LM7505 keeps shorting out...

[mnvelocitypilot]

Hi team!

I'm struggling to figure out why I keep frying a 5V regulator and a PIC controller on a very simple LED strobe system!

I designed two simple cards, one with a PIC controller, and the other with a linear current limiter (using an LM338) and a mosfet (IRF510A). The PIC just creates a flash sequence which drives the mosfet and flashes some Cree LEDs. My intention is to use this for both a flashing landing light and wingtip strobe system in my experimental aircraft.

Here's an instructable I published after getting the landing light to work:

https://www.instructables.com/id/LED-Landing-Light-for-Experimental-Aircraft/

Current limit is set at 5 amps, and the LEDs are flashed 5x with a pulse width of around 100ms.

The problem is... while my design is working just fine for the nose mounted landing light, it fails immediately when the same design is used for the wingtip strobes... on the first flash, the 5V regulator shorts, and the PIC controller gets fried!

I suspect the issue has got to be that the controller is mounted in the aircraft, while the strobes are out on the wingtips, and that the 15 feet or so of wire has enough inductance to create a high voltage spike after the 5 amp current pulse... However... I'd think that if that spike is there, it'd fry the mosfet, and not the 5V regulator/PIC controller...

So, I'm very confused. Any thoughts? HELP! This is making me crazy!

 

[esp1]

hi mn,
Is that a typo for a LM7805 5V regulator.?

What decoupling capacitors do you have 'around' the Vreg.?

Also consider a 1N4001 diode in parallel with the Vreg , input and output pins.

Cathode to the Input pin and Anode to the Output pin, this method is often used to protect the Vreg from high voltage positive pulses appearing on the regulated line.

E

 

[deepeshmishra]

Did you had used "mosfet driver" between your microcontroller and mosfet.
Friend Please post schematic so one can have deep think about your problem.

 

[mnvelocitypilot]

Thanks, esp1! Yes, it's a typo... should have said LM7805, but in actuality I'm using the small, 100ma version (L78L05). Regarding putting a diode in parallel, I'll try it - but thinking this through I'm not sure I understand the theory. The diode would conduct if the output rises to 13.8 + 0.7v, or 14.5V... the PIC microcontroller has an absolute max Vin of 6.5V... so it's fried. What I really need to do is to avoid the surge in the firstplace.

I'm thinking of trying a protection diode with the anode connected to the LED cathode, and the cathode connected to ground.... my idea is that it'd turn on and short the spike to ground.

Of course, all this is theory - I don't even know for sure if there is a spike! All I know for sure is that everything works perfectly on the bench, but fails immediately in the airplane. Geez!

BTW, for those that are curious, here's the schematic:





Thanks, deepeshmishra - I just tried to upload a schematic, hopefully it works. It's also available on the instructable (link provided in my original post).

THANK YOU! Much appreciate any and all suggestions!

 

[D.A.(Tony)Stewart]

This design although good on the surface is a little Mickey Mouse-ish when they neglect ESR of all components. Although you were correct to assume there would be ESL from the wiring, a 1A reverse diode across the strings will protect them.

Starting with the battery, estimate the ESR of all connections to the LEDs.

Load battery with 1A then 5A to calculate dv/di or ESR, where di in this example is 4A, choose any LOad resistor avail. between 1A and max load. (not no load or float level) usually 3.6V /per LiPo cell +/-.5V x N cells in series is the full range and +/-0.3V for 80% of charge usage under 5C load.

Same with LED's, although I know from experience a good 3W CREE LED is around 0.5 Ohm ESR.
The voltage drops with <rising> junction temp. so the cold room temp (25'C) ESR is around dV/dI = 0.4V/0.75A. from 100 mA to 750mA.

Then an array of 5A on some unknown LiPo battery voltage is possibly 6 in parallel with 3 or more in series or S3P6 giving a net ESR of 3*0.5/6 = 0.25 Ohms. So your battery and switch ESR goal is 1% of this or 2.5 mOhm.. The problem is most use LDO's for CC sources with 2.5V drop min and often choose more drop with fresh batteries. This is inefficient and HOT.

The biggest problem starts with 500 mOhm RdsON of the IRF 510. So get one with a few mOhm, then you wont need a heatsink , but you might need to learn to use solder paste on an alum. substrate board. USe the digikey search engine to filter RdsON , stock on hand and find the best fit for you. $2 1 mOhm FET looks like this http://www.irf.com/product-info/datasheets/data/irfh8202pbf.pdf

My trick is to permit lower current when the battery voltage drops below 30% state of charge thus only using load regulation of batteries (ESR) to and ESR of LEDs to regulate current within 30% which is barely visible change to the eye . ( unless modulated)

YOu can go for shorter durations and higher currents like 1.5 Amp easily and simulate the strobes used on actual aircraft. But more ESR calculations and voltage range and string size optimization needed. YOu need to specify LED array SxPy and battery voltage ESR capacity for me to optimize your design.

