# Battery Step Down Buck Converter - MOSFET problem

Discussion in 'Electronic Design' started by James A, Aug 26, 2003.

1. ### James AGuest

I want to use a buck converter to set-down a 4.8V battery to 3.5V
- Using the PIC pwm pin to drive the gate of the mosfet.

I've performed all the necessary calculation (L, C, Frequency, Duty
Cycle ect).

I was just looking into mosfets to use as the switch when i realised
that I am not going to be able to do it.

THe mosfet I found was a zemtech ZXm61N02F, the threshold voltage Vgs
= 0.7 V

So in other words if i want to be able to switch the MOSFET on/off I
will need to be able to Supply (4.8 + 0.7) = 5.5V to the gate in order
to do so. Since the PIC is powered at 5V this is not possible.

I was wondering if there is some way I can connect up the mosfet
switch so that i can perform this operation whilst maintaing the high
effeciency of the buck converter.

I have googled around without success.

I'm sure this is a fairly common application so I must be overlooking
something.

James A

2. ### Andrew PauleGuest

maxim 62x chips - (I think the 626 would be a good fit, but check the
data sheet)

Andrew

3. ### Fritz SchlunderGuest

The normal method to do this is to use a P-channel MOSFET instead of an
N-channel device. Something perhaps like the IRLMS6802 might be a good
choice since it is not all that different in packaging and threshold voltage
from your zemtech part. Datasheet available at:

http://www.irf.com/product-info/datasheets/data/irlms6802.pdf

Something to keep in mind when building switch mode power supplies run off
of batteries. The battery voltage decays when the batteries get low.
Assuming your MOSFET gate drive voltage is related to the battery voltage
(which it usually is and in this case it will be), the gate drive decays
sometimes enough to cause the MOSFET dissipation to become relatively high.
If the battery voltage is too low to sufficiently enhance the MOSFET, the
device will operate in the linear region and dissipate extra power.
Sometimes the battery is capable of producing enough power to overheat the
MOSFET while not being low enough to shut down your circuit. Therefore it
is often good practice to include some sort of undervoltage lockout for your
battery powered SMPS control circuitry.

Good luck making your whole circuit very worthwhile. Buck regulator
efficiency can range over quite a bit for different applications, but 80%
isn't that all uncommon. That corresponds to only a few points of gained
overall efficiency compared to a pure linear regulator approach for your
likely need to use a synchronous rectifier due to your low voltage output.

4. ### R.LeggGuest

If there's any possibility that the load will actually be damaged by
voltages greater than 3V5, using a PIC may not be such a good idea,
unless it is the only PIC function, and is suitably mated with
circuitry to ensure safe controlled output during start-up, shut-down
and even under 'normal' conditions of regulation.

Programmable IC's really aren't designed for settings where this is
supposed to matter.

The oscillator, output and UVLO circuitry of a pwm controller are
usually designed to provide predictable operation under these
conditions, and will be specified to do so.

I see no information re your semiconductor or it's manufacturer - I
assume a 20V part with a suitable Rds and non-logic-level gate
thresholds. Perhaps a P-channel device with logic-level rated
thresholds would suit the purpose better.

As is, you'd have to first generate a suitable drive supply voltage,
then bootstrap it and level shift the drive signal to the bootstrapped
rail.

Messy.

RL

5. ### John LarkinGuest

Note that the gate threshold voltage is where the fet just *begins* to
conduct, at some small drain current. One typically needs a lot more
gate drive - 5 to 10 or so volts Vgs - for serious turnon.

John

6. ### Dana RaymondGuest

You've run into the classical high side driver problem. Dedicated switching
supply ICs have internal circuitry to generate the higher than VCC gate
voltage required.

Of course, there are off the shelf chips that will do the job for you. Buck
voltage converters, high side driver chips, etc. Build? or Buy? is the
eternal question. Here's some suggestions relating to 'Build':

Is the supply driving the 5V regulator available, and is it of sufficient
voltage for FET saturation?
If so use it for gate drive and control it from the PIC using either an OD
output pin (capable of higher voltage operation), or use a 2N7002 fet to
control gate drive. The latter approach is slightly more costly, but it
allows for the gate drive to be disabled while the PIC pin is in tristate
during power-up (with a resistor to gnd before the 2N7002).

If a supply > VCC is not available, then make one. Use a logic driver driven
by the PIC's OSC OUT, or by a square wave output from the a PIC pin,
connected to a voltage doubler (consisting of a few small caps and signal
diodes) to generate the gate drive. Since gate drive current is likely to be
infinitesimal, then this approach is quite viable.

