LED Lamp intermittently turning OFF when it should not.

[eem2am]

led lamp circuit

Hello,

Please could you examine the enclosed circuit which is a parallel LED driver with LED failure detection.

If any LED fails open then the lamp must be turned off by taking the LED driver’s (LT3517) shutdown pin low. (i.e.”one LED off, all LEDs off”)


***Whole schematic of lamp and LED failure circuitry ...........

http://i27.tinypic.com/35jlef7.jpg


***Close ups of Lamp with LED failure circuitry.....

***BJT’S

http://i30.tinypic.com/206ewi0.jpg

***LEDs

http://i25.tinypic.com/9kbddh.jpg

***LT3517 LED driver

http://i25.tinypic.com/2qw2c2h.jpg




The LED current in each LED is 166mA .
Input volts (“BATTERY”) is 13.5V

When any LED fails open………
..Q2 goes OFF
..which allows C5 to charge up
..which switches ON Q3
..Which takes the LT3517’s SHDN pin low
..which turns OFF the LED lamp

Unfortunately, every about 1 in 10 PCB’s with this circuit do not stay ON when switched ON. (even with all LEDs “healthy”)

-instead the LEDs flash for about 400ms and then the LEDs just go OFF and stay OFF.
-Then, when switched ON at another time, it maybe works OK.

So , what is happening is that this circuit is giving intermittent faults

-that is , it is sometimes switching itself OFF when the LEDs have NOT actually failed.

Please – do you agree with this here conclusion about why these intermittent faults are occurring with this circuit………….
..in other words, this may be why the lamp is being turned off even when NO LEDs have failed…………….

Q2 (NPN) parts have variable hfe’s –indeed the hfe will vary with temperature also.
-So sometimes, a particular Q2 on a particular PCB will not be switched on by the current through R9 (220K) and so will not manage to stop C5 from charging up and turning Q3 ON , -which turns off the PWM controller..

Do you agree with this analysis?

---------------------------
here is the .asc file for the free LTSpice simulator...

**broken link removed**

also

**broken link removed**

 

[FvM]

intermittent led circuit

There's no appropriate delay in the current monitor, it's triggering a few milliseconds after battery on, most likely before the LED current has settled.

 

[eem2am]

ltspice

Thanks FvM,

-Though the cap C5 and Resistor R12 (100K) provides the required delay which you speak of.....

..then Q3 does not turn ON unless Q2 turns OFF.

..i think the problem is that even with all LEDs "healthy" , Q2 is sometimes not able to turn ON and so C5 does not get discharged......and then turns ON Q3 which shuts the LED driver down.

 

[FvM]

led cluster lamp schematic

You may want to increase the delay to a higher ms range (don't know the actual buck converter startup delay) by setting C5 to several uF. Or short C5 during startup and check, if the circuit is operating then.

 

[eem2am]

led lamp schematic

FvM, I do thank you for looking in to this,

When I first came to this circuit I had to stare and stare at it long-long-time before I got my head in to it.
-but then I realised how simple it was.

If all LEDs are healthy, Q2 will short C5 and the LED driver does not get shut down.

However, I suspect that certain Q2 parts cannot turn ON with the low base current and so the LED driver gets switched off. ?

Do you second this may I please ask?

 

[FvM]

led lamp schematic

If the dimensioning shown in the schematic is correct, Q2 ase current isn't low. Q2 is fully saturated with Vce most likely below 100 mV.

 

[eem2am]

intermittent led

…….if you review the schematic, you can see that Q2 is ON when the LEDs are “healthy”.

-if one LED should fail open circuit, diode D7 or D8 will forward bias and then a small proportion of R9’s current will go through D7 or D8.
(instead of going into the base of Q2)

-It is intended that this tiny reduction in Q2’s base current will make Q2 turn OFF.
-so you can know that when Q2 is ON, it is only just ON.

….And what I am suggesting is that “only just” being ON, is not really good enough, because the lack of base current will mean that some parts from the batch won’t actually come on when they should….?

Added after 3 hours 17 minutes:

sorry i must revise my last, virtually all base current is diverted away from Q2's base when a LED fails.

The problem, as before, is why Q2 does not always come on when all LEDs ae healthy.

(also, the "battery " voltage needs a ~4u7 bypass cap to stop noise in the failure detection circuitry.)