Overheating power regulator?

[hummingbird]

Dave, why not use a switching regulator like LM2576-5.0, and nip these problems in the bud?

My unit's regulator gets warm, and I have not seen how far cooked it becomes when I run with full backlight in the night, but I have not noticed any toasty smell or other signs that is acting up. But fixing issues one by one like the use of a switching regulator would solve issues one by one, particularly for folks that simply buy the unit and plonk it on the dash.

I know that was not the intention to begin with, but now making some such arrangements is really a form of public service. Wotsay?

[dcb]

Without proper diagnosis, we don't know what the problem is or how to fix it. I'm working with Tim offline to get a peek at his device.

[hummingbird]

How about a response to the switching regulator suggestion in post 21 above, Dave?

[dcb]

that was the response switching regulators are more expensive and complicated and I don't know if it will fix anything.

FYI, the newer revisions do not use the regulator to power the backlight.

[Magister]

To prevent overheating on my device, I used a top ground plane and the regulator is screwed on it to make a better contact, so the screw and the PCB act as an heatsink.

[dcb]

FYI, what I did was route the backlight through a 1 watt resistor and a npn, so the regulator and capacitors should power the cpu longer in a power glitch, and it also makes it easier to tie it up to an existing panel dimmer circuit or somesuch and still have cpu brightness adjustment on top. The most the regulator should ever see now is 20ma (even though 80ma is within "spec").

[dcb]

So I guess in essence, I am switching the power to the LED backlight, bypassing the existing regulator. It currently is in many respects a hybrid switching regulator that is powering the backlight, controlled by the cpu, limiting the current to 60ma when on full, and leaving more capacity to keeping the "brain" alive in a glitch.