setup shop:

I was able to test the output of the circuit: Could it really control the panel lights?

https://www.youtube.com/watch?v=r-Pr...ature=youtu.be

I also tried calibrating the sensor but soon discovered:

• I really need to do that at a campground with full hookup (so I can do multiple full/empty cycles and completely fill the black tank)
• I may need a bit more research as the measured value didn't change filling the tank to about a half

[FW28z]

One thing I noticed is that the commercial capacitive sensors use two identical size strips, about 3" apart. Looks like you have one huge strip and perhaps relying on the well nut for the other lead?

[JamieGeek]

In my testing with various sized vessels the most accurate readings were from grounding the fluid and using one sensor on the side. This makes the fluid one of the plates and the vessel wall the dielectric.
For the one test I did I simply used an alligator clip on the ground probe.
I did not disconnect the panel ground lead from the probe which may have affected the result.

[FW28z]

I'm sure you will figure it out.

[FW28z]

Well, we have not done much on this thread lately, so I thought I'd update it. I got a chance to work with some LCDs driven by an Arduino microcontroller today. After about 1 hour of work, I came up with this:



So far, it's static... meaning there is no changing of the display.

But there are a couple of interesting things.

First, if you use a parallel connected display, about 80% of the available I/O pins are dedicated to the display. Some displays also do not have any on-board memory, so the microcontroller has to provide it. This means that about 80% of the resources of the microcontroller are used just for the display.

But I am using a serial display, which has it's own memory, and only takes a single pin. However, you have to put in delays and change the baud rates or the display will write slower than the microcontroller sending data and will otherwise be garbled up. There is no bona-fide flow control to the display, so you have to throttle down the microcontroller so the display can keep up.

This sort of complicates things a bit, but far better than eating up the resources with a parallel display.

Anyway, this is not a final display design by any means. But it is a cool first attempt.

I can probably get more bar-graphs on the display if I run them horizontally rather than vertically. Unfortunately, you cannot make the fonts any smaller which means the text takes up a lot of space. Perhaps I can come up with some symbols for the various tanks...

I could use a flame graphic for LP, a water droplet for the fresh water tank, a toilet for the black tank, and I am not sure what to use for the grey tank.

[JamieGeek]

Originally Posted by FW28z

I am not sure what to use for the grey tank.

A sink or shower.

I haven't moved forward with my efforts simply because I need to have a full hookup so I can fill/dump for testing. This means we have to go camping. Unfortunately I haven't had the opportunity for that--as they say life intrudes.

[FW28z]

Version 2 (not yet completed).

Not sure what to put in the right side. I thought I'd make it a battery display, but with sensors to determine if the charger is running, and if it is in Bulk, Absorption, or Float mode (characters might be too big though), and possibly a thermometer at the bottom right.



The arrows work a lot better as they print much faster, so you don't need a bunch of delays.

I should do a video as I am putting the microcontroller into a loop, and using a random number generator to generate a number from 1-100 (to mimic the input value from the sensors), and the display does update to reflect that number.

So in real time, the display values change about once a second with the new number from the random generator.

The display is about 2.5 wide x 1.5 high, so it would easily fit into a panel.

I could also make sub-displays but I am just getting started.

[JamieGeek]

For battery, since you have the room, I would just put the voltage. You can tell when in different charging modes simply by the voltage of the battery.

I agree your version 2 looks better and the divisions make it easier to tell 1/2, 1/4, etc. with a simple bar graph with no graduations its difficult.

[FW28z]

I might tweak it a bit more, but so far, I am liking this graphic display.



Quirky little thing though. The display only takes integers or chars for the data type, and I was having a heck of a time trying to get the decimal into the voltage readout. I finally figured out that you have to use *10.0 rather than *10 when multiplying the fractional part of the decimal by 10 to get an integer representing the tenths digit.

Also, what nit-wit created the character map? The lower-case "t" looks awful.

I'm thinking of maybe tweaking it a bit more. I might replace the circle bullets in the battery status with the same triangle points on the graphs, or perhaps a check mark. Also, I might make a small bell icon to represent whether or not a graph is alarmed.

I'd like to add a % symbol to the numbers on the bottom of the graph, but it kind of clutters things.

Since I will be detecting voltages (and whether or not the combiner/charger is on), I can put in a different mode where the display is always on (with combiner running), or only on for a few moments (on battery only).

[JamieGeek]

Some displays allow you to upload your own font (granted that would take a lot of time designing your own font..perhaps you could design your own "T" and only use that font for that letter??)

Looks good.