Hope this helps.
Dana Frank Raymond

7. ### Dana RaymondGuest

Actually, a resistor to gnd won't do what I said it would. But you can using
a different configuration.
Sorry for the confusion!

8. ### James AGuest

Yes, using a PMOS makes alot of sense.

I've selected the PMOS I want to use: IRF9530N
http://www.irf.com/product-info/datasheets/data/irf9530n.pdf

Things that attracted me to this chip:
- Cost
- Fast Switching
- Low Rds(on)

I will hook it up in this fashion:

* Connect Source of MOSFET to +4.8V
* Connect Drain to the LED.
* Connect other end of LED to ground.
* Connect one end of a 100 ohm resistor (R1) to +4.8V.
* Connect other end of resistor R1 to MOSFET gate.
* Connect Gate of MOSFET to PIC pwm Pin

I've been looking at the datasheet and I'm extremely confused.

For example:

Electrical Characteristcs:
[VGS(th)] [Gate Threshold Voltage][-2.0 ––– -4.0 V] [VDS = VGS, ID =
-250µA]

I suppose this means that the threshold voltage is 2vmin and 4v max
when Vds = Vgs

For my circuit:

At start up, Vgs = Vgate - Vsource = 4.8 - 4.8 = 0V
(since gate is tied to source)
(this means gate is off)

Then when the PIC grounds the gate, VGS = 0 - 4.8 = -4.8V
(is the PIC capable of doing this?)

I'm not able to work out from the datasheet if this threshold is
enough to turn
the gate completely on and if it will charge the capacitance fast
enough for this purpose where f = 100Khz for the buck converter.

Also, another thing I'm a little confused about is Rds(on). Sometimes
on the datasheet it is more than 0.2. Why does the datasheet say max
0.2 and yet sometimes it is more.

Is there a better MOSFET configuration that I can use? Like possible I
need to use a NMOS digital FET to drive the PMOS.

Note: I have to use the PIC as I'm doing numerous other things in this
circuit.

Thanks for the help.

9. ### Winfield HillGuest

James A wrote...
- HUGE ugly old package
- specifies 10V to switch (barely turns on at 4.5V)
- high gate charge
- obsolescent

You need a "logic-level" switching FET, preferrably with
Rds(on) specified for Vgs at -4.5V or less.

The elegant irLMS6802 has already been suggested to you,
here are a few more choices taken from my notes,

| manuf p/n Vdss, Id - pulsed, Rds(on) at Vgs package
|
| IRF irLMS6802 20V, 4.5 - 45A, 0.10 ohms at -2.5V Micro6
| 2.9nC 0.05 ohms at -4.5V
|
| Fairchild nds336P 20V, 1.2 - 10A, 0.27 ohms at -2.7V SOT-23
| 1.8nC 0.2 ohms at -4.5V
|
| Fairchild nds332P 20V, 1 - 10A, 0.41 ohms at -2.7V SOT-23
| 0.9nC 0.3 ohms at -4.5V
|
| Sanyo 2sk2909 20V, 0.8 - 3A, 0.3 ohms at -2.5V SOT-23
| [ 2ns 4V switching time spec'd ] 2nC
|
| Philips bsh203 30V, 0.3 - 1.9A, 0.92 ohms at -2.5V SOT-23
| 0.25nC 1.1 ohms at -1.8V
|
| Supertex LP0701N3 16.5V, 0.5 - 1.25A, 1.7 ohms at -3V TO-92
| ~0.5nC 2.0 ohms at -2V
|
| Hitachi 2sj483 30V, 5 - 20A, 0.12 ohms at -4.0V TO-92mod
| Hitachi 2sj496 60V, 5 - 20A, 0.17 ohms at -4.0V TO-92mod
| Hitachi 2sj386 30V, 3 - 5A, 0.55 ohms at -4.0V TO-92mod
| Toshiba 2sj537 50V, 5 - 15A, 0.27 ohms at -4.0V TO-92mod
| Toshiba 2sj507 60V, 1 - 3A, 0.72 ohms at -4.0V TO-92mod
| NEC 2sj178 30V, 1 - 2A, 1.5 ohms at -4.0V TO-92
| NEC 2sj196 60V, 1 - 2A, 1.5 ohms at -4.0V TO-92
| Toshiba 2sj509 100V, 1 - 3A, 1.68 ohms at -4.0V TO-92mod
|
| Supertex VP0104N3 40V, 0.25 - 0.8A, 11 ohms at -5V TO-92
| Supertex TP0620N3 200V, 0.18 - 0.8A, 9 ohms at -5V TO-92

Thanks,
- Win