Hello Stef,

Was on a biz trip, this one w/o NG access.

> It is a separate 24VDC powerline, but it has up to 100 DC motors on
> it and there are sliding contacts. So it will be noisy. 

That is noisy but you can slide ferrite beads and stuff over the supply 
leads.

> Would you advise RF in a noisy environment? That does not seem logical
> to me. Can these phone tranceivers handle up to 100 simultanious
> connections or would I need to switch on one at the time? If one at
> the time, how fast can I expect to scan all nodes?

Above a GHz you won't see much of the DC motor noise anymore. 
Simultaneous xmit/rec isn't feasible since that would usually require 
100 modules. Polling could be done in milliseconds per node. You'd have 
to talk with the folks from Nordic or another manufacturer to find the 
most cost efficient modules for your requirements.

It's the same with all other protocols, even powerline. You can only 
poll one at a time or you would need 100 carriers and receivers. 
Theoretically you could send out on various frequencies and have a DSP 
figure out all those carriers. But at a hundred that would be a huge DSP 
job.

>>Also, think about IR for simplicity. If pointed upwards to a diffuse 
> 
> Yes, IR is on the list of candidates.

And it may be the most cost efficient candidate if you can make it work.

Regards, Joerg

http://www.analogconsultants.com

In comp.arch.embedded,
Joerg <notthisjoergsch@removethispacbell.net> wrote:
>Hello Stef,
>
>>>2400-9600 bps per node? I think that won't fly via powerline, at least 
>>>not in a noisy industrial environment.
>> 
>> No, not per node, in total. Most commands are broadcast, like: "All nodes
>> perform function A". Then there is some configuration and status checking
>> that needs to be done per node, but this can be relatively slow and uses
>> a request/reply protocol.
>
>Then it could be done via the power line but in a noisy environment I 
>still would not do that. You could send all this via a simple RF link. 
>For example, assume you had a cordless phone on the carriage and another 
>one on the floor. Via a modem you could now send all that data. This may 
>be a bit over-simplified and whether or not a spread spectrum phone 
>would work would have to be checked. You could use one of the Texas 
>Instruments or Nordic transceivers.
>
It is a separate 24VDC powerline, but it has up to 100 DC motors on
it and there are sliding contacts. So it will be noisy. 

Would you advise RF in a noisy environment? That does not seem logical
to me. Can these phone tranceivers handle up to 100 simultanious
connections or would I need to switch on one at the time? If one at
the time, how fast can I expect to scan all nodes?

>Also, think about IR for simplicity. If pointed upwards to a diffuse 
Yes, IR is on the list of candidates.

-- 
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

What foods these morsels be!

Hello Stef,

>>2400-9600 bps per node? I think that won't fly via powerline, at least 
>>not in a noisy industrial environment.
> 
> No, not per node, in total. Most commands are broadcast, like: "All nodes
> perform function A". Then there is some configuration and status checking
> that needs to be done per node, but this can be relatively slow and uses
> a request/reply protocol.

Then it could be done via the power line but in a noisy environment I 
still would not do that. You could send all this via a simple RF link. 
For example, assume you had a cordless phone on the carriage and another 
one on the floor. Via a modem you could now send all that data. This may 
be a bit over-simplified and whether or not a spread spectrum phone 
would work would have to be checked. You could use one of the Texas 
Instruments or Nordic transceivers.

Also, think about IR for simplicity. If pointed upwards to a diffuse 
reflector it could work. It's a matter of power and efficiency. Once in 
a hotel I could not believe it when the remote for the TV would still 
work when held at the ceiling instead of the TV. So I went into the 
bathroom and pointed it at the hallway wall. The IR had to bounce at 
least three times to get to the TV. It still worked.

Regards, Joerg

http://www.analogconsultants.com

In comp.arch.embedded,
Joerg <notthisjoergsch@removethispacbell.net> wrote:
>Hello Stef,
>
>> Required datarate is 2400 - 9600 bps. One of the operations is a sync
>> cycle that is currently handled by a separate hardwired signal to all
>> modules. The response here is in the us range and this is obviously not
>> possible through the powerline comms. I can compensate for a longer time
>> as long as the delay is fixed. I might be able to relief the jitter
>> requirement by changing/elaborating the sync cycle if needed.
>
>2400-9600 bps per node? I think that won't fly via powerline, at least 
>not in a noisy industrial environment.
>
No, not per node, in total. Most commands are broadcast, like: "All nodes
perform function A". Then there is some configuration and status checking
that needs to be done per node, but this can be relatively slow and uses
a request/reply protocol.

-- 
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

I have a very small mind and must live with it.
		-- E. Dijkstra

Hello Stef,

> Each rail (there are 2) is about 2 metres long, the ends are close together
> and from there, there is 2-3 metres of cable to the control cabinet. If
> needed, the powerline interface could be mounted close to the rails, but
> the preferred location would be in the cabinet.
> 
> Required datarate is 2400 - 9600 bps. One of the operations is a sync
> cycle that is currently handled by a separate hardwired signal to all
> modules. The response here is in the us range and this is obviously not
> possible through the powerline comms. I can compensate for a longer time
> as long as the delay is fixed. I might be able to relief the jitter
> requirement by changing/elaborating the sync cycle if needed.

