Strobing 5 amps down a long wire... and frying the supply

[mnvelocitypilot]

Hi team - I've been over in the electronic elementary questions forum trying to get some ideas on a problem I'm having where a homebrewed strobing 5A LED power supply keeps frying.

This is an LED strobe system I designed for use in my experimental aircraft. It's working fine in one application - as used to control a homebrewed LED landing light (which I can set to flash for better air to air anticollision lighting), but fails when applied to drive wingtip mounted LED strobes.

Here's the thread I've had running... several folks have made suggestions, but nothing is working:

https://www.edaboard.com/threads/319335/

To net this out - I designed two cards... a controller card, containing a PIC controller, the L78L05 regulator that drives the controller, and a 2N2222 transistor to uplevel the strobing output of the PIC to a higher voltage (to drive the mosfet switches).

The second card (which I call the power card), contains a LM338 used as a current limiter (set to 5A), and the mosfet (IRF510) used to strobe the Cree LEDs. The only difference between the nose mounted landing light system and the wingtip strobe system is that in the wingtip system I'm using one controller card to drive two separate power cards (one for each wingtip strobe).

The wingtip system works fine on the bench, but as installed in the aircraft, it immediately fails - the L78L05 regulator fries (shorts), which takes out the PIC controller. It's dead immediately after hitting the switch.

My theory has been that I'm getting a big reverse voltage spike in the aircraft installed system due to the approx. 14 feet of wire between the controller card + power card "power supply", and the LED. I've tried putting a protection diode (IN4005) across the LED line at the power supply end, and also using a IN918 across the L78L05 voltage regulator. It still fries.

The wire running to the LED is high quality belden, 3 conductor in a shielded wire.

I covered the specifics of the controller and power cards in an instructable:
https://www.instructables.com/howto/mnvelocitypilot/

I'd sure appreciate any ideas on how to make this work... very frustrating!

Thanks!

 

[esp1]

hi mv,
You do know a 78L05 has a maximum current output of only 100mA.?

Have got a 7805 1Amp Vreg you could try.?

E

 

[mnvelocitypilot]

Hi esp1 - yes, thanks, I actually chose the L78L05 because the current drive of an LM7805 was far more than the PIC controller required. The L78L05 is only used to drive the PIC12F1501 controller, which never gets close to 100MA of current requirement. I know this not just because of the spec - but also because of about 100 hours of bench testing prior to installation in the airplane, and also because of the operational results as used for a strobing landing light where it's working fine.

I don't know for certain why the system is frying... My thought regarding a reverse voltage spike causing a ground bounce, which in turn frys the weakest link (in this case the L78L05 & PIC controller) - is just a theory. I don't have the equipment to actually capture a scope trace.... so I can't be certain...

THANKS though - any / all ideas are welcome! Perhaps I'll try a 7805 just to be sure!

Dave

 

[esp1]

hi mn,
If you have a circuit diagram you could post, perhaps we can see a possible cause for the problem.
Eric

 

[mrinalmani]

Does the aircraft have a metallic body? If yes, connect the ground with the metal frame.
Have, you considered static charge build up (order of KV) in the frame, due to aerodynamic drag?

 

[mnvelocitypilot]

To esp1 - THANKS! Here's the schematic. I'm using a different Cree LED pn (a 3V device instead of the original 9V device), and an L78L05 instead of the LM7805 regulator, but the rest of the circuit is unchanged. It's pretty simple... the PIC controller strobes the 2N2222, which either pulls down the IRF1501 gate (off), or lets it get pulled up (to it's on state). The LM338 is configured as a current regulator. Pretty straight forward, and (as mentioned), it's working fine in the landing light (and on the bench).

To mrinalmani - THANKS! The airplane is all fiberglass... so ground wires have to be run separately (picture attached).

 

[D.A.(Tony)Stewart]

All CMOS is prone to the SCR latchup effect due to high current pulses coupled to stray wires connected to CMOS. There are internal ESD protection diodes rated for 15mA on every input, but this is obviously not enough.

Protection from these noise impulse events may include:

Reduce slew rate of current pulses (where V=L dI/dt )
Add filters to all I/O ports on long wires.
Shield all long wires.
Avoid high current returns on common signal paths ( DC ground shift)
TVS diodeson every I/O port( with long wires)

the other approach is to enclosed the power source and. return so high current is only a small loop between LEDs and. impulse power, while charge current is steadyDC andtrigger is immune to noise. ultracaps could be used with SCR's

 

[mnvelocitypilot]

mrinalmani - sorry, forgot to answer your other question. The strobe fails immediately upon power on ... in the hangar ... at airspeed zero! Static discharge isn't the issue... As I've said, I believe the issue is that I'm getting a reverse voltage spike after driving the approx. 14 foot long cable to about 3V / 5A - and then shutting it down with essentially a step-function waveform. Again, this is just my theory... I've got no facts to back it up, but it seems pretty plausible.

- - - Updated - - -

Thanks, SunnySkyguy - much appreciate your input. Space is very (very) limited in the wingtip, but I'll take a look at your idea of using supercaps to store the energy by the LED, and a more efficient mosfet to do the switching. I've never used a supercap, but if I can find ones that have enough capacity (and are small enough), that might just work.

Are there any issues with supercaps used in a configuration where there is a very fast repetative discharge rate?

 

[esp1]

hi,
Looking at your circuit I would say that the 220K from RA2 to the 2N2222 Base is far too high a value, I would use a 470R. I appreciate thats not the cause of the original problem

Regarding the 78L05 regulator I would add a 10uF across the 330nF and 100nF caps.

The fact that the PIC is not being destroyed suggests that regulated output of the 78L05 is not fault path.

It could be being damaged by a problem on Vin, perhaps a HV spike when the IRF1501 switches Off.

E

 

[rfmw]

I would add output-to-input diodes to both regulators, as described in most of the regulator datasheets. One idea that comes to mind is that your long wires and capacitance make some sort of a resonance LC effect, which you could damp by a low and series resistance at the wire start and end.