Really low bandwidth networking

This really has little to do with linux, except that the box it will run
on is linux-based....

I need to put sensors out on some equipment.... I can run a single cable
- say a few twisted pairs - but no more than that. I can't run a bundle
of cables - the maintenance folks nixed that quickly....

I have to collect 8 bits (1 byte) of data from 20 stations located along
this cable every minute or so. I need to know which station sent the data.

The cable is about 2,000' (600m) long. I'd like to have the absolute
minimum hardware out there - all I am doing is trying to report the
position of a few switches. I have poor quality 120 and 480 VAC power
available at each point....

Anyone know of a technology that can do this? We've looked at X-10, but
my previous experience with X10 indicates that it is not very robust when
running with poor quality power...

I will have a linux box collecting all of this data. Right now, I am open
to all suggestions....

Anyone have experience with modbus sensors? Anyone know if modbus sensors
would work?

--


use munged address above to email me
SpamTrap (E-Mail Removed)

No idea! programmable plc's doing simple serial comms
repeating/amplifying the incoming signal; each box with its own signiature?

Let us know what you come up with!

I suspect off the shelf youre gonna pay through your neck. hah, use 486
boxes with old lan cards!

Captain Dondo wrote:
> I need to put sensors out on some equipment.... I can run a single cable
> - say a few twisted pairs - but no more than that. I can't run a bundle
> of cables - the maintenance folks nixed that quickly....

> The cable is about 2,000' (600m) long. I'd like to have the absolute
> minimum hardware out there - all I am doing is trying to report the
> position of a few switches. I have poor quality 120 and 480 VAC power
> available at each point....

I think you should be more concerned that the maximum transmission
length for twisted-pair cable is about 100m. You'll need a repeater or
use another medium. You can get 185-500m with coaxial cable, and fibre
optic cable will work over 2 km.

-- shufler

shufler wrote:
> Captain Dondo wrote:
>
>> I need to put sensors out on some equipment.... I can run a single cable
>> - say a few twisted pairs - but no more than that. I can't run a bundle
>> of cables - the maintenance folks nixed that quickly....
>
>
>> The cable is about 2,000' (600m) long. I'd like to have the absolute
>> minimum hardware out there - all I am doing is trying to report the
>> position of a few switches. I have poor quality 120 and 480 VAC power
>> available at each point....
>
>
> I think you should be more concerned that the maximum transmission
> length for twisted-pair cable is about 100m. You'll need a repeater or
> use another medium. You can get 185-500m with coaxial cable, and fibre
> optic cable will work over 2 km.
>

This applies to twisted-pair Ethernet only.

There are other methods of communicating with widely different
requirements and maxima - think for instance about the POTS
(Plain Old Telephone Service).

For slow binary transmission, already Werner Siemens made a
telegraph link 500 km long (Frankfurt a.M. to Berlin).

--

Tauno Voipio
tauno voipio (at) iki fi

Coenraad Loubser wrote:
> No idea! programmable plc's doing simple serial comms
> repeating/amplifying the incoming signal; each box with its own signiature?
>
> Let us know what you come up with!

http://www.rs485.com/pat44r.html

Not too expensive, and with RS485 I can multi-drop and go 4,000'... Not
bad.

Now to get mgt to sign off on it so I can play with it....

;-)

Captain Dondo wrote:

> I need to put sensors out on some equipment....**I*can*run*a*single*cable
> - say a few twisted pairs - but no more than that.**I*can't*run*a*bundle
> of cables - the maintenance folks nixed that quickly....
>
> I have to collect 8 bits (1 byte) of data from 20 stations located along
> this cable every minute or so.**I*need*to*know*which*station*sent*the
> data.
>
> The cable is about 2,000' (600m) long.**I'd*like*to*have*the*absolute
> minimum hardware out there - all I am doing is trying to report the
> position of a few switches.**I*have*poor*quality*120*and*480*VAC*power
> available at each point....
>

Try RS-485. It's similar to RS-232, but uses balanced signalling and
multiple connections. It'll go up to 4000'. Here's some info.

http://www.rs485.com/rs485spec.html

shufler wrote:
> Captain Dondo wrote:
>
>> I need to put sensors out on some equipment.... I can run a single cable
>> - say a few twisted pairs - but no more than that. I can't run a bundle
>> of cables - the maintenance folks nixed that quickly....
>
>
>> The cable is about 2,000' (600m) long. I'd like to have the absolute
>> minimum hardware out there - all I am doing is trying to report the
>> position of a few switches. I have poor quality 120 and 480 VAC power
>> available at each point....
>
>
> I think you should be more concerned that the maximum transmission
> length for twisted-pair cable is about 100m. You'll need a repeater or
> use another medium. You can get 185-500m with coaxial cable, and fibre
> optic cable will work over 2 km.
>
> -- shufler

Where did you get these numbers ?????????????

It Might be true for high speed but NOT for his application
... read the Subject LOW BANDWIDTH

I'd suggest using a current loop with power supply at the receiving end
to over come any line losses

SEND NO SPAM wrote:

> I'd suggest using a current loop with power supply at the receiving end
> to over come any line losses

RS-485 is more than adequate for this job.

James Knott wrote:
> SEND NO SPAM wrote:
>
>
>>I'd suggest using a current loop with power supply at the receiving end
>>to over come any line losses
>
>
> RS-485 is more than adequate for this job.
>
>

Yes but he stated power avalible at the remote locations was
not very good. Using RS-485 you will need stable power at each
location ... special power conditioning at several locations.
Using a current loop eliminates that concern at the remote locations

James Knott wrote:
> SEND NO SPAM wrote:
>
>
>>I'd suggest using a current loop with power supply at the receiving end
>>to over come any line losses
>
>
> RS-485 is more than adequate for this job.
>
>

With those distances, the ground level differences will
easily grow to a problem. The common-mode tolerance of
RS-485 is not over 5 V, but the ground differences at
distances of hundreds of meters will easily be over
the limit.

At least, consider optical isolation at each station.

---

For long ( > 50 ft standard limit) links, the common
interface chips for RS-232 (MC1488, MC1489) were called
14-leg fuses, for a reason.

--

Tauno Voipio
tauno voipio (at) iki fi