3
\$\begingroup\$

I'm trying to control brightness of some commercially obtained instrument panels (for a mock aircraft cockpit). The panels have some internal LED strings (presumably, anyway) which are explicitly rated to a 2.8v input (and no more), and unfortunately must be switched on the high side. Maximum current draw I've measured on the bench is about 60-70mA.

I have an existing PWM module used for various other tasks in the system, which has spare outputs. The PWM chip on the module is a PCA9685 which can sink a decent amount of current, but 60-70mA is beyond what it can cope with.

Therefore my plan was to establish a 2.8v power rail using a small linear regulator, and use my PWM input through some kind of driver/switch to give the final 2.8v-PWM-ed signal to the backlight input. The problem is I can't find a suitable driver which works in the correct voltage range, but has a fast enough response time for PWM.

There's plenty of 'slow' solutions,(< 100 hz) even toggling the enable line of the 2.8v regulator, and there's plenty of fast solutions (designed for PWM) for automotive, but which need 12v output.

(I don't think I can use simple BJTs because of the junction voltage drop: below 2.8v the backlight is much less bright). In case it's relevant the PWM signal is at 5V.

\$\endgroup\$
3
  • \$\begingroup\$ Have you looked at using a single PMOS and just tie the gate to the GPIO of your MCU? You’ll need lower-than-normal Vgsth on your PMOS, but far from unheard of. \$\endgroup\$
    – winny
    Commented Sep 8, 2020 at 9:00
  • \$\begingroup\$ I worked with Osram with regard to aircraft panels (real ones.) I'd like to suggest the thought that one LED and another LED cut from the exact same wafer will not necessarily look the same brightness or color when driven with the exact same current. I worked on developing equipment designed to "bin" their LED devices (2D matrix, CIE color perception wavelength and CIE perception brightness) into boxes that they could sell as a group to aircraft instrumentation manufacturers who, only then, could count on the fact that adjacent LEDs within a digit, or adjacent digits would look the same. \$\endgroup\$
    – jonk
    Commented Sep 8, 2020 at 10:55
  • 1
    \$\begingroup\$ If you don't mind slight variations, but noticeable, in your displays then don't worry about it. \$\endgroup\$
    – jonk
    Commented Sep 8, 2020 at 10:55

1 Answer 1

Reset to default
1
\$\begingroup\$

Your existing module, according to you, meets all your needs but the voltage is simply too high. Presumably these LEDs require a 2.8V input voltage that for some reason needs to be low-resistance. It is hard to speculate without knowing the equivalent circuit for the LEDs, but based on this requirement it sounds like you can rule out being able to use a simple resistive voltage divider here.

The best solution I can think of would be to either use a high power op-amp (which is unnecessarily complex here) or simply drive it with a FET. An FDN337N FET would do nicely.

schematic

simulate this circuit – Schematic created using CircuitLab

Alternatively if you want a slightly simpler setup with few unknowns you can also use a solid state relay, they can handle the current and can switch fast enough to handle PWM.

\$\endgroup\$
3
  • \$\begingroup\$ Yes unfortunately the characteristics of the module backlight LED(s) are rather poorly specified, only the nominal voltage is given. I'll setup a test driving with a FET, it sounds like it should work. \$\endgroup\$
    – James Turner
    Commented Sep 8, 2020 at 14:35
  • \$\begingroup\$ I also didn't realise solid-state-relays had switching performance sufficient for PWM; I've used them in other applications but assumed they were 'slow'. \$\endgroup\$
    – James Turner
    Commented Sep 8, 2020 at 14:36
  • \$\begingroup\$ @JamesTurner Well depends on how you define slow I suppose.. but they are more than sufficient to PWM a LED without observing any flicker. \$\endgroup\$
    – Jeffrey Phillips Freeman
    Commented Sep 8, 2020 at 14:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.