Re: Idratek

[David Gumbrell <david.gumbrell@xxxxxxxxx>]

On 2/3/06, Chris Hunter <cjhunter@xxxxxxx> wrote:
> Yes, thanks ... I found that under (I think) the software category ...
> but I found it too often confusing (probably the way things were said
> more thananything else ) and it stopped rather too short on the detail
> ... if I'm about to set-out with Idratek, I need to know more about
how
> it works, how it's wired, how it's programmed, how easy it is to
adjust
> things as time goes on, how robust it is, how failure strategies have
&
> should be handled ... etc ...
>
> Chris

If you can, I'd visit one of their sites so that you can have the
stuff demoed. There's a  lot of functionality in there, and you would
also see how it's set up and configured. If you discuss your
requirements they can give you a lot of advice or even install it.

My attempts at short answers:

1. Control cabling is "free topology twisted pair". This means
that
you don't need to "home run" each connection back to your node-0.
I
typically do a home run from the ceiling sensors and then short drops
from there to wall plates. You need 2 pairs minimum, 3 if you want the
audio bus. Most people use Cat5, I use the C-Bus Cat5, since it's
mains rated insulation.
Mains cabling is up to you -- you can keep a fairly traditional wiring
system and use the onboard relays, or star wire loads back to a DIN
enclosure and use the DIN relay modules.


2. How it works -- that's rather open ended ! They have a proprietary
high-resilience data protocol on one of the pairs in the LV (Cat-5)
cabling. The various modules can send and receive packets on this
cabling. These are called Reflex packets. There is also an analogue
audio bus on the third pair, which can be driven/received at audio
modules, so you can use the MFPs as intercoms, baby monitors etc. The
Cortex program can inject audio and receive audio via the PCA module.
This means that you can have spoken menus and speech recognition.
Functionality can be defined in 3 layers --- PC-less "Reflex"
mode,
and in Cortex via the pre-canned functions and via user-created macros
(which are built as lists of built-in action-steps). There is another
layer that I can't talk about.

The Reflex functionality is mostly used as backup system that allows
you to have a functioning system even if the PC crashes or is offline
for some reason. Reflex packets are sent by sensors in response to
triggers (eg PIR detect, switch press/release or temp threshold
crossing). Received Reflex packets actuate stuff, like relay closures,
and can include timed functions and other stuff. However, it's a bit
cleverer than just trigger-->packet-->actuate. The trigger can cause
a
sequence of Reflex packets to be sent out, and the list can include
packets addressed back to the originator to execute various commands.
These packets can include rewriting parts of the list, or changing the
entry point. So you can write little programs that whizz round doing
stuff. However, I expect that most people would set up fairly simple
fallback Reflex functions (ie keep light switches working, keep boiler
control going etc) and define most of the functionality in Cortex.

In Cortex, there are a lot of pre-defined functions. If you look at my
blog, or this picture:
http://www.gumbrell.com/archives/home_automation/landing/landing-light-behaviour.gif
you can see a whole bunch of standardised behaviours for the light.
Depending on what sensors you have, some bits may be greyed out. Other
objects have similar "behaviour" dialogs. In my experience, if
you
find you want to do something more complicated, and can't see how, an
email usually results in either an explanation or a Cortex update to
add a new feature to achieve it.

3. Adjustments are made through modifying the behaviours or tweaking
thresholds or whatever - basically depends on how you configured the
functionality -- all through the GUI. I think all the sensors have
both activity lists and graphs of activity available to help you
understand what is going on. They also have "simulate" buttons so
that
you can test the functionality you have defined. The Cortex floor plan
is animated, so you can watch your simulation work through. You can
watch the non-simulated stuff too, like the other half coming upstairs
to tell you to get on with something else ...

4. Robustness: if you give Cortex a dedicated PC and don't fill it up
with crap it's going to be pretty good. I keep mine on a UPS to avoid
any mains glitches. You can create an automatic watchdog that will
fallback to Reflex if the PC stops controlling the network. You can
also have a backup PC with automatic Cortex failover, but I have not
tried this. There are very occasional software glitches. In my
experience these occur when I am messing about with stuff
experimenting or trying to set up weird schemes. I think I have had a
fallback to Reflex twice in two years. A restart of the network or the
Cortex software usually clears them up, sometimes a reboot. I don't
recall having significant problem that could not be rapidly fixed. The
only slightly irritating problem I have at the moment is restarting
via remote desktop (for which I have a workaround). Cortex can be set
to automatically update itself also.

5. Failure strategies -- as mentioned above, backup PC and/or basic
Reflex program + UPS for the PC and Idranet. Obviously a full power
failure means you have nothing to control, but partials may occur if,
for example, you have a split load CU with an RCD.

OK, so much for the short answer !

Here's a really short answer -- if you are planning a new build, get
Idratek to advise on modules, wiring etc and get them to commission it
to your specification. You can then go up the learning curve at your
own pace.

Since I'm doing a gradual refit I'm doing the learning curve as I fit
stuff. I cannot stress enough how excellent the support I have had has
been.

I hope that helps.

-- Gumby