[Message Prev][Message Next][Thread Prev][Thread Next][Message Index][Thread Index]

Re: one wire sensors



On Mon, 09 Jul 2007 06:10:27 GMT, Jeff <dont_bug_me@xxxxxx> wrote in message
<njkki.5741$tj6.3629@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>:

>Have you done anything with picaxe?
><URL: http://www.rev-ed.co.uk/picaxe/ />
>   It looks like that should handle the DMX also. Thanks for the tips.

No. PICAxe is Microchip PIC. I've been using Atmel AVR with BASCOM
www.mcselec.com.  Much more scalable than PICaxe.

DMX is more demanding of the MCU than it seems at first blush. For a few
inputs/outputs, it would not be worth the effort in my opinion and standard
8-bit data resolution is often insufficient for data collection.

>   I'm trying to go solar here. I've got a fairly large solar air heater
>on the south wall and am working on 120 ft2 of solar water. When I have
>some more time I'd like to run a dessicant solar AC. All of that needs
>some logic, some logging and some adjusting! Probably lots of adjusting.
>
>   I'm thinking now of a picaxe for each device and a DMX cable back to
>another picaxe for logging and master control:
>
><URL: http://www.rev-ed.co.uk/docs/AXE110.pdf />
>
>but I'm not crazy about that idea!

>Wonder what it would take to write to a USB memory stick as I see that
>picaxe can read to USB, but I don't know about the reverse. Could import
>the data to a PC later and sort it all out later, though I see that
>timestamping is not trivial.

As I understand it, you want both real time control and logging. This is
trickier than it may seem at first. What happens to the logging, for example,
while you are reprogramming?

This is one reason that I've decided to go with separate logging and control
systems using parallel connections to the sensors. Hence the PCBs that I am
making. They will also permit some backup sensors via Dallas 1-wire.

>   I expect the first boards back
>> in a week or so. If I like the  prototypes, I may put together a large
order.
>>
>> Most HA systems are not very good at data logging (checking, interpreting,
>> representing, and storing data). This system is designed to address that
need
>> in a robust and redundant way.
>
>   From the little I've looked into it, I can see that's a big complex
>problem. I suppose it would be easier to dump it off to a PC to process
>and store but that negates the whole high availability, low power
>concept. This gets hard fast!

It is simpler if you don't try to a)  measure, supervise, and control and b)
log with the same MCU/CPU. And there are some tricks to maintaining integrity
and usefulness of the data base of logged values.

For example, if one connects the output of the power-to-frequency sensor used
to monitor AC line loads I show here: www.econtrol.org/power_measurement.htm
to a Dallas 1-wire 32-bit counter (2^32), you can read whenever you want to
get instantaneous value of power because the output is about 1000hz
(counts/sec)  full-scale. But the 32-bit capacity also lets you go a month
without reading the data without losing the total for the month.. If you were
measuring voltage, and you read only once a month, you'd have next to no
data. And when (not if ;-) you lose some of the data, you have to figure out
how to guess ('bridge') what it was in order to estimate monthly total. And
so on.

So some forethought in how you will deal with the data in the PC ahead of
time will save you problems later that may be insurmountable.

... Marc
Marc_F_Hult
www.ECOntrol.org


comp.home.automation Main Index | comp.home.automation Thread Index | comp.home.automation Home | Archives Home