Soft starting high-power incans

I was thinking about the problems the USL folks have had with bulbs flashing on turnon and it occured to me that bulbs could be soft started with a large series mounted inductor. Is there some technical reason this isn't workable in situations where soft start is desirable but regulated power supplies aren't a viable option?

One word for you, enLIGHTenment:

sparkplug.

That's what you would turn your circuit into with a series inductor, because an inductor wants to keep the same current flowing through itself, and will adjust it's voltage accordingly, via the formula V=L di/dt. Thus a rapid change in current would mean a HUGE change in voltage, which would mean enough voltage to force a spark across the switch at turn off.

A capacitor in parallel is the simplest solution, circuit-wise, but it has to be freakin' HUGE to make a difference, because the "R" in the RC time constant is the small resistance of the cold filament.

The best "simple" solution (short of PWM regulation) is what Snoopy has done with MOSFETs and a separate RC circuit which uses an actual high value resistor, and thus doesn't need a large C. Check his thread out here in this forum by doing a search on his username.

You could also use a cap across the leads of the bulb too. This would also provide a discharge path for the cap when the light is turned off. I wonder what kind of an RC or RL time constant you would be shooting for.

I have read some discussion in the past on this topic and it was mentioned that there was a study done that determined that a soft start did not really affect bulb life. I am guessing that a solid power supply was used for this test. I think that one of the things that was overlooked was that most of us use batteries and when a switch is first thrown, the bulb sees the full voltage of the rested batteries (esentially a voltage spike). Once the bulb is on, the batteries tend to sag a bit and all is well. I am interested in hearing what some of the senior members have to say.

edit: above was being typed while js was responding...

So what time constant should one shoot for?

That and some folks are essentially "overclocking" their bulbs to start with....including letting their cells rest for a period off the charger, or even doing a short discharge, to avoid turning their spendy little hotwire into a flashbulb...

[ QUOTE ]
WhiteHot said:
You could also use a cap across the leads of the bulb too. This would also provide a discharge path for the cap when the light is turned off. I wonder what kind of an RC or RL time constant you would be shooting for.

I have read some discussion in the past on this topic and it was mentioned that there was a study done that determined that a soft start did not really affect bulb life. I am guessing that a solid power supply was used for this test. I think that one of the things that was overlooked was that most of us use batteries and when a switch is first thrown, the bulb sees the full voltage of the rested batteries (esentially a voltage spike). Once the bulb is on, the batteries tend to sag a bit and all is well. I am interested in hearing what some of the senior members have to say.

edit: above was being typed while js was responding...

So what time constant should one shoot for?

[/ QUOTE ]

I think anything more than 50 mS is enough, according to what I've been told.

I suspect that those studies of the soft-start impact were not done on highly over-driven, high-powered lamps run on batteries, but rather lower powered, longer life lamps run on voltage sources. But I'm just guessing.

In any case, it can't hurt, but there's no question that precisely controlling the effective average voltage at the filament is a far better solution than soft-start alone.

[ QUOTE ]
js said:
Thus a rapid change in current would mean a HUGE change in voltage, which would mean enough voltage to force a spark across the switch at turn off.

[/ QUOTE ]

Couldn't a shunt diode handle that? Even with something like the Sleeper, steady-state current over the inductor would be on the order of 30A. Power diodes capable of withstanding 600A surges are pretty widely available.

Again, I'm assuming there's a good reason no one has gone the inductor route. I'm just wondering what it is.

Snoopy's mod is the best I have seen so far. Even better than my idea /ubbthreads/images/graemlins/frown.gif ( Two 555 timers driving the mosfet )

Digikey has 100 amp mosfet 50 volt for under $3 TO 220 case I think. Off the cuff, .5 to 1 sec delay sounds good to me.

Les

enLIGHTenment,

Yes, a diode could be used to allow the current to free-wheel through and slowly fade off. It could be done.

But, ever check out the sizes of coils with any significant inductance? They aren't small. I suppose that's main reason, but I'm just guessing.

Hey, go for it and report back. I'd love to hear details and results.

/ubbthreads/images/graemlins/popcorn.gif

The issue here is that the bulb resistance changes (increases from almost zero) as the bulb warms up. So hitting the bulb with a constant voltage when it's cold gives it a huge and stressful surge of current. Adding something like a capacitor gives another place for current to go, making a time constant when combined with the battery's internal resistance, so the voltage hitting the bulb starts out a little bit lower. But with these ultra-high-current battery packs, the internal resistance is extremely low, and isn't always the same from pack to pack. Capacitors aren't the answer. The LVR or something like it is really the right way to do it. I haven't seen the LVR board, but such a circuit can be made very small (the MOSFET which is needed anyway, plus a few very tiny SMT parts) if desired.