Prev: Popular Science magazine have made 137 years of back issues available for free on the net.
Next: USB Host
From: djordj on 9 Mar 2010 07:02
After looking at several application notes (by National, TI, B&B
Electronic...) I can't understand if I can bias the two-wire bus
without inserting a 120 ohm resistor between A and B cables.
Reading B&B Electronics (http://www.dzjsw.com/shu/bz22.pdf) it seems
to be possible (pages 18 and 19).
Reading National (http://www.national.com/an/AN/AN-1057.pdf) not
(pages 8 and 9).

My design doesn't need terminators, as total cable length will be less
than 4 meters, but I have to guarantee fail-safe conditions in bus
idle state.

Thanks
From: Oliver Betz on 9 Mar 2010 09:18
djordj wrote:

>After looking at several application notes (by National, TI, B&B
>Electronic...) I can't understand if I can bias the two-wire bus
>without inserting a 120 ohm resistor between A and B cables.

you can.

>Reading B&B Electronics (http://www.dzjsw.com/shu/bz22.pdf) it seems
>to be possible (pages 18 and 19).
>Reading National (http://www.national.com/an/AN/AN-1057.pdf) not
>(pages 8 and 9).
>
>My design doesn't need terminators, as total cable length will be less
>than 4 meters, but I have to guarantee fail-safe conditions in bus
>idle state.

If you don't rely on the resistors in the transceivers, use one
resistor to 0V, one resistor to Vbias (e.g.5V).

Depending on your speed requirements, you might want to add one
resistor between the signal lines to speed up an eventual transition
from "space" to "idle" (if you can't rule out this possibility).

Oliver
--
Oliver Betz, Munich
despammed.com might be broken, use Reply-To:
From: djordj on 9 Mar 2010 09:25
On Mar 9, 3:18 pm, Oliver Betz <ob...(a)despammed.com> wrote:
> If you don't rely on the resistors in the transceivers, use one
> resistor to 0V, one resistor to Vbias (e.g.5V).
This exactly is the solution I'd like to use, as we don't want to
constrain the system builder to insert the final terminator (this
would be a dangerous option left to the final user........ sadly...).
So, only pull-up and pull-down: nothing else.

> Depending on your speed requirements, you might want to add one
> resistor between the signal lines to speed up an eventual transition
> from "space" to "idle" (if you can't rule out this possibility).

Well, because of underlying hardware (CPUs) the bus will run at a
relative low speed (115200 bps), so I don't think that we'll have the
need to speed-up "idle"-"active"-"idle".
Another question: this optional resistor doesn't need to be balanced
by a terminator?

Thanks!
From: Stef on 9 Mar 2010 09:50
In comp.arch.embedded,
djordj <grgmeda(a)gmail.com> wrote:
> On Mar 9, 3:18�pm, Oliver Betz <ob...(a)despammed.com> wrote:
>> If you don't rely on the resistors in the transceivers, use one
>> resistor to 0V, one resistor to Vbias (e.g.5V).
> This exactly is the solution I'd like to use, as we don't want to
> constrain the system builder to insert the final terminator (this
> would be a dangerous option left to the final user........ sadly...).
> So, only pull-up and pull-down: nothing else.

And where do you want to place these resistors? At every node, or
do you have a single 'master' node where you can put them?

There are also 'fail safe' RS485 tranceivers, but the fail safe
does rely on the differential signal being close to '0' -> terminator.

>> Depending on your speed requirements, you might want to add one
>> resistor between the signal lines to speed up an eventual transition
>> from "space" to "idle" (if you can't rule out this possibility).
>
> Well, because of underlying hardware (CPUs) the bus will run at a
> relative low speed (115200 bps), so I don't think that we'll have the
> need to speed-up "idle"-"active"-"idle".
> Another question: this optional resistor doesn't need to be balanced
> by a terminator?

And in which electrically clean environment will this be used?

In noisy environments, I would really add terminators. Educate your
customers/system builders. I've seen RS485 fail with 'just' 115k2 and
'only' a few meters of cable. Adding terminators can really make a
difference. And make sure you use nice, twisted, cable.

If you let others wire up the systems, document how it should be done.
Type of cable, lengths, stub length, terminator placement, etc.


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

It was all so different before everything changed.
From: djordj on 9 Mar 2010 10:48
On Mar 9, 3:50 pm, Stef <stef...(a)yahooI-N-V-A-L-I-D.com.invalid>
wrote:
> And where do you want to place these resistors? At every node, or
> do you have a single 'master' node where you can put them?
We have a master board, where bias resistors will be placed.

> And in which electrically clean environment will this be used?
It's a closed body machine, with no motors but some static relays, EVs
and heaters.

> In noisy environments, I would really add terminators. Educate your
> customers/system builders. I've seen RS485 fail with 'just' 115k2 and
> 'only' a few meters of cable. Adding terminators can really make a
> difference. And make sure you use nice, twisted, cable.
>
> If you let others wire up the systems, document how it should be done.
> Type of cable, lengths, stub length, terminator placement, etc.
I usually follow those rules: luckily boards will be cabled using
predetermined connectors, so the machine builder has only to follow
connectors color scheme in order to get the right cabling.
There will be non "visible" stubs, as RS485 wires will be routed to
transceivers really close to the connectors, via PCB tracks.

The only problem is the terminator placement: the first can be factory
placed on master board, but not the second, as the machine can be
built of two to four slaves devices.
We have to hope that the machine builder will not forget to close that
holy jumper... :)
 |  Next  |  Last
Pages: 1 2
Prev: Popular Science magazine have made 137 years of back issues available for free on the net.
Next: USB Host