watering system, second attempt.

after the comments on my first version, i've redesigned the circuit using
CMOS ICs and a mosfet to switch the coil. I was meaning to get it working
before i posted the new design, but i'm waiting for some parts, so i'm
posting it now to see what people think. as before, i would appreciate any
comments on how i've done it.

mostly what i want to know is if there's
anything that will stop it working the way i'm expecting, cause parts to
fail, cause a hazard, or if i've missed any obvious simplifications to the
design i've come up with. it's a bit more complex than before, because
i've added a feature to water in the morning, evening or both.

I've tested the part of the circuit that gets the light on/off signal from
the LDR using the 393, and xors it with the 'force watering' switch
signal, but then a chip failed and i'm waiting to get another one.

http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip.jpg
http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip.png

if you still think it's junk then say so, but i'd like to know why.

 

I'd have been interested to study your circuit - but it's too large
for comfortable reading. And if I try resizing it 50%, your dark
colour scheme produces a barely readable result. What's wrong with
conventional black on white, with occasional colour for special
purposes?

Also, if you really want the feedback you've specified, a full text
description of your objectives and approach would be helpful. Or are
you expecting readers to remember the previous thread, or search for
it?

 

A PIC circuit would take care of the size *and* be a simpler
approach. For all I know, there is a PIC, but the drawing is too big
(literally) of a hassle to look at.

LT Spice would be a good way to put this to the group.

 

You wouldn't be the first to publish a PIC design on the net by any
stretch of the imagination. I made my pic burner with a few
jellybean parts and the loader software was free.

The spice netlist won't let everyone and their family see the schem.
LT Spice, being free, has an ascii schematic format and we can load
it and it will do the netlisting and all that.

 

On Thu, 05 Aug 2004 01:36:35 +0100, andy wrote:

You could at least pick the smaller file size of the 2 images and
post one link. I don't see where the drawing is any smaller
physically.

It looks like a lot of board real estate. If you google on
irrigation timer, some dude in either Oz or Kiwi land has a PIC
based programmable irrigation timer he sells for around 150 of
someone's dollars and IIRC he's posted the code *as well* as the
schem. It appears to be not much bigger than a pager.

 

that's just the way the program i have writes out an image.
I've run it through a paint program to make it b/w and smaller:

http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip.jpg
http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip.png

It's meant to send a 1s 13A pulse through an electromagnet at the
beginning and/or end of every day, depending on whether the 'morning' or
'evening' switches are set. Will also trigger if the 'force' button is
pressed. there is a holdoff of up to 2 hours 40 min to stop the circuit
retriggering once a pulse has been triggered. The day/night sensing is
with an LDR, with a variable light level setting and a bit of hysteresis.
The 'reset' button resets the holdoff timer.

When the pulse is triggered, this will release a magnet which is holding a
ballcock valve closed in a water butt, and set off a single cycle of
filling and emptying the tank through some drip hose.

The design i'm going for is like this:


| :#: iron cored electro- |
| :#: magnet. |
| m |
========|X------O ballcock valve with |
| magnet (m) on top. |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| =============
| | slow drain
'--------------------------------' into drip hose.

The tank is normally empty, with the magnet holding the ballcock shut
against the electromagnet's core. Then a short pulse through the
electromagnet should make the ball drop down. The tank fills quickly,
the valve latches shut again, and then the tank drains slowly through the
drip hose.

the idea is:

a) a short pulse to cancel the magnet's field should take less current
than having to open a valve against friction. and much less than having to
hold a solenoid valve open for the whole watering cycle.

b) the system always comes back to a stable off (no watering) state by a
purely mechanical process once the initial pulse has been sent. i.e. if
the power goes, it can't get stuck open.

c) it's easy to build out of common or garden parts - don't have to buy an
expensive electrically operated valve.

andy.

 

doesn't that mean writing code and burning the EPROM in the PIC? This
isn't what i want because (a) this isn't something i have the equipment
for, and (b) i want to be able to publish the design on the net as
something anyone can build with just the parts and a soldering iron.

i tried to convert my schematic editor's file into spice format, but it
isn't working.

