Forums

RS485 is bidirectional does it mean it is fullduplex?

Started by Swizi June 15, 2005
"Floyd L. Davidson" <floyd@barrow.com> wrote in message 
news:87ekb1hmyy.fld@barrow.com...
> "Paul E. Bennett" <peb@amleth.demon.co.uk> wrote: >>Floyd L. Davidson wrote: >>>>> You *absolutely do* want a good ground connect at both ends. >> >>No, just the one end and preferrably the instrument rack/master end of the >>cable. In the rare circumstance that you cannot isolate the screen at the >>remote end put a break in the screen at a convenient point close to the >>remote end and make sure it stays broken. > > That is *not* true in the situation being discussed, where there > is a cable long enough to have the two ends connected to separate > ground systems and equipment supplied by separate power systems.
You keep saying this, but it's wrong. If you won't believe us, use Google to do some searching. A few seconds of searching turned up: http://www.hw-server.com/rs232_signals.html It's about RS-232, but the grounding issue under discussion is the same (except it's worse with RS-232, since ground differences are seen as signal). I'll quote a section:
>>
1 Protective Ground Name: AA Direction: - CCITT: 101 This pin is usually connected to the frame of one of the devices, either the DCE or the DTE, which is properly grounded. The sole purpose of this connection is to protect against accidental electric shock and usually this pin should not be tied to Signal Ground. This pin should connect the chassis (shields) of the two devices, but this connection is made only when connection of chassis grounds is safe (see ground loops below) and it is considered optional. Ground loops are low impedance closed electric loops composed from ground conductors. When two grounded devices are connected together, say by a RS-232 cable, the alternating current on the lines in the cable induces an electric potential across the ends of the grounding line (either Protective Ground or Signal Ground), and an electric current will flow across this line and through the ground. Since the loops impedance is low, this current can be quite high and easily burn out electric components. Electrical storms could also cause a burst of destructive current across such a loop. Therefore, connection of the Protective Ground pin is potentially hazardous. Furthermore, not all signal grounds are necessarily isolated from the chassis ground, and using a RS-232 interface, especially across a long distance, is unreliable and could be hazardous. 30 meters is considered the maximum distance at which the grounding signals can be connected safely. << I repeat: what you keep proposing makes no sense. I can only presume that you're confusing this issue with something else entirely. Consider what happens when lightning strikes one location - local ground potentials go *nuts*. Your (apparent) connection between two local ground systems will try to connect them together, and it will lose. Bigtime. You've mentioned telephone system cabling: I'm no expert, but I had thought telephone cables were current-driven loops, with the power supplied by a -48V supply and a local ground *at one end*. Perhaps this is only the subscriber loop. For *very* long trunk cable runs, I have seen what you describe - shield connected to local ground via a dirty great grounding pole - but I had understood that the series impedance of the shield was quite high, which is what saved it from becoming a very long and expensive fuse. This is not what we're discussing here (the need for all RS-485 signals to be within a defined common-mode range). Steve http://www.fivetrees.com
Grant Edwards <grante@visi.com> wrote:
>On 2005-06-16, Floyd L. Davidson <floyd@barrow.com> wrote: > >>>The RS-485 systems I'm talking about are all optically isolated >>>from frame, chassis, and earth. If you don't connect the RS-485 >>>commons together with the cable, then you end up with >>>common-mode voltages out of spec. Study all you want, that's >>>what happens in practice. >> >> What you are doing will result in equalizing the common mode DC >> offset from different grounds. > >Except that none of the RS-485 circuits were grounded. They >were all optically isolated and the RS-485 transmitter and >receiver grounds were floating.
I'm confused about what you actually have, see below.
>No, the opto-isolators were between the RS-485 >transmitters/receivers and the rest of the gear. The Rs-485 >transmitters and receives were galvanically isolated from >earth, chassis, and supply ground. Our experience was that >connecting two "floating" RS-485 interfaces together without a >common (connecting just the data lines) resulted in a lot of >problems.
I wasn't getting a clear picture of this yet... and I'm not sure that I am yet! :-) EQUIP ISOLATOR RS485 <===/ CABLE /===> RS485 ISOLATOR EQUIP If you are doing that, with 10 km of cable and *not* grounding the cable shield, it may well work, but it isn't the best way to do it.. You don't need to use a cable pair to connect the common mode grounds on the RS485 drivers, because the shield should have a separate ground strap going to a building ground at each end. The RS485 common mode ground points should also be connected to those same building grounds, with a separate cable. EQUIP ISOLATOR RS485 <===/ CABLE /===> RS485 ISOLATOR EQUIP | | | | \ / \ / | | <-- COMMON ----- ----- / / / / / / The part marked as COMMON must be sized appropriately to avoid ohmic losses from current in the cable shield from affecting the RS485 drivers. Generally that means either a very large cable, or a very short distance.