In my yard case I use 50ft of 18 AWG speaker wire to act as a 1-3 V drop around my yard lights with a 19.v laptop charger (CV) and 4 Cree string. with no active current source, just controlled distribute wire loss. I have 14V SMT strings on the rooftop antenna mast, and 4x3W strings on Al-clad boards around the garden flowers in the back. and then 9~15 rated SMT string dimly lit in the corner of the yard tool shed roof. running at 12V

The trick I used in my profile photo is use the 3 pin universal laptop chargers to dim the voltage with an external pot. (custom SMPS CV with no current limiter)

So in short,

- watch out for kickback on 5V regulator out to input and add reverse diode. (They will short out otherwise)
- get a much better MOSFET in the low mOHm range
- consider voltage swing from 100 to 20% state of charge
- allow LEDs to pulse shorter and harder like a xenon tube and add low ESR caps all around with ferrite to reduce stray inductive coupling and servo interference. (Use twisted pairs on all high pulsed power LED interconnects)

 

[mnvelocitypilot]

Wow, thanks to SunnySkyguy! Great response.

I guess if the design is a little "mickey mouse-ish" due to lack of consideration of ESR, that's my bad - the design is mine. And yes, believe it or not, I did spend some time looking at alternative mosfets on Digikey, but because I'm pretty sure others will follow my footsteps (on building the landing light configuration), and I figured that some of my less electrically inclined aircraft homebuilding buddies wouldn't be comfortable with surface mount components. PTH is complex enough for some. So, given that, and given that the IRF510 is available at radio trash, it was an easy choice.

Regarding heat, outside of the LEDs themselves, the hottest component isn't the mosfet, it's the LM338 linear regulator. If I ever get the time, I'll start looking at designing a switcher. I'd like to do that anyway, since that'd allow for much higher drive currents, but I've been reluctant because of the electrical noise they can generate (and the potential for impact to aircraft radio and intercom systems).

I chose around 100ms for the pulse duration because in experimenting, it seemed to me to be a good comprimise between apparent brightness and power/heat.

I'll try a 1A reverse "protection" diode across the LED string and see if that helps. It's getting painful replacing all the PIC controllers and the L78L05 regulators... I suppose I should just use a socket!

Anyway, thanks. Any and all input is much appreciated! I guess I still don't fully understand why the L78L05 regulator and the PIC controller are the components being blown, but if there's a simple solution I'm all in. I'd still appreciate any thoughts on the true root cause.

 

[D.A.(Tony)Stewart]

Linear regulators only source current and will fry with very small reverse current of output voltage greater than the input. Then they short out. If they don't have OVP and /or a reverse diode clamp.

I guess with a 7V drop with a few watts of heat from 12 to 5 you are wasting too much heat, which you could be using for red power LEDs in series just above the dropout.

I would strongly suggest you buy the OKI/MUrata 3 terminal SMPS if you only need 1A for 5V.

Make a professional drawing package and BOM at Digikey and they will ship all parts in 1 day from your BOM and save yourself some money from RadShack, except for maybe 5~10 in shipping. Buy enough to make a dozen kits and get your friends to pay for yours. You got a plane worth a few pesos., its worth it to get better parts.

THis MOSFET is good for 3 mohm (90A) for <$4 https://www.digikey.com/product-detail/en/IRFP3206PBF/IRFP3206PBF-ND/1894149.. Diodes Inc (nee Zetex) makes better ones for a couple dollars more.


For SMPS, don't design, buy these MUrata/OKI 3 terminal regulators 5V@8W $4
https://www.digikey.com/product-detail/en/OKI-78SR-5/1.5-W36H-C/811-2692-ND/3438675 you can connect them with 3 pin header cable from Digiikey too and hot glue any place or a tiny dot of polyurethane adhesive.

For root cause, the LDO's will fry from overvoltage kickback spikes or induced transients. They have very high impedance above 5V since they only source and not sink current. THus lack reverse diode protection on LDO with <0.1us inductive spikes might have caused it. The LED's can handle -5V each, but the diode with the least reverse capacitance will shunt the most reverse voltage, so the best protection is ESD protected LEDs but adequate (for some) is one reverse diode to protect LEDs in future. 10uA is tolerable at -5V whereas 50uA at -10V ( or +22V kickback will blow one of them if unusually low capacitance , so you were lucky. THis would be a double mode failure, not your root cause.

