Overheating power regulator?

[bobski]

Car horns throw out huge amounts of electrical noise in operation. They combine a big electromagnet with a set of contacts. When power is applied to the horn, it flows through the contacts to the electromagnet coil, which pulls on the contacts as well as the horn bell/diaphragm. When the coil has been powered long enough, the contacts are pulled open, which breaks the circuit. Momentum keeps the contacts open for a brief period, but spring tension eventually pulls the contacts and horn bell back to their original position, re-making the circuit and restarting the cycle. This happens several hundred times per second, producing the horn tone.
The electrical noise comes from arcing across the contacts as they open, as well as inductive flyback from the electromagnet coil.

Inductors (electromagnet coils) act like capacitors in that they store energy, but rather than trying to smooth out voltage, they try to smooth out current flow. If a charged capacitor is shorted, it will push as much current through the short as necessary to maintain capacitor voltage across the short, until the capacitor is discharged. Conversely, inductors will ramp the voltage up as high as is necessary to maintain the original level of current flow through a high resistance load, such as an opened set of contacts, until the inductor is discharged. This voltage spike is known as inductive flyback.

Anyway, to deal with the horn issue, you need to filter out that noise with a capacitor or something on the power input lines. I would put it on the car side of the voltage regulator.

The overheating regulator is a load issue. Regulators are spec'd to handle a certain amount of power dissipation (watts of heat); this number is generally higher with a heat sink.
[power dissipation in watts] = [voltage across the regulator in volts] * [current through the regulator in amps]
Voltage across the regulator should be supply voltage (car battery voltage) minus the regulator output voltage (5V). Since these voltage regulators are basically resistive elements, current through the regulator should be the same as the current drawn by your MPGuino circuit.
The real question here is what has changed to make the circuit overload the regulator. Unless your alternator is failing, we can reasonably assume that supply voltage and the regulator output voltage haven't changed, which only leaves an increase in current draw by the MPGuino circuit. I would poke around and see if any other components are heating up (drawing more power than they should). If not, go over everything with a magnifying glass hunting for shorts (stray wire strands, solder bridges and such).

[tim3058]

So I tried the heatsink trick... cut out a section (3/8" by 3/8") of an old pc video card heatsink and mounted it on the regulator with some heatsink paste. The heatsink gets hot, and I am still having the same issues. Next step is either a larger sink or trying a fan. Almost wondering if something in the MPGuino/my wiring went bad? I went the first 3 or 4 months without problems save for a random reset once every other tank or something. Contrast was on 25, and the backlight was off this morning, I barely made it 6 miles with the heatsink and it started resetting every 30s or so til I unplugged it.

[dcb]

can you measure the current draw? it should be about 20ma with the backlight off, and 80ma with it on full. Something sure doesn't sound right.

[spitsnrovers]

Since I noticed that there are only 10 wires between the main PCB and the dislplay, would it be possible to mount the two units separately?

The advantages I'd think might be gained are:
1. thinner flat unit display could mounted in or on a dash, or in a smaller case,
2. perhaps less heat in the overall unit,
3. PCB mounted in perhaps a more convenient location?

Perhaps the downside would be that the 3 buttons to change displays would be separate from the display.

I've seen the Volvo units mounted in the dash clock while the control knob is on the right of the steering wheel, on the dash.

[bobski]

Originally Posted by spitsnrovers

Since I noticed that there are only 10 wires between the main PCB and the dislplay, would it be possible to mount the two units separately?

Yes. The communications between the processor and display are relatively crude, so simply extending the wires should work fine. Same goes for the buttons actually.

[tim3058]

Originally Posted by dcb

can you measure the current draw? it should be about 20ma with the backlight off, and 80ma with it on full. Something sure doesn't sound right.

Checked the current last night, looks to be 20ma with the backlight off, and 40ma with it on low. The car sat running for just a minute or so and the gauge started resetting again. I had the door open the whole time, and it was snowing while I worked, so ambient temperature ~ 30* and windy... the regulator felt a lil warm, but no way could it be overheating that fast in freezing temps.

[Magister]

Why are you not using a low power LCD display? I am using some from Newhaven like the NHD-0216K1Z-FSB-FBW (blue), and it's an FSTN so display is black and not "grey", and at full power the LED takes only 20mA so I can drive it directly from a pin of the AVR, no transistor, no resistor (it's built in the LCD to drop to 4.2V max).

[dcb]

Not blue, that proved to be very unergonomic especially at night, but there may be some reduced power options available. But still tims unit seems to be making more heat than normal, even with the backlight on low (I assume).

[Magister]

Originally Posted by dcb

Not blue, that proved to be very unergonomic especially at night, but there may be some reduced power options available. But still tims unit seems to be making more heat than normal, even with the backlight on low (I assume).

Of course, I have used white, red, amber, etc, all the colors are available in low power, but I agree something is wrong with tim unit.

I used my first OBDuino with an Arduino (and its 2 LEDs) and the same first blue LCD as yours, so all in all taking 80+mA, in a black fully closed enclosure in South Carolina for 2 weeks of summer, and it never thermal shutdowned.

[tim3058]

Magister:

I bought it through the link on here and it worked fine for months, its the display it had on it, I was happy with the night visibility so I left it that way.

4 thoughts on the problem, maybe someone gets an idea:
1) The device randomly totally locks up, no response hitting buttons, numbers are frozen while car is driving, then randomly reboots at some point and sets to zero. Once it starts rebooting it will do it randomly every 20-60s, after a few of those the display goes a darker shade and numbers are illegible (usually unplug it right after it does this).
2) I notice also during the brief time the MPGuino is functional, if I start pressing near the small silver device soldered into the back center of the board near the base, I sometimes get a more-often reset, and can alter the display appearance (gets dark/illegible numbers). Coincidence? nearby parts being affected?
3) Also tried removing the Atmega chip the other day after a series of resets, checked the pins and replaced it firmly in its socket... worked good for 10 mins and then back to rebooting (maybe not a good idea, but I was curious).
4) As Magister said, his worked fine in the SC summer heat in a box, mines in NY, its been 30s-70s tops so far this spring, and the whole back of the MPGuino is exposed. It's been in the same setup for months, why would it be overheating now?


Thanks for the ideas thus far, that thing was handy, I miss it