2400-9600 bps per node? I think that won't fly via powerline, at least 
not in a noisy industrial environment.

I would consider an ISM band transmitter for the (hopefully few) signals 
that require microsecond response times.

> Most other commands require a response within 10 - 100ms. All commands
> are only a few bytes long (2 - 5).

With that much allowed latency you could consider WLAN for these 
channels. It provides lots of bandwidth for your data rates. Then there 
is Bluetooth, spread spectrum such as used by upscale cordless phones 
etc. Best is to stick with something that has already undergone agency 
certification for other uses. Otherwise this project could become quite 
expensive.

Regards, Joerg

http://www.analogconsultants.com

In comp.arch.embedded,
Rene Tschaggelar <none@none.net> wrote:
>Stef wrote:
>
>> Hello NG,
>> 
>> In a machine we currently use up to 100 nodes (most installations use under
>> 64 though) on a CAN bus. These nodes are moving along rails and are
>> connected with wire loops. For ease of installation (and node replacement)
>> and cleaning up the wire tangle, we are investigating a wireless or power
>> line system. The 24V DC power can be supplied with sliding contacts but
>> due to space restrictions it is not easy to add more contacts for
>> communications.
>> 
>
>Powerline communication on a closed system is doable. FSK modulation
>in the 100kHz range gives excellent results also on a bigger systems.
>
>You should come up with some information regarding the system size in
>mete r& squaremeter, the required datarate and the responsetime.
>
Each rail (there are 2) is about 2 metres long, the ends are close together
and from there, there is 2-3 metres of cable to the control cabinet. If
needed, the powerline interface could be mounted close to the rails, but
the preferred location would be in the cabinet.

Required datarate is 2400 - 9600 bps. One of the operations is a sync
cycle that is currently handled by a separate hardwired signal to all
modules. The response here is in the us range and this is obviously not
possible through the powerline comms. I can compensate for a longer time
as long as the delay is fixed. I might be able to relief the jitter
requirement by changing/elaborating the sync cycle if needed.

Most other commands require a response within 10 - 100ms. All commands
are only a few bytes long (2 - 5).


-- 
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

Women are just like men, only different.

Stef wrote:

> Hello NG,
> 
> In a machine we currently use up to 100 nodes (most installations use under
> 64 though) on a CAN bus. These nodes are moving along rails and are
> connected with wire loops. For ease of installation (and node replacement)
> and cleaning up the wire tangle, we are investigating a wireless or power
> line system. The 24V DC power can be supplied with sliding contacts but
> due to space restrictions it is not easy to add more contacts for
> communications.
> 
> Most of the communication is broadcast with a few bytes command every
> second or so. But we do need to be able to read back the status from each
> node once in a while. So we need some sort of half duplex transmission
> channel. speed need not be very high, but we do need a fixed response time
> to some commands. We are not stuck with CAN for communication with the
> nodes, as long as we have half-duplex we're OK.
> 
> At first glance, some sort of powerline communication should be possible.
> Any experience here with such systems?
> 
> Another option would be wireless HF such as zigbee or bluetooth, but I
> don't know if these systems will work with such a high node count. The
> comms are half duplex so a single channel would be enough?
> 
> Or maybe infrared could be an option.
> 
> The maximum distance is only a few meters. In an ideal world, the comm
> interface would be an add-on for the existing nodes (serial and CAN
> available).
> 

Powerline communication on a closed system is doable. FSK modulation
in the 100kHz range gives excellent results also on a bigger systems.

You should come up with some information regarding the system size in
mete r& squaremeter, the required datarate and the responsetime.

Rene
-- 
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

Hello NG,

In a machine we currently use up to 100 nodes (most installations use under
64 though) on a CAN bus. These nodes are moving along rails and are
connected with wire loops. For ease of installation (and node replacement)
and cleaning up the wire tangle, we are investigating a wireless or power
line system. The 24V DC power can be supplied with sliding contacts but
due to space restrictions it is not easy to add more contacts for
communications.

Most of the communication is broadcast with a few bytes command every
second or so. But we do need to be able to read back the status from each
node once in a while. So we need some sort of half duplex transmission
channel. speed need not be very high, but we do need a fixed response time
to some commands. We are not stuck with CAN for communication with the
nodes, as long as we have half-duplex we're OK.

At first glance, some sort of powerline communication should be possible.
Any experience here with such systems?

Another option would be wireless HF such as zigbee or bluetooth, but I
don't know if these systems will work with such a high node count. The
comms are half duplex so a single channel would be enough?

Or maybe infrared could be an option.

The maximum distance is only a few meters. In an ideal world, the comm
interface would be an add-on for the existing nodes (serial and CAN
available).

-- 
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail)

I use technology in order to hate it more properly.
		-- Nam June Paik