Capacitor Type question

[dannix]

That makes so much sense now!

My reasoning for the cap was that once charged up, and the mosfet on, when there is no pulse (the gap) the capacitor would continue to feed the mosfet, keeping it out of the Linear Region until the next pulse which would top up the cap and so on.

I can see how the cap would delay the first transition to ON but would it not keep it on after that?

 

[dknguyen]

I can see how the cap would delay the first transition to ON but would it not keep it on after that?

If the PWM duty cycle from the fan was high enough to maintain the capacitor voltage it would always stay on (least likely). If it was low enough it would always stay off (least likely). If it was anywhere in between it would be dissipating a lot of heat (most likely). So even if you assumed it would never be in the most likely situation, you lose the ability of variable heating. The heater is either always on or off. No pulsing for efficient, variable heat. You also lose predicatability.

You seem to be forgetting that it's much more likely that the PWM won't be sufficient to maintain the capacitor voltage to keep the gate on. It's much more likely it will slip into the linear region and stay there most of the time. WIth regular switching there is no risk of that happening- it may enter the linear region more, but it spends far less time there, and it's predictable.

 

[dannix]

I wont be having pulsed variable heating, the heater will be either on or off, depending on the fan being on or off, the pulsing is just from the fan signal. I dont want the heater on if the fan has failed or is rota locked.

If we can move away from this fan / heater issue as you say the current won't be enough anyway for it to work, the cap can only go upto what supplys it, which is a peak pulse, if for argument sake the pulse was enough to turn the mosfet on then a capacitor inline would be a good thing - right?

It would keep the mosfet on during the no pulse - or slow its transition to off thus disapating more power and out weighing the benifit. Is it only a bad use if the supply is not enough?

dam electronics, goes deeper and deeper!

 

[dknguyen]

I see. So then since you are doing high frequency MOSFET switching, no current driver is needed since fast rise times are not needed since switching is infrequent. You may need a driver, however, if the voltage that switches the MOSFET is not sufficient since a driver lets interface the two. However, in this case (without current requirements) you could use a level converter or whatever else...

Use a 555 timer pulse stretcher circuit (and supply it such that the output voltage is sufficient to drive the MOSFET, but make sure it can also accept the control signal voltages from the fan as well). Have it stretch out pulses it detects, no matter how short, so the period of the output pulse is longer than that of fan's output period. That way as long as their are pulses the output signal will stay high to drive the gate.

There are a lot of unpredictable problems if you try and use a capacitor for this. I'll briefly list some
-delay time turning on
-delay time turning off (like when the fan has stalled)
-frequency dependent (frequency too low, means larger capacitor means increased delay times for the above two points).
-dependent on the average value of the fan's minimum duty cycle
-probably needs a driver of some kind for increased voltage since the fan pulse minimum duty cycle probably has an average voltage far too low to drive the MOSFET.
-because of the point above, analog methods probably aren't going to work, you're better off trying to detect the presence of a pulse and stretching it out at a sufficient drive voltage with a "watchdog timer" of sorts, like the 555 timer method. I can't say what this will do to power consumption.

 

[dannix]

Thanks for all your help.Im going to have tpo look into this a bit more but this thread is not the place now.