A bit of background first:
I'm trying to make a multipurpose module that will be capable of:
- Flashing two independent outputs at about 0.5 - 1 Hz
- PWM one output at a few kHz
- RS-485-control of two independent outputs. Then it could potentially be instructed to do anything - flash the outputs, continuous-on/off, or PWM. (The RS-485 will go through a transceiver chip, then to the PIC, which will then control the outputs per its programming.)


- The MOSFETs/gate driver must be able to handle anything from 12 to 48VDC.
- Peak current loading per FET will be <2A.
- Logic circuitry runs at 5V.

I'm looking to control a bunch of strings of LEDs, either a cluster of regular T 1-3/4, or else 12 high-brightness 1W LEDs.
Notes: The high-brightness 1W LEDs are driven by a constant current supply; these will not be PWM'd, but may be flashed.
The outputs are to be connected to two loads, which will have a common GND between them.

I've been looking at many different high-side N-channel MOSFETs, and how to go about driving them. Their allure is of course their low on-resistance, but driving the darn things on the high-side seems to have all kinds of challenges, and the fact that they will need to serve in continuous-on duty seems to put a damper on using a driver with a bootstrap type of arrangement, something like the LTC4440-5. They appear to require off-time in order to build up sufficient voltage at the gate output.

In the attached schematic, that shows a general view of what I'd like to do. The control signals will be coming from a PIC chip. V+ is the input voltage, which will be 12V or 48V, depending on what it's powering. NC1 and NO1 are the connection points for the load.
(Note that the "NO" and "NC" designations are to indicate the default states of the MOSFETs when the circuit is powered-up - the PIC will be programmed to turn on the "NC" MOSFET by default.)


So, given the requirements of PWM and continuous-on control, up to 48VDC input, coupled with the gate drive voltage requirements of high-side N-channel MOSFETs, am I just going to have to bite the bullet and use P-channel MOSFETs? (I think I'd still need some kind of driver chip in order to get acceptable turn-on/off times.)


Thank you in advance.
If any further details are required, I'll be glad to provide what I can. If I need to do some additional reading, which is a distinct possibility, please point me in the right direction.

 

Why don't you switch the ground side instead of the supply side?
That way you would not need a high-side driver.

 

Why don't you switch the ground side instead of the supply side?
That way you would not need a high-side driver.

Believe me, I'd love to. It would make this so much easier, in that it would be done already.

But that'd require changing a bunch of existing circuits that need to be able to attach to this, and that's not an option.

 

I want to see if I'm understanding some of this stuff correctly here.
LT1910 - it looks like this will accept 48V on the high side.
(Datasheet PDF)

It seems like the LT1910 could keep a MOSFET fully-enhanced indefinitely, as I don't see any graphs there that show decaying output vs time, or mention of maximum duty cycles.

And I see a turn-on time of up to 400uS when VGate = 32V, CGate = 1nF. Most of the FETs I've considered have a gate capacitance right around 1nF. So that wouldn't allow for terribly fast PWM speeds; hopefully fast enough to avoid visible flickering.
For example, the FDD86326 (Datasheet PDF):
- 37mΩ max rDS
- 80V VDS
- ±20 VGS (rated as a logic-level gate)
- 8A ID continuous


LT1910 + FDD86326? A feasible combination for this? Or is there a huge glaring issue staring me right in the face?


Edit, and an update on the "huge glaring issue":
The LT1910 is a no-go - the on-time is way too slow. I did some testing today with various PWM frequencies on an LED, just to get a feel for what kind of frequency I'd want to ensure a good look. I had to ramp it up to about 2000Hz to make any signs of flickering/strobing go away. And I'd prefer to push it considerably higher than that, just to be sure it would look good.
So, the on-time of 400uS wouldn't really work out well with high-frequency PWM, and it might look especially bad at lower duty cycles.

So, I ended up finally stumbling across this thread, which had a link to this handy PDF, which combined the bootstrap circuit with a charge pump. The bootstrap circuit turns on fast, and while it's discharging its capacitor, the charge pump is powering up. Result: Fast on-times, and indefinite run-time. I'm putting together a prototype circuit currently to give it a test-run. Hopefully all goes well.