433MHz to LAN gateway or miniserver?

[Joerg]

Is there a cheap (as in mass produced and <$100) pod that plugs into a
LAN port of a regular router and have a, say, 433MHz radio link?
Something that could then connect to several thermometers or weather
stations and ask for their readings.

It would have to be accessible from the web by a remote server for
polling purposes. Maybe it could just be a mini webserver in itself that
gets polled. 2.45GHz often wouldn't work all that well because of range.

 

[Martin Riddle]

Is there a cheap (as in mass produced and <$100) pod that plugs into a
LAN port of a regular router and have a, say, 433MHz radio link?
Something that could then connect to several thermometers or weather
stations and ask for their readings.

It would have to be accessible from the web by a remote server for
polling purposes. Maybe it could just be a mini webserver in itself
that gets polled. 2.45GHz often wouldn't work all that well because of
range.

A weather base station with a serial port and this
<http://www.lantronix.com/device-networking/external-device-servers/uds1100.html>

Over $100

Cheers

 

[Frank Buss]

Is there a cheap (as in mass produced and <$100) pod that plugs into a
LAN port of a regular router and have a, say, 433MHz radio link?
Something that could then connect to several thermometers or weather
stations and ask for their readings.

It would have to be accessible from the web by a remote server for
polling purposes. Maybe it could just be a mini webserver in itself that
gets polled. 2.45GHz often wouldn't work all that well because of range.

I would build it with modules. 433 MHz transceiver, 5 euro:

http://www.octamex.de/shop/?page=shop/flypage&product_id=1418

Ethernet module, 25 euro:

http://www.olimex.com/dev/enc28j60-h.html

I thinks this would be sufficient for the LAN port side. The schematic for
the ethernet module is free, so maybe you could create a better PCB for
mass production with the RF module.

For the sensor boards you can use any cheap microcontroller, which has
sufficient resources for your sensors and the RF module. And the RF module
has already an integrated temperature sensor.

The software would be a nice project for a weekend

 

[markp]

A weather base station with a serial port and this
<http://www.lantronix.com/device-networking/external-device-servers/uds1100.html>

Over $100

Cheers

There is a specification used by the likes of Jennic and others called
6LowPAN (http://en.wikipedia.org/wiki/6LoWPAN) which allows IP addressing of
radio nodes specifically to allow them to be accessed in ths way. There are
also free 6LowPAN stacks on sourcefourge. Jennic certainly do 2.4GHz radio
modules with this, and there may well now be some ISM modules with it.

The Holy Grail for me thouigh would be a very low power mesh network, such
as DigiMesh, using something like 6LowPAN to connect it to a LAN.

Mark.

 

[Joerg]

A weather base station with a serial port and this
<http://www.lantronix.com/device-networking/external-device-servers/uds1100.html>

Over $100

For just the port? That would be too much, and we'd still need a radio
module. Ideal would be one that has the radio part in the box.

 

[Joerg]

It is easy to make, a PIC 18F67J60, a 430 MHz module, magnetics, some programming,
maybe 10 $ in parts.

Sure, we can make it, but that's what we wanted to avoid. Just the radio
certs can become a pain in the neck. In most countries you must have a
blessing from an EMC lab before marketing it.

 

[Joerg]

There is a specification used by the likes of Jennic and others called
6LowPAN (http://en.wikipedia.org/wiki/6LoWPAN) which allows IP addressing of
radio nodes specifically to allow them to be accessed in ths way. There are
also free 6LowPAN stacks on sourcefourge. Jennic certainly do 2.4GHz radio
modules with this, and there may well now be some ISM modules with it.

The Holy Grail for me thouigh would be a very low power mesh network, such
as DigiMesh, using something like 6LowPAN to connect it to a LAN.

Thanks! Now I know the phrase for searching, LoWPAN.

