i think you didn't see the version i posted after john popelish's comments:
http://www.niftybits.ukfsn.org/electronics/daily-water3.png
http://www.niftybits.ukfsn.org/electronics/daily-water3.gif
Instead of three TIP3055's beasts being driven as emitter followers,
why not use a single MOSFET which can handle a 13.3 amp load, (which
is what your three windings will look like in parallel) and make it
common source?
like i said in my reply to john popelish, they aren't meant to be NPN
emitter followers but PNP switches - I just bought the wrong part by
mistake, and then copied the part number into the diagram without noticing
it was the NPN version. my mistake, but that wasn't the mistake.
+12>-------+
|
S
VIN>-----G IRF4905
D
|
+----+----+------+
| | | |
[L1] [L2] [L3] [1N4001]
| | | |
GND>-------+----+----+------+
paralleling the coils and using a bigger transistor is fair enough -
that's just me not thinking. I used bipolar ones just to stick with what i
know for this project - it's only the second i've built.
The only current you need to drive it with is what's needed to charge
and discharge the gate capacitance, and with a channel resistance of
20 milliohms and a 13.3 amp load, the IRF4905 will be dissipating
about 3.5 watts for the 1 second it'll be on, which is nothing.
That does sound like a better idea given the number of bipolars i've had
to use to amplify the signal enough to drive the output stage.
Next, let's look at what you're using to drive the 2N3055's:
+12>------------------------+
|
[50R]
|
+----+------+
| | |
C | +--[10R]--->TO 2N3055'S
VIN>---[340K]--B Q1 | |
E | |
| C |
+---B Q2 |
| E |
[2K2] | |
| | C
+----+-+---B Q3
| E
[1K0] |
| |
GND>-------------------+----+
Notice that in order to keep the 2N3055's turned off, Q3 has to be
turned on, which will be doing your 1.2AH battery no favors since
it'll have to supply current through the 50 ohm resistor ***ALL THE
TIME***.
That's still the same mistake/misunderstanding over PNP/NPN transistors.
yes i did draw it wrong, but the version i put up in reply to john
popelish is right i think. in this version, that triple darlington is
meant to be to switch /on/ a PNP switch, not switch /off/ an NPN follower.
Assuming Q1, Q2, and Q3 are collector-to-emitter shorts, 12V
through 50 ohms is 240mA, which means that the battery will be drained
in about 5 hours after the circuit is connected. Period. End of
story.
Like i say, i did think about this, and apart from the mistake with the
part number (should have been tip2955 for the PNP design i drew), i think
the original circuit was OK for this. What i was trying to do was keep the
current through the two input stage transistors as low as possible (hence
the high resistor values), but not worry so much about the output stage,
because most of the current in this would be used by the coil anyhow.
But, if you change your driver and output stage to look like this:
+12>-----------+----+
| |
[1K] |
| S
+---G IRF4905
| D
C |
VIN>--[10K]---B Q1 +----+----+------+
E | | | |K
| [L1] [L2] [L3] [1N4001]
| | | | |
GND>-----------+----+----+----+------+
You could use one of your 2N5551's for Q1, and then the only time
current will be drawn from the battery is during the time when the one
second pulse will drive the base of Q1 high.
Assuming no self-discharge and a fully charged battery, with a
capacity of 1.2AH and a load of 13.3A for one second every 24 hours,
the battery would last for 0.09 hours, which is 324 seconds. That
translates to 324 days... How big is your solar panel?
Thinking of using a 50mA panel, which should give a long term average
current of about 1.8 mA minimum in uk winters, according to the table i
have.
---
If you take 13.3 amps out of the battery for 1 second it should take
7389 seconds to fill it back up at 1.8mA, but it'll take longer
because there is no free lunch. Figure about 50% more, so that'll be
about 11083 seconds, which is about 3.08 hours. Not bad, but that's
not counting the quiescent current of the circuitry which, if it's
1.8mA will keep the battery from charging in winter...
---
that is what i wanted to know.
---
You're welcome?
Oh, well, it's a slow day, so here are some thoughts about the front
end...
You've got basically four approaches to choose from when deciding how
to generate the 1s pulse:
1. 7555/7556
2. 4538/HC123
3. LM393 or some other comparator array.
4. Discretes/glue logic
The 7555/7556 route looks attractive because of low quiescent current
and a totem-pole output which would eliminate the wasted current drawn
by a pullup resistor-open collector feeding the driver.
Unfortunately, the slow-rising DC voltage from the LDR would have to
be differentiated and AC coupled into the timer, which would require
additional circuitry to do the job.
The 4538 looks very attractive because of a totem pole output and DC
coupled Schmitt trigger inputs, which could help to prevent
retriggering because of noise on the output of the LDR circuitry.
Unfortunately, the amount of hysteresis isn't specified. That is, I
couldn't find it in the spec's. Still in all, there are two timers in
each package, so the second one might be able to be used to hold off
the first for long enough until it gets dark enough outside that there
won't be a chance for a retrigger. That would also eliminate the need
for hysteresis and you could use Fairchild's 4538BC part which has no
hysteresis, but some other nice stuff.
Using LM393's looks attractive because of the flexibility allowed, but
because of no totem-pole output there'll always be a drain on the
battery which can be avoided by using one-shots like 4536's. Also,
LM393's have higher quiescent current requirements than CMOS does.
Discretes... Unless you're doing this because you want to learn how to
do it with discretes, forget it.
So, my vote is for the 4538, and I'll post a schematic of how to do
the whole thing to alt.binaries.schematics.electronic under "Water
timer" in a few minutes.