>> On the other side, they are *all* connected to a common >> ground, if they are properly engineered.
-- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
"Floyd L. Davidson" wrote:
>
... snip ...
> > The other part is just not being exposed to the full expanse of > what is involved in data transmission over longer lengths of > twisted pair cables.
Yet use of the old fashioned 20 ma current loop, driven through photo diodes/transistors, is virtually immune to these problems. It is biased in only one place. If you are going to inject signal at every station the interface is harder. For point to point or broadcast you can hardly beat it. -- Chuck F (cbfalconer@yahoo.com) (cbfalconer@worldnet.att.net) Available for consulting/temporary embedded and systems. <http://cbfalconer.home.att.net> USE worldnet address!
"Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:
>"Floyd L. Davidson" <floyd@barrow.com> wrote in message >news:873brik81n.fld@barrow.com... >> And absolutely unnecessary too. RS-422 works on two pairs. The >> master transmitter and all slave receivers are on one, and the >> master receiver and all slave transmitters are on the other. >> They are *all* high impedance devices, relatively. There has to >> be a 100 Ohm load resistor on each pair. > >Errr... no, the 100R is there for transmission line matching. It should >match the characteristic impedance of the cable.
So? You seem to miss the fact that its DC characteristics are just as important a its impedance matching function. That's the same as the 50 Ohm termination on a 10base5 ethernet. Ever try using RG59 cable instead of RG58? It works... as long as you stick with 50 Ohm terminations. Same point... The fact is the 100 Ohm load resistor, being of lower resistance than any other load on the cable, is what determines the voltage levels. The receivers are high impedance voltage sensing devices. -- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
"Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:
>"Floyd L. Davidson" <floyd@barrow.com> wrote in message >news:87d5qmi3i3.fld@barrow.com... >>>So, I repeat: one *has* to consider common-mode. That 3rd or 5th wire must >>>be there, whether implicitly (via a common local ground) or explicitly >>>(via >>>a physical cable). >> >> You haven't caught the significance of what I've been saying either. >> >> Your "3rd or 5th wire" is *not* the way to deal with ground potential >> differences. What has been described causes more problems than it >> cures. > >I'm afraid I'm not just not catching you, I'm not following you at all. >Unless I'm missing something, you seem to be suggesting dangerous practices, >e.g. connecting a long cable to local ground at both ends. At that point I >started skipping your posts...
I've explained that *in detail*. That *is* the way telephone cable systems are installed.
>As I've just said elsewhere, this is simply nonsense.
And you are out of your field...
>Consider: you're >trying to short out a part of the mains utility distribution system, which >may be carrying significant leakage currents over long distances, with a >flimsy little signal wire. Ground potential differences can be significant, >and the source impedance is very low indeed - certainly capable of >delivering tens of amps. I've seen guys who should know better staring at >melted cables, scratching their heads...
If they had a melted cable, it wasn't simply the difference in ground potential. (Telecom cables generally are spliced and grounded every 6000 feet, max.)
>Please see my (and others') post(s) re isolated comms interfaces.
See the tutorial I posted: Message-ID: <87vf4egfri.fld@barrow.com>
>Use one >ground; isolate from the other - i.e. keep both Tx and Rx relative to one >ground.
That is fine within one building, and is not the best practice for longer cable runs, for reasons explained in the message listed above. -- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
"Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:
>"Floyd L. Davidson" <floyd@barrow.com> wrote in message >news:874qbyi2cb.fld@barrow.com... >>>> And absolutely unnecessary too. RS-422 works on two pairs. The >>>> master transmitter and all slave receivers are on one, and the >>>> master receiver and all slave transmitters are on the other. >>> >>>And all those slave transmitters should be disabled, except the one that's >>>talking to the master. >> >> They are high impedance devices. The only "disable" they need, is >> to not be sending. > >You've just confirmed my suspicions that you have no clue what you're >talking about. RS-422 and RS-485 drivers are low-impedance in both MARK and >SPACE state, and do indeed have a third (tristate) condition.