The workhorse 7805 (not 7505) LDO has OCP and OTP but not OVP. Somehow a microsecond or longer spike above 12V got into the 5V out, maybe thru your MCU loads elsewhere. So following the design guide and using a reverse protection diode or better a TVS clamp diode for OVP (eg 5.5 or 6V) would have saved both LDO and MCU.

If you ever need 5mm LED's I get them in bulk ( several kg) The White are 20,000 mcd 5000K 30 deg, also 14,000 mcd in both yellow and amber and some red. (500/bag min.) They are really easy to hook up with 30 AWG magnet wire, a few turns and solder right thru the varnish. in <4 seconds (never 10)

For aircraft EMI protection you should try to use CM ferrite beads sized for the wire in bulk to reduce interference with radio. any current pulses can be picked up on AM or SW between radio stations, if you aren't sure or don,t have a scope. It's also possible your transmitter or servo can be injecting many volts into the parallel wiring.

BTW nice plane. A buddy of mine could sell his vintage Cesna fully restored for $17K so he donated it to a women's flying club. I suspect you got about the same investment or more in yours.

BTW a small zenon flash bulb is not that hard to make and has as much efficacy as an LED, but also much more shielding, filtering protection needed and HV supply is the issue for efficiency.

I've seen xenon flash bulbs wipe out cell phone reception inside a building at 1 pps. (Drives the phone's AGC crazy when the current resonates during avalanche discharge)

 

[mnvelocitypilot]

Wow, once again, SunnySkyguy rules!

Just background - I'm driving the LEDs at 5 amps - two separate LM338/IRF510 power boards each driving 5A, driven by a common PIC controller... one LED string for each wing (8 LEDs/string in parallel).

Regarding SMPS - the Murata/Oki 78SR series looks really cool - but there are two things... first, with an integrated solution like this, I'm not sure how to configure it as a current limiter as opposed to it's design purpose as a voltage regulator. How would I control the current?

Second, they max out at 1.5 amps... I'd like to see 5 amps minimum, more if I can keep the LED heat under control. Can I put 'em in parallel? What would be the considerations?

The mosfet family you suggest also looks awesome (IRFP3206 lead free). Great suggestion!

BTW, thanks for the comment on the plane. It's a labor of love... we've got about 1100 - 1200 hours on it now... flown it all over the place. Here's a short video of one of our trips last year: https://www.youtube.com/watch?v=X8jpKQFkxxo

BTW... I'm doing all this to replace a Whelen strobe system... the zenon flash power supply failed ($500+ to replace)... PCB is burned through, several components obliterated. I'd never replace it ... new LED strobe systems are out there for purchase at about the same price. I'm just hard headed enough to DIY (after all, I did build the airplane!).

THANKS AGAIN!

 

[D.A.(Tony)Stewart]

Supply voltage range? Max/min
LED voltage Max/min

THe OKI 3 terminal 5V reg was intended for uC, but if the 'L05 works ok, no bother.

 

[mnvelocitypilot]

(sorry if this gets posted twice!)

Supply voltage: 12 - 14V (13.8V nominal)
LED (XMLBWT-00-0000-000LT60E3) voltages:
Vr: 5V
Vf: 2.9 - 3.5V (3.1V @ 1.5A)

There are two LED assemblies, 8 LEDs per assembly
Each LED assembly is driven by it's own LM338/IRF510 "Power" card. Both power cards are strobed by one PIC12F1501 "Controller" card.
There is approximately 14 feet of 18 gauge shielded belden cable connecting each wingtip LED assembly to it's power card.

Right now, the LM338 drives 5 amps to each LED assembly, resulting in (ideally) 625ma/LED. I'd like that to be more like 1A/LED, 8A total. Obviously, that is much more heat... which is why a switching supply makes sense.

Any & all ideas are very welcome - THANKS!

 

[mnvelocitypilot]

Finally had a chance to try a modification again today... I put IN4005 protection diodes across the LED string (on the power boards, between the LM338 and the IRF510, cathode to the LM338 side), and I also put a IN918 protection diode across the L78L05 regulator. Once again, it tested fine on the bench, and once again it failed immediately in the airplane. I'm starting to think that this may not be an inductive field collapse creating a reverse power spike, but something else. VERY FRUSTRATING!... I had hoped to have this new system working for the Oshkosh airshow next week. Oh well...