On Jun 22, 9:39*pm, Tim S wrote:
Jules coughed up some electrons that declared:
On Mon, 22 Jun 2009 09:54:21 +0100, Tim S wrote:
Just had a call back from the designer (who I guess is also the owner)..
Very interesting... They use a custom design latching solenoid valve,
driveable by pulse in the range of 6-24v. And he seems happy to sell the
valves on their own.
Nifty. It's probably worth checking what testing's been done in terms of
estimating the number of cycles it'll stand before failing, as that'll
dictate what applications it can be used for.
This is a good point.
Like a PIR in the bathroom to operate the extract fan, but at teh sam
etime cut off the water to prevent flooding.
Stuff like that makes me nervous in case the valve fails - it doesn't look
like it has any feedback as to current state, so you really need extra
gubbins 'downstream' to check that the valve has actually done what
you've asked it to.
Indeed. Regarding the "flood prevention" - my view is that it doesn't make
the situation any worse than not having a valve (unless the valve itself
suffers catastrophic failure and ****es water everywhere - which hopefully
is a very unlikely scenario compared to "fails to control the water" mode
of failure[1]). So it's a net win.
[1] I'm expecting to see something loosely based on a washing machine valve
design with a clever drive mechanism - but I might be wrong.
However the lack of positive feedback is a pain, especially for diagnosing
problems - lack of control on a radiator (my main intended application)
will result in the room beiong either too hot or too cold.
I was thinking about this in the car this evening... The obvious solution is
to put a flow switch in.
This however is expensive and also prone to fail being a mechanical part.
So, I thought - seeing as I'm going to whack an AVR microcontroller next to
the valve, why not stick a temperature sensor on the output pipe next to
the valve? Could be a Dallas 1-wire jobbie (I like them and have driven
them from AVRs with excellent success) - or even a thermistor and use the
AVR's A/D converter.
In any case, one would expect 2 things of the temperature on a pipe:
a) When valve is closed, output pipe temperature should tend towards room
temperature over time, where I might expect a time constant of minutes;
b) When open *and* downstream is accepting flow (always with a rad,
intermittently with taps/shower/etc) then the pipe temperature is either
going to hit perhaps 10C or less (mains cold) or 50C+ (hot water). These
are pretty extreme temparture changes to be looking for and I think should
give *a reliable indication of the radiator valve working. For flood
prevention, one should expect a reliable indication that there is no flood,
but one will never be sure *all the time* if the valve is opening
correctly, but you will get indications *some of the time* (tap on) so
could do long term analysis and flag up a failed valve.
SMWBO taking a cheesegrater to my nuts after the shower lost hot water of
course is the 100% reliable indicator the valve died(!)
I'll ask the bloke if I can buy one for a sample and test it...
Cheers
Tim
lol, maybe thats where you should put the thermistor
Thermistors are very simple. A more basic option is to repeat the
drive pulse a bit later, which is trivial in software. A mechanical
flow meter is just one more relatively vulnerable failure point.
And ultimately some failures are easier to let happen than get into
complex solutions, depending on your app and control method.
And of course sometimes software can deduce failures from system
behaviour.
NT