TheSamba.com :: View topic

Fri Dec 31, 2010 6:49 pm

busman78 wrote: Suspect that the 5V for buses and 6V for bugs is nothing more than factory max volts, cause no vibrator I have tested reaches the 5V range.

Did some more testing with the 5V Regulators today, seems after a minute or so the voltage begins to climb, the LM7805 triped past 10V and the NTE climbed to 9V, so fixed regulators they are not. And with the voltage increase the needle rises on the gauge.

something's not right there. I've used linear regulators a fair bit and their voltage has been pretty consistent unless you are measuring them in a no-load condition ie. completely disconnected. Even with the low draw of a microcontroller they stay stable. I used them in embedded projects and even a butchered micro atx power supply which I needed a linear reg to add a -5v rail to for a substitute power supply for my apple 2e.

Fri Dec 31, 2010 7:01 pm

Quick self reply / correction. The "other vibrator" is just a zener and light bulb. I have the schematic somewhere. From this thread I'll assume the zener has a 5v or 6v breakdown voltage.
The light bulb is a current limiter. Connecting a zener from +v to ground without a current limiter is ...not good. This gives a current limited stable low voltage supply for the sender and gauge

Fri Dec 31, 2010 11:03 pm

Not sure what you mean by loaded but the 5V regulator was screwed to the back of the housing, 12V battery in, center leg grounded, and power out to the gauge, for about a minute, maybe a little more while the gauge moved to full the volatge was at or slightly below 5V, after the gauge moved to full the voltage out started to increase.

Sat Jan 01, 2011 12:54 am

busman78 wrote: Not sure what you mean by loaded but the 5V regulator was screwed to the back of the housing, 12V battery in, center leg grounded, and power out to the gauge, for about a minute, maybe a little more while the gauge moved to full the volatge was at or slightly below 5V, after the gauge moved to full the voltage out started to increase.

That seriously should not be happening. If you are using a multimeter could you please humour me and check the current flowing through the output of the regulator over a couple of minutes?

Sat Jan 01, 2011 5:10 pm

Quote: Did some more testing with the 5V Regulators today, seems after a minute or so the voltage begins to climb, the LM7805 triped past 10V and the NTE climbed to 9V, so fixed regulators they are not. And with the voltage increase the needle rises on the gauge.
Three-terminal regulators need typically need bypass capacitors from their input and output pins to ground to ensure stability. See the device datasheet. Driving an inductive load like a fuel gauge without them may be problematic. FYI

Sat Jan 01, 2011 5:29 pm

telford dorr wrote: Quote: Did some more testing with the 5V Regulators today, seems after a minute or so the voltage begins to climb, the LM7805 triped past 10V and the NTE climbed to 9V, so fixed regulators they are not. And with the voltage increase the needle rises on the gauge.
Three-terminal regulators need typically need bypass capacitors from their input and output pins to ground to ensure stability. See the device datasheet. Driving an inductive load like a fuel gauge without them may be problematic. FYI

I never thought to ask if the supporting components were there. Good call.

Sat Jan 01, 2011 6:07 pm

Did a little more playing today. Set the test up, with battery connected and sender in the full tank position the gauge rose quickly to full, meter indicated 4V, after 42 seconds climbed to 5V and at 1min 22sec meter read 6V, gauge needle was pegged to the far right side of the gauge. Moved the sender arm to the half way position, immediately the volt meter went past 10V and pegged, before I could switch the meter to the next scale smoke poured out of the gauge. Kiss that gauge good-bye.

Set it up again with another gauge and new 5V VR, got the same full tank volt readings, falling within seconds of the 1st test, before moving the sender arm switched the meter to the next scale, moved the arm, volts went past 12V smoke started to emit from gauge. This gauge did survive, tested it with a vibrator and got good stable readings.

Now I am no electronics person, vehicle electrical I understand, I think I will stick with the vibrators.

Sat Jan 01, 2011 6:15 pm

busman78 wrote: Did a little more playing today. Set the test up, with battery connected and sender in the full tank position the gauge rose quickly to full, meter indicated 4V, after 42 seconds climbed to 5V and at 1min 22sec meter read 6V, gauge needle was pegged to the far right side of the gauge. Moved the sender arm to the half way position, immediately the volt meter went past 10V and pegged, before I could switch the meter to the next scale smoke poured out of the gauge. Kiss that gauge good-bye.

Set it up again with another gauge and new 5V VR, got the same full tank volt readings, falling within seconds of the 1st test, before moving the sender arm switched the meter to the next scale, moved the arm, volts went past 12V smoke started to emit from gauge. This gauge did survive, tested it with a vibrator and got good stable readings.

Now I am no electronics person, vehicle electrical I understand, I think I will stick with the vibrators.

You didn't check the current like I asked, did you?
I'm pretty sure because you don't have a current limiter you are overloading the regulator, it misbehaves when you move the sender changing the load on the circuit and POOF! you've cooked it.

In this setup you need a current limiter. Need.

Sat Jan 01, 2011 7:58 pm

Check the current? The meter does AC/DC volts, it will do Ohm's, there is no scale for current.

