# high-side mosfet scheme

Discussion in 'Electronic Design' started by Tolstoy, Aug 25, 2007.

1. ### TolstoyGuest

I used to work on a bike with an old-fashioned generator that
had a segmented commutator and put out dc. That means that a
voltage regulator has to have a rectifier in it, to block
the battery from discharging into the armature when you
turn the bike off (they used a relay in the old days).
I built a solid state regulator for the bike using a
shottky rectifier, but now I'm thinking about ways to use
a low-resistance mosfet in the high side so it doesn't
require a heat sink. The suggested current limit for this
generator is ten amps. A ten-mOhm mosfet would only make
one watt of heat. The MBR3035PT shottky created five
and a half watts at ten amps. So now I"m trying to figure
out how to control a mosfet in the high side as
a blocking diode and have it turn on and off, like a
syncronous rectifier.
The circuit will have a comparator that controls the
field current, so I figured I might use a comparator for
controlling the power mosfet as well. Since I'll have a
charge pump for driving the mosfet gate, I can run the
comparator off it. This means I can use an ordinary
comparator to sense the voltage across the high-side mosfet.
Here's an illustration of the basic topology I envision

,-----------,
| |
||-+ |
||<- |
charge pump>---+-R1-+-||-+ |
^ | | | |
| | | | |
| | | | |
| | | | |
generator | | | |
armature--+----|----|----' |
| | | |
R2 | '------, |
| | | |
| +-R4-, | |
| | | | |
+----|-||-+ | |
| |\ | | | |
'-|-\| | | |
| \ | c |
LM393| >---+-----b |
| / e |
,-|+/| | |
| |/ | gnd |
| gnd |
'--------------R3----------+
|
| +
-----
---
-----
---
-----
---
| -
|
gnd

The only problem is that the comparator may chatter
at low charging currents. For example, if the comparator
has an offset voltage of -5mV and the mosfet has 5mOhm
Rds(on), then at currents less than an amp it will
oscillate. I put the negative feedback capacitor there
to alleviate it as much as possible. And though it's
unlikely the generator will have to put out very low
charging currents for any length of time,
I still would like to find a way around the oscillation.
Any suggestions?

2. ### John O'FlahertyGuest

Why does it need to be such a sensitive switch? Couldn't you use an
absolute reference for the turn-on, depending on the armature being
well above some threshold while it's working?

3. ### TolstoyGuest

It seems like they would have done it that way
back in the day they used relays.
They must have set the relay to engage at 15 volts or something
on the armature and cut out at say 11.5, or whatever the armature
would drag a fully charged battery down to, once the generator
stops. I don't actually know how the
cutoff relays in the old mechanical regualtors worked.
Anyhow, taking that approach would call for a fixed voltge
reference and a comparator with hysteresis.
I'll have to think about it.

4. ### ehsjrGuest

Yes - add a resistor for hysteresis. See
http://www.maxim-ic.com/appnotes.cfm/an_pk/3616

Ed