'power supply control'
|G'day to all!
I've finaly gotten around to working on a project I mentioned a few months
ago - a little board that receives and sends 9600 baud comms as well as
outputting continous PWM, monotoring battery voltage, pushbutton, and
talking to 28 LEDs. The whole thing wound up fitting into an 8 pin PIC -
I'm using a 12c671 since I need a/d input but the code works just fine in
the 12 bit 12c509.
This whole thing has to run from a pair of AAA cells. I'm using a LT1303-5
switcher - seems to work just fine. My problem comes from turning the PSU
on and off.
My 1st attempt used the low battery detector built into the LT1303 as a
latch. This detector becomes disabled when the LT1303 goes into shutdown -
make a nice low power latch. Serial input or a button press starts the PSU
- the LB output is tied to the shutdown pin (10K pullup); the LB input goes
to gnd with a 10K pulldown. Once the PSU starts, the LB stage becomes
active and holds the shudown pin LO which keeps the whole thing running. An
extra shift register output is tied to the LB input - a HI allows the LB
output to go HI and shuts the PSU down.
The problem I'm having is on startup from serial input (button startup is
just fine). The start bit is only 105 uS long - it takes at least 1 mS for
the PIC to come to life (the windowed 12c672 parts require the power up
timer to be enabled). I'm keeping the SR outputs disabled during startup
(so that the latch stays on once it gets triggered) but its not reliable.
PIC is awake but the PSU is not - as a result, the PIC goes into lala land
(supply voltage with weak battery is WAY too low).
So I had another brainstorm - it seems to work but I wanted to bounce it off
the gang to make sure I'm not gonna ship something flakey.
The idea is to make the PIC decide whether the PSU should be operating or
not. The easiest method seemed to be to monitor OSC2 - if it has signal,
the PIC is awake. I know this has been discussed in past - buffer OSC2 and
use a charge pump to control the gate of a FET. But the prototype boards
I'm about to ship to the client don't have that much room on them; and I
really hate to ship that OSC2 signal anywhere (EMI considerations).
Instead, I made use of the *average* DC level at OSC2. I'm using a 1 meg
resistor tied to the gate of a VN2222L mosfet. The gate capacitance of the
FET and the 1M0 resistor form a nice low pass filter that drives the FET
quite nicely. The FET has to sink about 100 uA to keep the PSU alive - the
curves in the Siliconix data book show that the minimum gate voltage
required (worst case) is less than 1.5V. OSC2 is solidly at gnd when the
PIC is in sleep - should be no problem with turnoff. The extra capacitance
caused by the 1M0 resistor (1-2pF) shouldn't affect the resonator and there
is no DC loading by the FET gate - I think it should work just fine.
As it stands, the whole PSU control is the FET and gate resistor, as well as
a diode from the pushbutton and a NPN transistor monitoring serial input.
There is a 47K pullup on the shutdown pin which is also tied to a 2.2uF cap
to keep the PSU alive until the PIC's oscillator starts. Either a button
press or serial input pulls the shoudown pin LO and discharges the cap, the
PIC's oscillator keeps the pin LO once it starts.
I'm open to other ideas as well as hearing any criticism.
Dwayne Reid <planet.eon.net> dwayner
Trinity Electronics Systems Ltd Edmonton, AB, CANADA
(780) 489-3199 voice (780) 487-6397 fax
Celebrating 15 years of Engineering Innovation (1984 - 1999)
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Do NOT send unsolicited commercial email to this email address.
My posting messages to Usenet neither grants consent to receive
unsolicited commercial email nor is intended to solicit commercial
> The problem I'm having is on startup from serial input (button startup is
> just fine). The start bit is only 105 uS long - it takes at least 1 mS for
> the PIC to come to life (the windowed 12c672 parts require the power up
> timer to be enabled). I'm keeping the SR outputs disabled during startup
> (so that the latch stays on once it gets triggered) but its not reliable.
> PIC is awake but the PSU is not - as a result, the PIC goes into lala land
> (supply voltage with weak battery is WAY too low).
I am developing something similar at the moment except the circuit is
for a smart energy transfer system.
My method is to buffer and charge pump the serial input to turn the
reg on, the only trouble is I lose the first byte/packet as the
circuit doesn't wake up in time to use it. The system however is aware
of this short coming and the TX sends the info twice and at a lower
baud rate. From reading your text I understand there is some power
available to the uC to wake up before the PSU?
My application is totally off in standby (contradiction in terms?) so
maybe the comparison is useless.
Uni of Auckland
More... (looser matching)
- Last day of these posts
- In 1999
, 2000 only
- New search...