 

[Joerg]

I would build it with modules. 433 MHz transceiver, 5 euro:

http://www.octamex.de/shop/?page=shop/flypage&product_id=1418

Ethernet module, 25 euro:

http://www.olimex.com/dev/enc28j60-h.html

I thinks this would be sufficient for the LAN port side. The schematic for
the ethernet module is free, so maybe you could create a better PCB for
mass production with the RF module.

One challenge would be that it has to work standalone on the LAN, so it
can be polled from the web even if no live computer is connected to this
LAN other than a DSL or cable modem. It would have to act as its own
little web server or at least auto-connect to a site on the web.

For the sensor boards you can use any cheap microcontroller, which has
sufficient resources for your sensors and the RF module. And the RF module
has already an integrated temperature sensor.

On the sensor board there's already a uC.

The software would be a nice project for a weekend

If this happens I'll tell the client to throw it over your fence. uC
programming ain't my cup of tea ;-)

 

[Frank Buss]

One challenge would be that it has to work standalone on the LAN, so it
can be polled from the web even if no live computer is connected to this
LAN other than a DSL or cable modem. It would have to act as its own
little web server or at least auto-connect to a site on the web.

Then maybe a bigger module would be a good idea:

http://www.olimex.com/dev/pic-mini-web.html

This is all you need on the LAN side, and a connection to the RF module,
which I assume has already all passed all the EMC requirements, so the
final device should pass the EMC tests.

If this happens I'll tell the client to throw it over your fence. uC
programming ain't my cup of tea ;-)

Thanks. I don't have much time, but if you do the hardware, it should be
not too difficult, because the hard programming parts are already done,
e.g. TCP/IP stack, web server and there are sample code for initializing
and transfering data with the RF module. So I just need to write some error
correction (there is already code for this part at
http://www.mikrocontroller.net/topic/67273 , if I want to do something
challinging and using Hamming error correction for such a low bandwidth
application) and a simple protocol with collision detection and maybe
channel hopping for querying the sensors.

 

[Nico Coesel]

Is there a cheap (as in mass produced and <$100) pod that plugs into a
LAN port of a regular router and have a, say, 433MHz radio link?
Something that could then connect to several thermometers or weather
stations and ask for their readings.

Thats on-route. When do you need it and in what quantities?

 

[Nico Coesel]

Thanks! Now I know the phrase for searching, LoWPAN.

We're working on similar devices for domestic use. LoWPAN seems to be
a pain in the ass to implement. We go for full IP including real
security (thats whats missing in other products).

 

[Frank Buss]

We're working on similar devices for domestic use. LoWPAN seems to be
a pain in the ass to implement. We go for full IP including real
security (thats whats missing in other products).

LoWPAN Seems to be a complicated protocol. This article:

http://edageek.com/2008/05/19/low-power-wireless/

says there is an implementation which needs a 32 bit microcontroller with
192 kB ROM and 96 kB RAM. Looks like the IEEE 802.15.4, which is the base
standard for it, is not easy to implement. Same if you would use something
like Bluetooth-Piconet.

If there is no interoperability requirement with other devices, would be
much easier to implement a protocol just for the task at hand, e.g. a
simple polling protocol with one master, which makes things even easier.

 

[Frank Buss]

Those modules are certified.
What else if no design?
Try shopping unlimited.

This doesn't matter. If you are assembling multiple modules to a new
device, you have to certify the whole device again (but chances are good,
that it passes the tests without problems, if the components passed the
test already).

 

[Joerg]

We don't need much security for the RF link, and only very low data
rates. Even 300bd would be just fine but must be bidirectional.

LoWPAN Seems to be a complicated protocol. This article:

http://edageek.com/2008/05/19/low-power-wireless/

says there is an implementation which needs a 32 bit microcontroller with
192 kB ROM and 96 kB RAM. Looks like the IEEE 802.15.4, which is the base
standard for it, is not easy to implement. Same if you would use something
like Bluetooth-Piconet.

Yikes. That sounds like Windows or MS-Office where you need at least a
gigabyte of RAM to write "Hello World".

