Boost Converter Schematic Comments?

[William At MyBlueRoom]

I'm working on a PIC (16F873A) boost converter for running LED strings. The converter in theory should provide approx 15v @ 300ma to 24v @ 190ma

I've used the excel calculator provided in the zip file on this project page to make Nixie drivers.

**broken link removed**

I would like to avoid using the TC4427a and use a resistor & zener instead as a clamp (I've seen similar used on a AVR version of a boost converter)

I do have a question on the math, the Duty cycle for max power from the inductor is 21us and the Period is 28us. I assume this means the inductor is charged for 21us and the discharge only take 7us. The discharge seems really short to me or do I have my charge/discharge reversed? Duty when on should make the HEXFET conduct...

 

[justDIY]

is a fet driver really needed, given you can commonly buy (or sample) fets with very respectable switching times given a Vgs = 2.4 or even 1.8v? With such a low gate threshold voltage, that makes the 5v from the pic all that much more powerful?

also, isn't the output of the PIC push-pull (drives high and low) ... it's not like the output stage is an P channel open collector which needs a pull-down to go low?

I've never heard of a "flyback" surge from a fet, I wonder what the author of that instrucables means by this?

sorry to ask more questions than provide answers

 

[William At MyBlueRoom]

HEXFETs are the common way to do this, same with the schottky diode. The maximum voltage I need is about 25v and the minimum about 11v. It seems the very low voltage drop on the HEXFET makes the circuit more efficent than using a power transistor. It should also run cooler than a power transisitor.

Thats the great thing about these forums, sharing knowledge. I've learned plenty here.

 

[justDIY]

hexfet is just a brandname, trademarked by IRF ... like Fairchilds' UltraFET and TrenchFet

Now something like a FETKEY (also IRF trademark (maybe expired)) contains a 'hexfet' brand power mosfet and a schottky diode, keeping the distance between the switch and the rectifier to an absolute minimum.

I dabbled using a pic as a smps controller, but switched to dedicated purpose chips instead (with integerated switches!).

see here: http://projects.dimension-x.net/archives/57
and here: **broken link removed**

 

[William At MyBlueRoom]

JustDIY neat article, did you read the app note from microchip on PWM?

It's not so simple as just tossing a square wave at a fet & inductor. You need to know both the impedance and power the inductor is rated for. This determines the peak power, frequency and duty cycle for a given inductor. You also need a fast diode schottky preferred. I didn't notice any feedback system in your diagram. How do you control the voltage?

I've attached the excel spreadsheet from the article in the earlier message. Just plug in your values and he computes period & duty cycle & max power.

 

[justDIY]

feedback was described in the ramblings, but never drawn in the schematic. I used a small value series resistor inline with the load (LEDs) to sense current, and adjust the pwm accordingly.

I agree some serious math and fine tuning would improve things, but it's not "required" to make things work ... just tossing a square wave at the switcher works, as evidenced by my glowing string of 10 LEDs glowing brightly.

Having learned a great deal about smps between my first dabblings with that pic micro and now, there are a number of things I would do differently. The biggest change would involve using something a bit shorter than an 18" piece of wire to connect the pwm output from my pic to the gate of the switch... the shorter the traces for everything involved with the switcher, the better as I understand it now.

 

[William At MyBlueRoom]

Here's an application note from IR about gate clamping with a zener