RS485 connector 5

[RyreInc]

I'm looking for a plug and receptacle to use with RS-485 serial communication. From what I can tell there does not appear to be a standard connector. RS-485 half-duplex cable consists of a single twisted-pair w/ optional shield - I would prefer to include the shielding. The connector could be located outside an enclosure in an industrial environment, so it should have some sort of mechanical retention as well as cable strain relief. A DB9 connector could probably work but is larger than what I'm hoping to find, plus I don't like the retention method. It would be nice for us to be able to make our own cordsets so we can customize the wire length. It would also be nice to avoid soldering.

Anything come to mind?
Thanks!

 

[ScottyUK]

MIL-C-5015 connector? I like the convenience of the bayonet type, and they're pretty robust

 

[danw2]

No advice on connector, but I have run into common mode issues with earth grounds at different potentials with RS-485 comm, where the cabling was 2 wire TSP and the 485 signal ground became earth ground connection.

If the 485 vendors offer a signal ground connection, my advice is to use it with a 3rd conductor. But, a lot of 485 vendors make no provision for a signal ground.

Here's what Robust Data has to say:

Figure 1:

 

[ScottyUK]

I agree with Dan - we had a lot of problems with RS-485 on a long distance multi-drop bus within a power plant environment caused by noise in the earths. In the end we installed galvanic isolators at each drop and earthed the screen at the source end only. We didn't have long Tee's off the bus as shown in the second diagram though - it was wired on a loop-in, loop-out basis at each drop. I suspect this is only a significant problem on very large installations or as you push the Baud rate up.

 

[catserveng]

For connectors I am a fan of Deutsch connectors, a lot of OEM's use them and there are now a number of vendors who carry a good selection at reasonable prices, at least in the US.

The DT series is used quite a bit by CAT and others, and offers a lot of options, here is a source,

http://www.deutschconnector.com/products/deutsch_connectors/deutsch_dt_series_connectors/

Amphenol now makes a compatible series of connectors called the AT series,

https://www.amphenol-sine.com/at-series

Hope that helps, MikeL.

 

[JG2828]

You could use RJ45 connectors like used on this Comtrol 32-port 1 rack unit serial port server.

https://comtrol.com/ethernet-device-servers/devicemaster-rack-mount/devicemaster-rts-32port-rj45

 

[IRstuff]

there are subminiature versions of the DB9 connector; they're about 1/4 the size of the standard sub-D connector:

http://www.airelectro.com/manufacturers/glenair/mil-dtl-83513-micro-d/3-4-micro-d?ref=x

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[ScottyUK]

If size is a consideration, Lemo make some very nice miniature connectors. Top quality. Price to reflect the quality!

 

[RyreInc]

Thanks for the feedback everyone. Some good connector solutions, as well as some interesting info on common mode voltages. I need to look further into each of these yet.

I had been considering an RJ11 type connector (we want to avoid RJ45 since Ethernet will be present as well), but I had planned to use a shielded twisted pair cable and it seems like it would be tricky to crimp the drain wire. Maybe this is still the best option? Maybe I could tin the drain wire before crimping? What do you think?

For this application the nodes will be pretty close - within a few meters of each other - and will be powered from a single power source, so I don't think common mode voltage will be an issue. Nevertheless I will look into this more since I should understand this. The concept is counter-intuitive to me since 485 is differential: adding a ground wire would seem to defeat that, even if the ground wire is a twisted pair or a shield...

 

[IRstuff]

"485 is differential"

As always, "Yes, but..."

> No 485 circuit has infinite common mode rejection
> Transients can occur between circuit partitions, particularly if powered separately. Without grounding, power-on transients can make short work of frying the 485 inputs, particularly if they're bipolar.
> The ground actually needs to be rock solid, to suck up differential currents from differing ground potentials, particularly during power-on conditions.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[LionelHutz]

485 is a differential signal, as in the signal is the difference between 2 wires, but the circuity is not differential. The signal wires are not in a loop. Each of the signal conductors is referenced to the same common. The actual circuitry, usually inside an IC, has circuits that produce 2 output signals with respect to the common pin of the IC. Sometimes, the common pin of the IC is connected to earth ground, other times the common (and whole circuit power supply) is floating and the common is brought to a terminal the custom needs to connect. Floating ports are more desirable with a dedicated wire connecting the commons so that local earth noise doesn't affect the communications.

