Driving power FET from LV AVRs...

Go To Last Post
10 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,

I'm looking for ways to drive an IRF730B or similar from a low-voltage AVR - it must be very simply and cheap (its to drive a boost converter).

I was going to drive the gate directly from the PWM, with a series 200ohm resistor and a diode from the gate to ground (C to gate, A to GND). This would work fine for logic-level FETS, but I need to switch peaks of 250V DC at about 50kHz.

I'd prefer not to use an extra chip like the Maxim FET drivers, so was looking for ideas... The IRL630S looks a possibility... using an ATMega16L

Thanks

Nick

Nicko

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The IRL630S is only a 200V part. Choose a part with a higher rated voltage than the working voltage. Also, the gate capacitance is very high and the circuit will require a high current driver to get good rise and fall times. Fast rise and fall times increase the efficiency of the circuit and keeps FET resonably cool during operation. However, without controlling the slope of the edges, you can get increased harmonics and generate lots of EMI.

Most dedicated switching power supply controllers take all this into account so that you do not have to reinvent your own switch mode controller. Let us know how it works out.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Getting (reasonably priced) logic level parts much above 200V is a problem.

I have been modelling this in SwitcherCAD (LTspice), but really have no feel for the source and sink impedence of an AVR port (I'm rather new to AVRs). Any pointers to this would be appreciated. Once the drive to the gate is not "ideal" (i.e. not "stiff"), you see all sorts of artifacts on the gate, including some quite strong -ve pulses (due to switching transients and the FET's internal capacitances), hence the smal signal diode (1N4148 or similar) - whilst the AVR might not die immediately, it will in time ;-)

I actually only need 180V at 20mA, so the IRL630S is not so bad. with a 100uH inductor, my figures give a peak I(L) of about 1.1A at 50kHz with a 75% mark time, which means that in practice the IRL630S is fairly conservatively used.

I realise about the switching speed and the cleaness of the edges, which is why the impedence of the port is important. The Qg of the IRL630S is 40nC, which is not bad - the best non-logic FETs are about 30nC (only expensive exotice seem to be better), and at only 50kHz switching, I'm hoping the port bit will dump enough charge very very quickly. I'm aiming for about 85-90% efficiency - the same as the MAX1771 I'd normally use for this type of application.

Just as important is the use of a low ESR output cap and a very fast diode.

Nick

Nicko

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To get a general idea of output impedance, draw lines from Vcc to Voh and from GND to Vol. The slope of these lines should be a pretty good substitute for an "impedance" since CMOS I/Os look a lot like resistors until they reach their current limit..

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Gates of switching FETs can take amps during the transition. Even at 50KHz, I'd be concerned that a normal port pin would not drive it hard enough to get reasonable efficiency. You certainly won't get reasonable performance with a 200 ohm series resistor!

There are logic-level FETs with thresholds under 2.0V, but the combination of that with high Vds ratings may be pretty tough!

What system supply voltages do you have?

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You can generate a gate drive voltage from the avr by generating a square wave on one of the output pins and double the voltage with a couple of diodes and caps. That has worked well for me in past where a higher than Vcc voltage is needed. Then you could build a totem pole driver with discrete parts for driving the gate of the FET. There are plenty of discrete sot23 bipolar transistors that will handle an amp of pulsed collector current.

Another option may be to use a high voltage bipolar transistor. If you only need 20mA of collector current and not several amps, then a small bipolar part may be the way to go. It will operate at the voltages and currents that the avr is capable of outputting and the circuit is simple and should be reasonably priced.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmmm, sounds a lot like the NIXIE clock project I just finished up. If you are running from a mains supply you should have acces to the unregulated supply. (If batteries maybe not). A couple of transistors should be able to level shift the PWM out for the fet drive

BTW, I don't think you'll be able to approach the effeciency of the max1771. They use a few tricks to get the effeciency way up there.

-carl

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yeah, dedicated regulators are the way to go unless you are trying to do something quick and dirty for parts reduction where performance can suffer. Even then, the headaches of making it work are often not worth the effort. I have built some very 'clever' circuits that took a lot of time to tweek all because the parts count could not be increased. I have also attempted a several 'clever' circuits that did not pass the tests and had to be redesigned with the more traditional tried and true approches.

BTW, I remember when we used to throw away NIXIE displays in favor of the more modern and much cooler looking LED displays. Now I really wish I had some of those old displays.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm very familiar with several vendors' SMPS chips, in particular the MAx1771 (e.g. see one of my project pages: http://www.desmith.net/NMdS/Electronics/NixiePSU.html, I just wanted to do this myself with a single chip if possible. I suspect that I'll protect the AVR with a MAX5048 series FET driver - at about $0.75 it's a lot cheaper than a 1771 (typically $1.80/1k or $4.50 one-off) and requires fewer components. I can then use a straight IRF730AS or similar cheap power NFET for the switch. Very cheap and lower component count. Efficiency will be pretty good - I expect at least 80%, possibly more...

I have a very large quantity of nixies, and want to use them to best advantage...

Nick

Nicko

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a couple switcher situations where I am driving "logic" fets directly
from port pins. Seem to work just fine with nice brisk rise and fall times.

True, peak gate current is limited by the avr's port driver on resistances, but
speeds still seem more than adequate. If you are worried about currents
getting back into the port, just use a couple of small schottky diodes to
augment the avr's internal clamps.

If it is not a misprint, my catalog lists a IRFSL9N60A which purports to
be a 600 volt logic fet in a TO-262, but I would think a 200 volt device would
be fine, too.

Tom Pappano
Tulsa, Oklahoma

Tom Pappano
Tulsa, Oklahoma