If there is no interoperability requirement with other devices, would be
much easier to implement a protocol just for the task at hand, e.g. a
simple polling protocol with one master, which makes things even easier.

Looks like it :-(

It's strange, everyone is talking about ambient assisted living but it's
all just PowerPoint jockeying and talk, no real action.

 

[Joerg]

Then maybe a bigger module would be a good idea:

http://www.olimex.com/dev/pic-mini-web.html

Nice!


This is all you need on the LAN side, and a connection to the RF module,
which I assume has already all passed all the EMC requirements, so the
final device should pass the EMC tests.

Well, this client definitely does not want to do any sort of EMC test.
Has to be pre-certed and ready to go. Software writing and (very minor)
asembly work would be ok, but not a whole EMC cert session.

Thanks. I don't have much time, but if you do the hardware, it should be
not too difficult, because the hard programming parts are already done,
e.g. TCP/IP stack, web server and there are sample code for initializing
and transfering data with the RF module. So I just need to write some error
correction (there is already code for this part at
http://www.mikrocontroller.net/topic/67273 , if I want to do something
challinging and using Hamming error correction for such a low bandwidth
application) and a simple protocol with collision detection and maybe
channel hopping for querying the sensors.

Yes, it would need channel hopping because 433MHz is pretty crowded in
many metro areas. It's the only band that is international enough and
offers a few hundred meters range for really low bandwidths. 2.45GHz
doesn't work well. Initially I thought maybe some DECT device could be
used but I am rather disappointed with that standard.

 

[Joerg]

Thats on-route. When do you need it and in what quantities?

Realistically around next winter, so there is time. But it must be
something where we don't have to EMC-cert it for every market.

Quantities depend on user acceptance. I'd have to ask. Initially
probably a few hundred, later thousands per year. So it can't be an
elaborate full-custom solution, must be almost off-the-shelf because of
the low quantities. There could also be a huge initial demand if
existing systems are to be upgraded with this feature.

 

[Frank Buss]

Use 2 boxes.
Boxes are cheap.

I'm not an expert, because I'm just a programmer (and Ocarina player on
iPhone since yesterday

, but
at least the RF module needs its own PCB and I think as soon as you solder
some module to another PCB and put it in a box, you have to test it again.

 

[Frank Buss]

Yikes. That sounds like Windows or MS-Office where you need at least a
gigabyte of RAM to write "Hello World".

No, just 96 kB RAM But you are right, maybe too much for low cost mass
production.

It's strange, everyone is talking about ambient assisted living but it's
all just PowerPoint jockeying and talk, no real action.

LoWPAN doesn't sound bad and maybe is better, if you have multiple devices
from different vendors, but unversal multi layer protocols are complicated
and devices are more expensive.

 

[Joerg]

No, just 96 kB RAM But you are right, maybe too much for low cost mass
production.

This time it doesn't have to be super low cost. But at the same time it
shouldn't be over-engineered. This easily happens when standards
committees get together. What's the old saying? Too many cooks spoil the
broth.

LoWPAN doesn't sound bad and maybe is better, if you have multiple devices
from different vendors, but unversal multi layer protocols are complicated
and devices are more expensive.

Well, if there were off-the-shelf units at reasonable cost I wouldn't
care whether there is a 32bit ARM in there of some lowly 8051.

 

[Joerg]

I'm not an expert, because I'm just a programmer (and Ocarina player on
iPhone since yesterday

, but
at least the RF module needs its own PCB and I think as soon as you solder
some module to another PCB and put it in a box, you have to test it again.

Yes. Even mounting the antenna yourself can already put you into a legal
gray zone. Doing the box and antenna mount yourself definitely will.

For hobby projects all this may sound easy but when it's a consumer
product that's a whole different ballgame. Sometimes younger engineers
tell me "Oh man, it's so easy, you simply do this, that and the other
thing and you're done". Until they have to trudge through the paperwork
of EMC cert themselves for the first time and a check for a princely sum
is plopped onto the table at the compliance lab. After that they usually
never say this again