Sat Jan 01, 2011 8:21 pm

The gauge being heat driven, you really need to calculate the wattage. High wattage output [resulting from no/low resistance] will fry the gauge and/or sender. With your current readings for Ohms and Volts you can figure out the Current[amps] and Watts[power], the site below may help.
http://www.ohmslawcalculator.com/ohms_law_calculator.php

I highly recommend adding some form of resistance.

Sat Jan 01, 2011 9:11 pm

timo78 wrote: The gauge being heat driven, you really need to calculate the wattage. High wattage output [resulting from no/low resistance] will fry the gauge and/or sender. With your current readings for Ohms and Volts you can figure out the Current[amps] and Watts[power], the site below may help.
http://www.ohmslawcalculator.com/ohms_law_calculator.php

I highly recommend adding some form of resistance.

Absolutely. With my experiments on this I tried using a resistor mostly for tweaking the range of the gauge I was using, but it just wasn't the right thing for the job. A light bulb is a dynamic resistor which by design can deal with energy dissipation effectively. Why dynamic? because the more current that flows, the hotter it gets and the greater the resistance. Great for keeping a circuit like this in check.

I know I sound like a broken record, but Zeners are cheaper than linear regulators, and little light bulbs are cheap too. not only that, but this was the configuration VW used as one of their "vibrators".

Sat Jan 01, 2011 9:25 pm

Augh! I did a write up on the gauges and vibrators, and detailed how I got the whole mess working with an aftermarket gauge but it seems to have fallen off the internet. Typical.

Sat Jan 01, 2011 11:38 pm

busman78 wrote: Check the current? The meter does AC/DC volts, it will do Ohm's, there is no scale for current.
Buy a 0.1 ohm, 1 watt, 1% resistor [Fry's, or some on-line parts vendor (Mouser Electronics)]. Connect the resistor in series with whatever current you're trying to measure. Measure the voltage drop across the resistor. Good to 3 amps: 0.1 volt = 1 amp; 1 volt = 10 amps, etc.

p = i * i * r; p = 3 * 3 * 0.1 = 0.9 watts

If you need to measure more current, use a 0.05 ohm, 25 watt, 1% resistor [good to 20 amps: 0.1 volt = 2 amps; 1 volt = 20 amps]

In either case, make sure the voltmeter connections to the resistor are physically closer to the resistor body than the power connections passing current through the resistor. Never use the same points for both connections.

-----

Sorry to hear you're having so much of a problem getting the series regulator to work - it should be fairly easy.

Most series regulators are current-limited and should not run away (as long as you keep the device cool by heat-sinking it to something. Also, the regulators need some bypass capacitance on the input and output of the device to ground (at least 0.1 uf / 25 vdc; 1.0 uf works better) Without it, the device may be unstable with a non-resistive load.

If you have any regulators left, breadboard one on the bench with a couple of caps soldered from the two end legs to the center leg. Mount the device to a piece of sheet metal (say, 6" x 6") for a heat sink. Apply power to the device's input pin from a 12 volt source through a rear tail light bulb (to act as a current limiter; wire both tail light and brake light filaments in parallel.) You should measure 5 volts on the output pin (even unloaded.) Now load the output with a small light bulb, like a dash gauge lamp. You should still measure 5 volts, and the bulb should glow. The current limiter bulb might also glow slightly.

As long as the regulator doesn't get too hot (like burn your finger hot), it should operate like this indefinitely.

If it does, it should be able to power a fuel gauge just as easily as the dash bulb.

The case of the regulator is connected to the center lead of the device (TO-220 case), and thus is electrically active. Make sure whatever you mount it to is either insulated, or at the same potential as the regulator case.

Keep trying - it shouldn't be this hard to get it working. Something is screwy somewhere that we're not seeing.

Pics?

Fri Sep 23, 2011 6:14 pm

I have since found and lost the article I wrote on gauges again.

Resurrecting this because I'm considering using a micorocontroller so do a nice low current reading and PWM output to drive my gauge correctly. This also has the added advantage that if I cared enough I could probably calibrate the sender and gauge and even compensate for fuel tanks shape and slosh. If I did that it'd also be easy to add a trigger to one of the axles to measure speed and calculate economy, but it all adds up in time. I already have everything I need so $0.

I dug up this thread because I need values. Sender for a '75 max and min and vibrator mean voltage, which I can see is 10v.

I have a gauge pulled from a truck. Ages back I semi calibrated by initially trying to calculate the needed resistance to get range and level correct, realising my math is terrible and empirically doing it. Some things have changed and now it's out of whack so I need to fix this. I'm strongly considering mounting a few gauges between the ashtray and the stereo, because random gauges everywhere is ugly and my dash is hardly pristine anyway. I'd rather neat with roughly period correct added gauges mounted in the dash than a mess with exposed wiring blistering from various places.

Fri Sep 23, 2011 10:59 pm

Since you have to build your own interface circuit to connect the stock sender to the micro's A to D converter, you might as well use the resulting system to generate a calibration table. You can estimate the sender resistance range by measuring the sender resistance with the tank empty. Then assume somewhere near zero for a full tank. This gives you a baseline for your A/D front end circuit, which you want to design to use as much of the A/D range as possible.

