Switch in series with LED load gicing problems

[treez]

Switch in series with LED load giving problems

Hello,

Ive just doen a 5W sepic led driver (vin = 5V, vout = 40V).
Its current regulated, but has an output overvoltage trip comparator.

When i test the ovevoltage comparator, i switch the load open with a series switch. i use the following switch.

**broken link removed**


..but when i switch on, and at randon other times, i am getting overvoltage faults even when the switch is closed, -is this switch not fit for purpose?, does it take time for its contacts to weld together, and in that time the overvoltage trip happens?

(as you know, overvoltage in current regulated led drivers happens when the load goes open circuit.)

 

[chuckey]

I would rate that switch at no more then 50V at .1A. if you are switching 40V @5W I = .125A, its all a bit close to its limit? if the switch is going open circuit the LED would go out? it might be that the trip is set too close to the actual voltage and a bit of noise pickup is enough to trigger the over voltage?
Frank

 

[treez]

if the switch is going open circuit the LED would go out?

....yes but i think the open is occurring over just a few 10's of milliseconds, so its enough to trigger the overvoltage comparator.....which shuts down the led driver.

Max led voltage = 40V
Overvoltage divider which feeds overvoltage comparator is set up for a 51V trip.

 

[chuckey]

Well, the first thing is , is the PSU going over voltage for a few 10s of milli secs?, try putting a big capacitor (100MF?) across its output, that should slow it down. I.e. soak up the excess voltage to reduce its amplitude and prolong the pulse length. - you might then be able to catch it on a DVM. if its noise pick up (or a noisy pot), then put a .1 MF cap across the bottom half of the voltage monitoring divider.
Frank

 

[dick_freebird]

It seems to me that you have no real data for us, only
conjecture. A 'scope shot would be worth all of the words
to date.

You could look at adding some time-domain filtering to the
overvoltage comparator to make it more resilient to things
like peak current overshoots. And look at using something
like a power MOSFET, properly driven, to get a clean open
circuit event with no contact bounce etc.