 

i did that because the png is a better image, but not everyone can read it.

if you still have the link, could you post it - i can't find it with

'irrigation timer pic site:.au'

 

I tried both and wheter I zoom in or out, it's a bitch to read. Must
be those colors and/or that schematic capture program you're using.

Oh. It was propagation misting timer and he no longer has a link to
the plans. $130. You can see a picture of it and seriously, the
learning curve for working with PICs is not steep and it's worth the
effort.

Good resource on PICs is www.piclist.com Thoroughly read the first
page and FAQ about subscibing and posting if you want to get on the
list and either way, there's enough info and links there to control
the world.

 

Can't see any change! Still that awful colour scheme, and still too
large to view comfortably.

 

sorry, i re-did it and then forgot to change the link. it should be:

http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip-bw.png
http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip-bw.png

 

i'm using gschem for linux, which is part of the gEDA suite.

i've converted it to b/w and rescaled it - should be easier to read now.

thanks for the info, but i want to stick with a separate chip design.
partly because i don't want to learn a whole new thing to get this project
done, and partly like i said because i want it to be possible for someone
else to build it from the plans without any fancy equipment - i'm talking
about people who have hardly any electronics knowledge but can follow
instructions to assemble a circuit on stripboard.

 

That's more like it. After increasing contrast and tweaking the size a
bit, I now have it printed in landscape on A4.

At first sight it does seem a bit complex, given its fairly simple
intended function. But I recall that curiosity (about your
electromagnetic approach) was a major motivator, so I guess that may
prove a fair price.

Nice timing chart, BTW.

Do you have it working yet?

 

Looks the same to me.

I've seen posts on PICList from people who can't even bias a
trasistor or select a resistor for an LED

But I understand. It's something you can mess with later.

 

<snip>

That's loads better. I see you read up on solenoinds. There's an app
called Maxwell SV (that's the free stoodint version) at
http://www.ansoft.com/

That might help you with those kinda EM things in the future.

I don't see anything obviously wrong, except I would've eliminated
that glitch and there's gotta be a simpler way with fewer gates, but
I wasn't here whenever you laid down the specs.

You may as well build it and try to beat it to death.

 

sorry, it's in the other thread.

http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip-bw.jpg
http://www.niftybits.ukfsn.org/electronics/daily-water-with-chip-bw.png

 

On Thu, 05 Aug 2004 20:06:29 +0100, andy wrote:

That'd be enough for me to look for a better way.

^^^^^^

I tried that spelling and got tired of it and went back to Active8 -
before deciding to hang out here Seems like some people think
your version is better. I used to introduce myself as Michael so I
wouldn't feel the need to react to "Mike" when there's 20 Mikes in
the room.

"Hi, Bob. I'm Michael."

"Nice to meet you, Mike."

So much for that.

"No, the other Mike. No... not you. The one over there."

Maybe if I knew the specs and why all the delays I could suggest
something. I'd think a comparator followed by an integrator would
deal with varying light levels. We might get it down to 3 chips or
less.

 

i was looking for something a bit more specific than that, like
suggestions on how to make it simpler if that's what you think. it doesn't
seem /that/ complex to me - it's only 5 chips after all.

the only things i can see that would make it simpler are:

- maybe get rid of the 393 and just use one of the schmitt trigger nands
as the comparator.
- use some kind of direct wired logic (switches rather than gates) to do
the morning/evening selection.
- or go for the PIC approach like Activ8 said, which i don't want to do.

not yet.

 

<snip> - Like it's not obvious.

One further thought. It depends on at which light levels you
consider indicative of day and night, but perhaps a window
comparator is the ticket. Then you can set thresholds for 2
different light levels. A quad op amp would serve the purpose and
leave you 2 amps for integrators to filter out disturbances like
casting a shadow on the photocell.

Then maybe and only maybe 2 gates for AM/PM- not needed if you
switch the comparators inputs to force them active/inactive.

Then a one shot to keep the valve open long ewnough. I think I'd use
the water utility, though, and work in a soil moisture detector.
Then go on vacation and bugger worrying about AM/PM switches or
filling your tank.

 

OK, my first thoughts are that it could be implemented with one cheap
chip, along the lines shown here:
http://www.terrypin.dial.pipex.com/Images/Greenhouse1.gif