I'd then use the resulting circuit to gather readings, starting with an empty tank, then adding a gallon at a time. This gives you the start of a data look-up table (which may not be linear) to generate the gauge output drive from.

Now you manually drive the gauge with a current source, plotting the values needed to linearly move the gauge from empty to full. You then design a PWM circuit to generate this current range (which again may not be linear), noting the PWM values required to move the gauge to the desired intermediate points.

Now all the microcontroller has to do is pass the A/D values through the final look-up table which equates the input values with the output values, then use he resulting values to drive the PWM.

Using this technique, you can literally interface any sender or sensor with any gauge or readout device.

Fri Sep 23, 2011 11:28 pm

I noticed this topic came back up, and have something to add on the solid state replacement. It has a tendency to induce gauge needle swing. Such as when you make a corner the sloshing gas moves the sender arm, which in turn makes the gauge needle swing up, or down a bit. I now believe one of the functions for the stabilizer/vibrator is to correct this. No biggy to me, but figured I'd let y'all know for reference.

Sat Sep 24, 2011 2:33 am

telford dorr wrote: Since you have to build your own interface circuit to connect the stock sender to the micro's A to D converter, you might as well use the resulting system to generate a calibration table. You can estimate the sender resistance range by measuring the sender resistance with the tank empty. Then assume somewhere near zero for a full tank. This gives you a baseline for your A/D front end circuit, which you want to design to use as much of the A/D range as possible.

I'd then use the resulting circuit to gather readings, starting with an empty tank, then adding a gallon at a time. This gives you the start of a data look-up table (which may not be linear) to generate the gauge output drive from.

Now you manually drive the gauge with a current source, plotting the values needed to linearly move the gauge from empty to full. You then design a PWM circuit to generate this current range (which again may not be linear), noting the PWM values required to move the gauge to the desired intermediate points.

Now all the microcontroller has to do is pass the A/D values through the final look-up table which equates the input values with the output values, then use he resulting values to drive the PWM.

Using this technique, you can literally interface any sender or sensor with any gauge or readout device.

Yes. Yes. Yes. It would seem as though you read my mind. Although gallons is a useless measurement for me and it may actually be easier to fill the tank and decant it out by some unit at a time from the tank to lessen errors from pouring.
It is my understanding that an effective method of reading the sender would be a similar method to reading a photoresistor.

PWM either hardware or bitbanged should be more than sufficient to drive a gauge.
If I don't have any spare microcontrollers I may have to bite the bullet and try one of those kit ones like an arduino. I've always built my hardware from scratch so that'd be a bit odd for me. I cooked a few microcontrollers working on one project before I found the issues. One thing I learned was something very strange about those linear regulators. If they are mis-wired they still work effectively albeit differently. Nearest I could tell is it was pulling the ground potential up to make the correct differential. This makes them run hot but they still work accurately. Then you add a second, correctly wired regulator into the mix, perhaps providing a different voltage. All hell breaks loose and the electronics in the middle get very toasted. Anyway now I re-re-re-re-re-check the exact package type on regulators.

If I build from scratch again I'll have to drag out my testbed computer. A modified C=64 :lol:

Sat Sep 24, 2011 10:44 am

Quote: It is my understanding that an effective method of reading the sender would be a similar method to reading a photo-resistor.
Yes, except the resistance will likely be a lot lower, so compensate. Be sure to design it to use as much of the A/D converter range as possible. Also design it so any inputs can be grounded or hooked up to +12 without frying anything (use a series resistor, bypass cap, and diode clamps on all inputs).

Quote: PWM either hardware or bit-banged should be more than sufficient to drive a gauge.
Definitely. Many micros have counter-timers and generate a PWM output. Run that into a R-C low-pass filter network, buffer with an op amp or transistor, and you have a nice analog signal to drive the gauge with.

Quote: If I don't have any spare microcontrollers I may have to bite the bullet and try one of those kit ones like an arduino. I've always built my hardware from scratch so that'd be a bit odd for me. I cooked a few microcontrollers working on one project before I found the issues.
I've had good luck with PICs, as I have a programmer for them. They're fairly easy to use, and have A/D converters on board.

Quote: One thing I learned was something very strange about those linear regulators. If they are mis-wired they still work effectively albeit differently. Nearest I could tell is it was pulling the ground potential up to make the correct differential. This makes them run hot but they still work accurately. Then you add a second, correctly wired regulator into the mix, perhaps providing a different voltage. All hell breaks loose and the electronics in the middle get very toasted. Anyway now I re-re-re-re-re-check the exact package type on regulators.
All the problems that I've seen from them are usually related to inadequate bypassing of the input and output, making them oscillate like crazy and get really hot. Easy to see with a 'scope.

Fri Nov 04, 2011 2:39 am

This evening I made a thing.

It's pulling around 40mA so that's not so bad. As for calibration, short answer, no. But it is better than scorching a sender.