[ha] Re: Good ideas at work here, but...

Mon, 23 Aug 1999 00:24:43 +0100

On Sun, 22 Aug 1999 17:13:50 -0400 John Mullan <jmullan.at.cgocable.net> wrote:
> The whole theory here sounds like it is developing well.
Ah, if only that were true. The list has been very quiet for months.

> ... For the last year or so I've wanted to develop some of my own
> stuff based around X10 but find it difficult getting anywhere hardware
> wise because, in part, of my own lack of abilities and the inability
> of getting any proprietary integrated circuits.
I think this is a problem we share. There don't seem to be any people
on the list who can design the circuits we need to actually shift bits
around the place.

> I think we should get a group started on the actual interfaces that will be
> used to integrate the various planned media types.
Good plan.

> This will enable people to start testing various protocol ideas. No sense
> testing protocols without working transmission types (in my humble
> opinion).
Well, I think it would be sensible to get some ideas for the higher
level protocols in the absence of of real transmissions, given that we
still don't have any mechanism for trasmissions after more than a year.
The high-level protocol aspects should be pretty much ignorant of the
hardware used to shift the bits about anyway. But I do agree that
having working hardware would help to focus the mind in a way that a
paper protocol design would not.

> I am more than willing to build some device/interface prototypes if someone
> is willing to help design some. I imagine that powerline interfaces will
> be a good one to try but not sure how to develop modulated carrier
> circuits.
Same here. I would love to have a play with some power-line modems, but
don't feel confident designing one myself.

> If we use a multi-mastered protocol and a lamp switch can talk directly to
> the lamp, there needs to be a way to tell the switch/lamp that they are
> related.
I had a notion that the two units would be introduced to each other
by some sort of unusual operation sequence. For example, switching
lamp module off and on within a second, then operating the relevant
light switch three times. For a simple switch operation, this is enough.
For more complex operations, an intelligent interface unit could use the
unusual operation to select the nodes to be programmed and then offer
the user a menu of possible connections to be made.

> Since any one switch may need to be any address (per it's users
> wishes), there needs to be a method of loading at least a few basic
> parameters into the module.
I'm not sure a special loading interface is required---if every module
has a unique serial number (either built in by a DIP switch or EEPROM
location, or acquired by random selection followed by collision detection)
then all other parameters may be loaded through the communications medium
just by addressing the module.

> Perhaps (space permitting) a 3 conductor mini
> stereo jack would suffice to input data directly (as if it were coming from
> the power-line or other medium).
It might be a good idea to have a standardised way of doing this, but
I don't think it would be necessary. A way to manually reset a unit in
an emergency or just to start again from fresh would be sensible---just
a simple reset button would do and could surely be fitted in more easily
and with fewer safety concerns than an electrical connection. The unusual
operation mentioned earlier could then be merely pressing the reset button.

Like John, I feel the need to get some hardware built to prove I'm
actually getting somewhere, so I have just started on my design for
a simple touch-sensitive wall-mounted LED control and display unit,
even though I currently have nothing to display or control:-) So far,
I have the touch sensor bit partly working. One of those little OPB706B
infrared reflective proximity sensors and a 4070 hex inverter can
reliably detect a finger from a distance of a few millimetres through
a dark red bezel which fits the front of a standard 96x48mm DIN case.
This needs to be generalised to several sensors mounted near each other.
Currently, the modulated light from one sensor would trigger the others,
so I need to add a multiplexing mechanism. I also have some cute 5 by 7
red LED dot-matrix displays (Kingbright TA07-11) which I need to figure
out how to drive. Does anyone have experience of this sort of thing?
They run indefinitely on 20mA or take a maximum of 150mA at 10% duty
cycle. The LEDs are connected by row and column pins, so they have to
be multiplexed. How many rows can I run before the multiplex time makes
it too dim to see? I hope to have six of the displays (and six of the
photosensors) side-by-side, so they will be visually 7 rows by 30 columns.
I could wire them this way too, or group them in other patterns, such as
2*7=14 by 3*5=15, which might be easier to drive from a microcontroller
with a limited number of ports. I'll probably use a 68HC11 for now as it
has enough spare capability to do the sensor scanning and bus protocol
(whatever it turns out to be). Until we get a power-line interface,
the unit could respond to and issue IR remote control signals.

I reckon the cost of the control/display box will be comfortably less
than 30 GBP (50 USD), so I can afford to have a several of them dotted
around the house. I would really like to have touch-LCD panels, or even
LCD panels with soft buttons alongside, but the cost of pre-built units
is prohibitive and the practicalities involved in making a home-built
unit look nice and be visible in all light conditions are daunting (and
it still wouldn't be cheap). The LEDs in my box are large enough and
bright enough to be seen across the room in full sunlight.


R.M.O'Leary <robin.at.acm.org> +44 7010 707044  PO Box 20, Swansea SA2 8YB, UK