PWM and Quartz Halogen bulb life

[SherpaDoug]

I have a PWM circuit driving some quartz halogen bulbs. The bulbs are rated 1000 hours but they are only lasting 10 to 20 hours.
The bulbs are speced at 75W at 130V.
The driver supplies 152V at a maximum of 83% duty cycle at 75kHz.
85% of 152V is 126V so the bulbs should be fine.
The waveform has a little ringing but not bad. The switching FETs (IRFBA22N50A) are running barely warm to the touch.

What am I doing wrong?
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Neutone]

I'm no expert but would guess that the majority of the stress on the element is caused by changes in filiment temperature. Being that you are using a square wave to power the bulb the temperature I would guess has impulse changes that would cause the fractures in the filiment. See what happends if you use a choke coil to prevent the ringing. Also the resistance of the filiment is probably changing with temperature. Compare the current with the voltage on a scope. The filiments have a complex impedance.

[KerryW]

Try a 100uH inductor in series with the bulb. Be sure you use a snubbing diode around the inductor. The internal diode on a spare FET (IR HexFET) should work fine. Drain to + side, Gate and Source to - side. DO NOT try to use a 1N400x series diode, it must be a high speed switcher. A capacitor in parallel with the bulb would also be a good idea.

You can get the inductor at Radio Shack.

[david smith]

[mcafzap]

Years ago I used 2 techniques to extend lamp life. The obvious one is to reduce the voltage which is probably not possible for you, the second was to have a bleed resistor in parallel with the switching transistor. From memory, it was 100ohms 4 watt component for a 12v 5W lamp.


HTH
Steve

[SherpaDoug]

So far there seem to be two problems:
1) We've got some 110V bulbs mixed in with our 130V bulbs.
2) All 130V bulbs tested under bench conditions have lasted at least 24 hours, so we suspect some sort of physical shock or vibration problem in the field.

The bulbs are used on a ROV (wire guided remote control mini-submarine) and the customer has also bent the steel frame, so we are exploring shock mounting the lights & video camera rather than an electrical problem.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Tomi]

Hi Doug,
Your lamp is overdriven.
The effective voltage of an square wave is
Ueff = Umax * sqrt(k)
where k = t/T is the duty cycle
In your case
Ueff = 152 * sqrt(0,83) = 138.5V
If you want to make 130V you have to use
k = (Ueff/Umax)^2 = (130/152)^2 = 73%

Brg,
Tomi

[Matthew]

Doug,

I was thinking about you guys when I saw this topic then realized it was you. I spent about 2 years working for the father-company of the ROV you speak of.

Regards,
Matthew