Impedance is relative. To an 8 Ohm speaker, 600 Ohms is high impedance. -- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
"Steve at fivetrees" <steve@NOSPAMTAfivetrees.com> wrote:
>"Paul E. Bennett" <peb@amleth.demon.co.uk> wrote in message >news:d8sihj$ffk$1$830fa7a5@news.demon.co.uk... >> >> By way of re-inforcing the point here, consider that the screen is only >> meant to act as the notional extension of a metal enclosure out along the >> wires. It should not carry any current at all (except maybe for the >> tiniest >> leakage current capacitively coupled from the signal wires - and even that >> should be miniscule). Dealing properly with the screens is a safety issue >> as well as a circuit protection and noise reduction issue. > >Absolutely. Connecting to local ground at both ends of the run is a recipe >for HUGE ground currents. I've seen (very) melted ground wires. It is indeed >a safety hazard.
Bullshit.
>See my other post re isolated comms interfaces.
Why not do some real research, and find out how it is actually done. -- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
On 2005-06-17, Floyd L. Davidson <floyd@barrow.com> wrote:

> EQUIP ISOLATOR RS485 <===/ CABLE /===> RS485 ISOLATOR EQUIP > > If you are doing that, with 10 km of cable
I don't think I ever saw runs longer than 2-3km.
> and *not* grounding the cable shield, it may well work,
The cable shield was grounded. What wasn't grounded was the RS-4xx driver/transceivers. ----+ +------+ +------------/ /-- Shield | | Opto | +-----+ \ uP|-Tx--| Iso |--Tx--| 485 |--|---A-----/ /-- |-Rx--| |--Rx--| xcvr| | |-dir-| |-dir--| |--|---B-----/ /-- Mirror -++-+ | | +-++--+ | Image |`--Pwr--| Pwr |-IsoPwr-'| | `---Gnd--| Iso |-RScom---+-----|---com---/ /-- | +------+ / /// +--+--------/ /-- | /// Or something pretty close to that. In some pieces of gear the uP was floating (uP ground was RS485 common). In all cases the RS-485 interface was floating with respect to earth/chassis ground. I don't remember if people were told to ground the shield at one end or both. There were A, B, common, and chassis ground lugs on each of our bits of gear (in addition to some other stuff unrelated to the discussion).
> You don't need to use a cable pair to connect the common mode > grounds on the RS485 drivers, because the shield should have > a separate ground strap going to a building ground at each end. > The RS485 common mode ground points should also be connected > to those same building grounds, with a separate cable.
That's sure not the way I remember it in the process control systems I worked with (it's been 6-7 years). The RS-485 bus was pretty much always floating w/ respect to ground, with A, B and a floating common wire between the two RS485 transceivers. There may have been people that grounded the RS-485 common node at some point, but it was expected to work if it was floating. -- Grant Edwards grante Yow! Psychoanalysis?? I at thought this was a nude visi.com rap session!!!
"Paul E. Bennett" <peb@amleth.demon.co.uk> wrote:
>Floyd L. Davidson wrote: > >> However, if the cable is a long run, and particularly if there >> is exposure to power lines, if the ground potential is different >> at the two ends, or if there are any other sources of induced >> noise in the cable, this arrangement has the best effect: >> >> +-------+ +-------+ >> | | >--------- tx wire/pair ---------> | | >> | EQUIP | <--------- rx wire/pair ---------< | EQUIP | >> | | ========= cable shield ========= | | >> +-------+ | | +-------+ >> | | | | >> | | | | >> o------+ +------o >> | | >> ----- Earth ----- Earth >> --- Ground --- Ground > >If you really do need to connect at both ends then you may need to consider >inserting some impedance in the screen connections at both ends (usually a >capacitor and resistor in parallel). The resistor is large enough to >prevent high current flows but needs to be small enough to provide an >effective electrostatic drain. The capacitor provides a low impedance at >higher frequencies.
No, the whole idea is that you *want* that current to flow. In particular it is the 60 Hz power line induced current that makes up most of the current flow. Keep in mind that the whole idea is to allow the current flow to generate an equal and opposite induction into the signal pairs. -- Floyd L. Davidson <http://web.newsguy.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd@barrow.com
"Floyd L. Davidson" <floyd@barrow.com> wrote in message 
news:87is0dg2oe.fld@barrow.com...
> >>As I've just said elsewhere, this is simply nonsense. > > And you are out of your field...
I've pointed out several factual errors with your posts. I've demonstrated several areas where your understanding of RS-422/485 is, at best, incomplete, and at worst, downright wrong. I (and many others) have provided you with ample details of why what you're saying is just plain wrong. "Out of your field", you say? To be real clear about all this: I'm sure you're sincere, but you sound confused. RS-422/485, and its practical applications, happens to be an area I know *real* well. I'm a bit embarrassed for you. Steve http://www.fivetrees.com