7
\$\begingroup\$

I would like to ask something about RS485 minimum node spacing which is described in (Page 5)

TI Design guide

I would like to create device which consists from 3 existing RS485 devices. My plan was that I put these 3 devices together on single 2 layer PCB and connect them via single RS485 line (A, B). According to material above this shouldn´t be a good idea because there will be very small length traces between these devices.

Do you have any experience with this approach of placing more RS485 devices on same board? Is there any technique which could help the signal integrity?

Thanks

\$\endgroup\$
  • 1
    \$\begingroup\$ Take a look at same question for CAN network. Since we are talking about distributed capacitance, it might be possible to compensate for it by adjusting trace impedance by changing trace geometry, as discussed in that question. \$\endgroup\$ – Maple Aug 25 '18 at 20:22
  • 1
    \$\begingroup\$ Should just work. Maybe issues may arise when all transmit, which may happen if protocol allows it. With long cable no one knows, when all are together.. Maybe they can heat each other. But generally just don't worry, try it. \$\endgroup\$ – Gregory Kornblum Sep 26 '18 at 9:57
  • \$\begingroup\$ At speeds under 10MHz it should always work fine. With slower speeds you only need one set of termination as well at any point on the bus. If you want to go crazy look up differential ECL and successors for what is used in the trade. :-) \$\endgroup\$ – KalleMP Sep 26 '18 at 11:15
0
\$\begingroup\$

RS-485 is a differential bus. This means it is very noise immune. There are two lines A and B. You need to bias the bus so the (A-B) is around -200mV when the bus is idle. Try a potential divider like: 5V->1k2 B 120 A 1k2 0V. That's the biasing for the RS-485 bus its self. That will mean when no RS-485 chip transmit is enabled the bus will be in idle state.

A small bus length does not matter. What does matter is the termination. You have to make it so that when idling both A and B are near the middle voltage with ([(A-B) < -200mV] see above) but when operating does not have open ends to the bus. Put 120 Ohms at either end of the bus as well.

This is pretty good https://www.maximintegrated.com/en/app-notes/index.mvp/id/763

\$\endgroup\$
  • 1
    \$\begingroup\$ on a short bus you can just put 60 ohms at a single location instead of terminating both ends, \$\endgroup\$ – Jasen Oct 31 '18 at 4:57
0
\$\begingroup\$

It depends on your data rate. Lower frequencies can be closer together, what you are trying to avoid is crosstalk. Use termination resistors on the lines and you should not have a problem. Most commercial solutions are internally terminated to some degree. That said, if the bus length is long, especially if its longer that the 'official' maximum length, you get into a grey area where terminations and line conditioning play a larger role. Always terminate at the receiver, never at the transmitter. The purpose is to change an LC circuit into an RLC circuit, which generally has better frequency characteristics. Terminating at the transmitter makes the problem worse though.

\$\endgroup\$
  • \$\begingroup\$ the transmitter should be strong enough that nearby termination has no effect. \$\endgroup\$ – Jasen Dec 21 '18 at 22:03
  • \$\begingroup\$ Its a matter of noise immunity. terminating at the receiver provides much better noise immunity than terminating at the transmitter \$\endgroup\$ – Anthony Bachler Jan 1 at 18:04
  • \$\begingroup\$ usually both ends of a transmission line are terminated. especially if it's a bus. \$\endgroup\$ – Jasen Jan 2 at 22:41

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.