Ideas For Bulb Flash Inhibition Circuit

[star882]

"Do you know how appropriate a thermistor would be for use in a flashlight? How long does the resistance remain at a high value after powering up? If it's on the order of seconds, or tens of seconds, that's too long for me. Are there thermistors available with time constants of around 100 milliseconds?"
100ms is about right.

 

[PaulW]

Star882,

I appreciate the answer, but I don't understand. Do you mean that there are thermistors with 100 msec time contants? Do you know who makes them?

Paul

 

[Jonathan]

Sorry that I've been away from this thread...I went back and looked at the datasheet that I linked above. I am quite certain that the 'time constant' value is the thermal time constant for the device temperature to respond to ambient air temperature changes, or similarly the time constant for the device to cool down in air when power is taken away. This is not the 'warm up' time when the device is turned on.

Think about it; these devices would be _useless_ for their intended task of limiting inrush current in power supplies if they took a minute to 'warm up'; could you imagine plugging your computer in and having to wait a minute for the green light to come on...windoze is bad enough /ubbthreads/images/graemlins/smile.gif

These devices have a mass of perhaps 1 gram, and when you first turn them on they will be resistive heaters generating several watts of heat. The will heat up _quite_ rapidly. Figuring out how rapidly is another, more difficult problem. You would need to know the 'heat capacity' of the device (essentially how many Joules or Calories it takes to raise the device's temperature), you would need to know the operating temperature of the device, and you would need a model of your power supply and load.

I believe that one could figure out the heat capacity from a combination of the dissipation factor and the time constant; imagine the device is an RC circuit. The analog of charge in the capacitor is heat in joules, the analog of conductance (inverse of resistance) in amps/volt is the dissipation in W/°C, and the analog of volts is temperature difference in °C.

I can't follow this up further at the present time, but it should give you some ideas.

-Jon

 

[PaulW]

Jon,

Yes, that does help. What you say makes sense. They will be fast, but it also makes sense that they would be a little slower than incandescent bulbs, which is what I want. The best thing to do now is to test. I'm going to just buy some and try them out.

Thanks,

Paul

 

[Tomas]

A thermister is simply a negative temp coefficient resister. As such it will be wasting energy, just like any resister, all the time it is in the circuit. It's only output is heat. They are possibly a poor choice for a battery-powered device.

What you are hoping to do is to extend the rise time (make the VA rise more slowly) than it normally does at the filament so here is a "soft" turn-on rather than the bulb-killing massive inrush until the filament heats.

While a thermister will do this, it's response time, and especially it's "reset" time (very slow, usually) would probably put it out of contention. It's constant power use while the light is on would also be a negative ...

 

[PaulW]

Tomas,

Yep, about 5% to 10% of the power of the lamp will be devoted to the thermistor in a design I'm exploring. That's a lot, but I'm losing it from only the viewpoint of efficiency and battery cost. It should allow me a higher steady-state current through the bulb and therefore more Lumens. The bulb efficiency (Lumens per watt) would also rise somewhat. I don't know how much.

But the reset time issue is a more serious problem. I hadn't thought about it. To take it into account, I would have to refrain from too quickly restarting the light after it's off. I wonder if I could do that?

Thanks for your thoughts. I'm back in the contemplation mode. /ubbthreads/images/graemlins/smile.gif

Paul

 

[Doug Owen]

My money's still on the transistor, cap, resistor deal. I fiddled around a bit with this just after I posted this suggestion, and it seems to have promise, but needs to be 'tuned' to the specific use.

I recall I got the best results using one of "MrAl's" Zetex NPNs (nice low Vsat) and the 'split resistor'. I was using a 6V half Amp bulb and my bench supply.

Doug Owen

 

[PaulW]

Doug,

I think you're probably right. It seems I have been avoiding the inevitable.

So, I did some searching and found that the Zetex NPNs (the ones from Digikey) can carry 1.5 amps at best. Alas, my beafiest application is rated at 9.6 volts and 3.16 amps. I'm currently overdriving that by 3% in voltage and would like to increase the overdrive to 10% or more. It's the high-power configurations that I want to push the most and hence need the circuit to permit a higher of overdrive.

Thus I need a larger transistor. I realize that I'll have to provide some moderate amount of heatsinking for the short transition time during which the transistor is operating linearly. Do you know, offhand, of any suitable transistors with higher current capacity?

Paul

 

[Doug Owen]

[ QUOTE ]
PaulW said:
Doug,

Thus I need a larger transistor. I realize that I'll have to provide some moderate amount of heatsinking for the short transition time during which the transistor is operating linearly. Do you know, offhand, of any suitable transistors with higher current capacity?

Paul

[/ QUOTE ]

Sure, there's a bunch. The best part in the end may easily turn out to be a FET, but for a start I'd go with a '2N3066 class" part. Probably in a TO-220 package. Jameco will sell you a MJE3055T for just over half a buck. Gains over 20 at 4 Amps. Couple that to a high gain driver (like 2N3904 or 2N2222) and pick up enough gain for our use.

The 'trick' is a low Vsat. Like I said, that may end up a MOSFET. But I'd start with bipolars and a couple of resistors and a cap. BTW, I forgot to mention, when I was fiddling with this I added a signal diode across the charging resistor (reverse biased) to aid in rapid resets (not that that may be an issue if the filament is already hot.

Doug Owen

 

[PaulW]

Doug,

That's very helpful. And the addition of the diode acroos the resistor is great. It looks like I'm going to have a little fun here. Thanks.

Paul