"BUCK" voltage converter questions

Bean

I want to try my hand at makign a "buck" voltage converter to go from 12V (close to 24v open circuit) solar panels to 7.2v to run a robot.

The schematics I've seen use a high-side switch (FET), is there any reason I can't use a lo-side switch ? It seems the FET would be easier to drive if it was on the lo-side.

I want to have a propeller chip contol the switching frequency and duty cycle to get the most power. Is there a FET that will turn on at 3.3V ?

Bean.


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Does that byte of memory hold "A", 65, $41 or %01000001 ?
Yes it does...


·

Phil Pilgrim (PhiPi)

You could do it with a low-side switch, but you would lose the advantage of having a common ground between your power source (and switching controller) and load. Many buck converters still use an nMOSFET on the high side, but they have to use a charge pump to boost the gate voltage above the supply voltage. As to an nMOSFET switch that fully conducts at V_GS = 3.3V, try the IRF3708.

-Phil

Beau Schwabe

Bean,

"Six of one, half a dozen of the other..." Yes you could use a LOW side driver (an N-MOS), the reason that it's done the other way is to keep the Grounds common and it's much easier to read the output voltage. However as you stated, to ensure that the 3.3V signal will turn on the MOSFET into saturation can be tricky, even with some of the logic level mosfets available. In this since, it makes more since to use a HIGH side driver with a P-MOS so you have adequate drive strength. IOW - easier to take the Gate to GND with a P-MOS than to drive the gate to VSupply using an N-MOS.

I know that I have talked in circles, but it depends on the topology that you want to use... If it's ok to NOT use a common Ground and you can measure the output voltage with reference to Vsupply, then a N-MOS would work. If you need common Ground, then P-MOS is the way to go as well as measuring the output voltage with reference to Ground.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Phil Pilgrim (PhiPi)

Even with a pMOSFET, the output swing of its gate-driving circuit has to be to reach the full input supply voltage of the converter in order to shut it off. The simplest approach to this is to use a designated MOSFET driver, such as one of the Micrel devices. They're spec'd for driving capacitive loads, switch with logic level inputs, and can provide a full voltage swing from V_supply to ground.

-Phil

Bean

Does this look right ?

Bean




▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Does that byte of memory hold "A", 65, $41 or %01000001 ?
Yes it does...




Post Edited (Bean (Hitt Consulting)) : 11/20/2009 3:56:43 PM GMT

Phil Pilgrim (PhiPi)

No. The FET will shut off too slowly (via R2 and the gate capacitance), which will compromise efficiency. You really need to drive the gate low and high. Also, you swapped the gate and source connections.

-Phil

BrooksL

Also, C1 goes on the input side of Q1 (esp because solar panels have relatively high source impedance) and D1 should be Shcottky type (fast, low Vf). Your sense sampling rate will need to be much higher than your Q1 driving adjustment rate - and you'll need to be careful that your sense and control are in correct phase to prevent instability. Looks like fun to play with.
-- BrooksL

Phil Pilgrim (PhiPi)

Small OT diversion here (sorry Bean): BrooksL, is your last name Leffler by any chance, the KAP guy?

-Phil

Bean

Okay, here is my second try...



Bean




▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Does that byte of memory hold "A", 65, $41 or %01000001 ?
Yes it does...

Phil Pilgrim (PhiPi)

That could work. You have to be very careful with your programming to make sure both Q2 and Q4 don't conduct at the same time! To remove this possibility (and save a pin in the process), you could connect R4 (or maybe even the base of Q4) to the collector of Q3.

But you could easily replace all three transistors and four resistors with a cap and one of these:

····www.micrel.com/_PDF/mic5018.pdf

The nice thing about this part is that it includes a charge pump to drive the gate of a nMOSFET in a high-side configuration. However, if you still want to use a pMOSFET, there are other driver chips that don't even need the cap.

-Phil

Beau Schwabe

Caution when connecting ... "R4 (or maybe even the base of Q4) to the collector of Q3" ... you will create a current loop between the B-E junctions of both the NPN and PNP that will cause them both to conduct slightly.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Bean

Phil,
Thanks for the link. I assumed there HAD to be a part that did all the dirty work.

Bean.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Does that byte of memory hold "A", 65, $41 or %01000001 ?
Yes it does...


·

Phil Pilgrim (PhiPi)

Bean,

I've probably steered you wrong on the MIC5018. I just noticed that its supply voltage is limited to +9V max, and your panel probably puts out more than that. Instead, take a look at the MIC4416. This is a low-side driver, but it could be used on the high side with a pMOSFET, since the output voltage covers the full Vss - Vdd range.

-Phil

Philldapill

I know National Semiconductor makes some really good high-side drivers that will allow you to use an N-channel FET. I've ordered a few as samples and they work great! Give me a sec and I'll find the PN...

Phil Pilgrim (PhiPi)

Micrel also makes some high side drivers that have a wider input voltage range than the 5018, but they don't list them as recommended for DC-DC converters. I haven't bothered to investigate further, though. I suspect it may have something to do with switching frequency and the capacity of their charge pump to keep up.

-Phil