As for a terminal. I would just use a terminal strip. The problem with connectors is that you can't daisy chain the wires. With a connector like RJ11, you have no easy way to daisy chain so you do stubs which then means you will likely be installing a terminal block close by anyways. So, might as well put a small terminal block that can allow the daisy chain connection right at the device which reduces the number of connections and makes wiring simpler.

If you don' use terminals, then using a RJ45 would be my next choice since I do believe there is a standard for the pin layout in that connector.

 

[JG2828]

LionelHutz makes a good point above. Are you running a single cable between each device and a port server, or are you daisy-chaining all the devices?

Either way, this PDF does a good job at explaining wiring of RS-485 devices:

http://www.bobtech.ro/get?download=21:rs485-wiring-guidelines

 

[RyreInc]

Very interesting. LionelHutz's statement that while the signals are differential but the circuitry is not is key for me: I was thinking the circuitry was also differential. So that makes a lot more sense now!

In this application 485 is used to daisy chain between a parent and multiple child controllers. Each child controller is powered from 24VDC, whose common/0V is typically tied to ground at the power supply. The RS485 chip (ISL8485EIPZ) on these boards is powered from an onboard 5V non-isolated DCC converter, so it shares that same common. I do not want to create a ground loop by having two grounds go to the board. But I also cannot guarantee that each child will be powered from the same source (although that SHOULD always be the case...). It seems like galvanic isolation is the only real answer to this conflict. Would you agree that isolating the 5V board power is preferable to isolating the 485 signals?

Regarding a connector vs. terminal blocks: The plan is to provide a pair of connectors to allow daisy chaining, so e.g. two RJ11s per node - I think that solves the daisy chaining issue.

Regarding RJ11 vs RJ45: While I definitely see the merit in using a standard pinout with RJ45, again there will be Ethernet present and we want pokeyoked connectors. Also an important detail I haven't mentioned is that this is a closed system; 3rd parties will not be adding devices to this network, so I think that weakens the standard pinout argument.

 

[itsmoked]

I like Linoel's description LPS.

There are lots of 485 isolaters out there. If you fear ground loops put a few into the system to break things up.

And I'd not want to incur the wrath of future controls people blue-air by using some crimped connector.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[RyreInc]

itsmoked, can you please elaborate on this? Are you just saying crimped connectors are bad generally? Also I'm not familiar with "blue-air"...

And I'd not want to incur the wrath of future controls people blue-air by using some crimped connector.

 

[itsmoked]

blue-air = Cussing

In the future people will have to service this field of 485 instruments, moving them, adding to them, and replacing them. Having to understand and acquire odd -for RS485- connectors needing special crimping tools or soldering will make that service considerably slower and more expensive. Faced with that, I'd be 'turning the air blue'.

485 in field applications should remain screws, either cage-clamps, spring clamps, or standard old terminal blocks. If they need weather protection put them in small boxes with bottom drip loop entries. This way anything can be used for wire, terminations can be added, taps can be accommodated, isolation can inserted, wires can be disconnected for testing, and things can be probed with meters or scopes - none of which can be done with crimped termination like an RJanything.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[LionelHutz]

I still prefer terminal blocks or a standard connector layout. Otherwise, not knowing the connector layout adds yet another issue to any future troubleshooting or re-working.

The communications would be more robust if you can change to an isolated 5V converter and put galvanic isolation between the RS485 chip and the rest of the circuit. I'd highly recommend doing this if the board is still in the design stage.

 

[RyreInc]

Thanks everyone for your feedback, some great food for thought!