In Reply to: Circuit for constant power source posted by Rao on 01/24/00 at 6:29 PM:
: I am looking for a Circuit for constant pulse energy control switching modew power regulator using LM 317E with vco
: Pulse width of the Bulb voltage/current out out will change inversely with the prevailing battery
: voltage. Thus energy per pulse is constant. Reruired for a constant beam light intensity for a hand torch with 4 Zinc cardon or Penlite cells.
: Bulbs rated 3.2/3.8/4.8/6 with 5 to 10% tolerance.
You shouldn't have much of a problem. If you have equipment you should probably
pulse the bulb and examine the resistance decay and build up time constants. This is
typically a monotonic function of temperature. Purpose of the measurement is to set
the lower pulse rate so that you can ignore (but know) the resistance variation during one
cycle. In other words set the pulse repitition rate about 5-10 times the measured time
constants; always try to simplify or design to ignore things. Besides, if you allow the
peaks to overheat the filament the life will go down _dramatically_!!
Remember that the bulb resistance is higher when sarting and make sure that the power
dissapation during startup is accomidated. I can't remember why right now but a constant
(average)voltage is better than constant current drive. Assuming that you do that the lower
bulb voltages are easy to deal with using some form of; averaging (RC), comparing, and
switching (FET). The higher bulb voltages will require a switching regulator subsystem.
In any case you have to accomidate the internal compliance of the batteries this might
imply some energy averaging on the battery side (a capacitor). The fact that the bulb
resistance is constant (your pulsing much faster than the bulb can see) helps here. At
this point you should be able to write down the equations for the voltage and current
through each component of the circuit during a cycle and also under all operating
conditions; cold bulb, hot bulb, new battery, old battery, low/high ambient. As a part of
this you should buy several (preferably a couple hundread but ~20 will do) bulbs and
estimate the Probability Distribution Function of the hot and cold resistance. Just a CYA
that occasionally has surprises.
If you need to estimate the temperature there are techniques to estimate it visually based
upon color but that is probably too hard. To find out temperature/resistance variation
(that is to find out the temperature meaning of resistance readings taken earlier) just
measure the resistance in normal operating conditins and than change the ambient while
keeping everything normal. The filament temperature will go up and down with
ambient temperature on a one to one basis (good enough for your needs anyway). Be
aware that this is a linearization and if the bulb temperature changes by over 10-20degC
you would have to retake the measurement or extend the measurement range. Hopefully
this is just to allow you to _know_ the temperature profile during one pules cycle and isn't
really needed for the design.
Let me know if I have overlooked something or am incorrect.
I usually get one or more things wrong the first time through a design problem so
make sure that you understand things before you need to not after.