rs485 questions [SOLVED]

[1ABengineer]

it is possible to make a T-junction in a twisted pair cable for rs485.
for a project on my boat it is useful if I can hang a node (a lamp) on a T junction that is 2 meters from the main twisted pair cable.

if its possible does it need a terminating resistor.
An answer is greatly appreciated.
Kind regards, Bert Meijer

[tooki]

See e.g. page 3 of https://www.ti.com/lit/an/slla272d/slla272d.pdf

[Sorama]

In real world it is not a problem if not too long.
Definitely no extra termination.

[1ABengineer]

tooki, Great article , thank you. At the moment, I have no idea about the rise time. in mi circuit but i'm going to dive into the datasheet of the transceiver to find out.
Sorona, thank you. If I can't calculate it, I'm just going to try it

[Sorama]

tooki, Great article , thank you. At the moment, I have no idea about the rise time. in mi circuit but i'm going to dive into the datasheet of the transceiver to find out.
Sorona, thank you. If I can't calculate it, I'm just going to try it

I see it often, as well in industrial as in residential automation system where the baudrate is 385kb.

Mostly because they (designwise) forgot to pass the cable at this particular location.

[tooki]

In real world it is not a problem if not too long.
Definitely no extra termination.

Usually.

Some sources say you “can” add a terminator to each stub to prevent reflections, e.g. section 7 on page 4 of https://www.renesas.com/en/document/apn/rs-485-design-guide-application-note#page4

The issue with terminating each stub is excessive loading.

This page has some nice rules of thumb: https://www.felser.ch/profibus-manual/der_einsatz_von_stichleitungen.html

[Sorama]

…and that extra load make those drivers burn up way too soon.

I even have to remove the terminators  to have a decent amplitude at the bus.

[Siwastaja]

In real world it is not a problem if not too long.
Definitely no extra termination.

Usually.

Some sources say you “can” add a terminator to each stub to prevent reflections, e.g. section 7 on page 4 of https://www.renesas.com/en/document/apn/rs-485-design-guide-application-note#page4

The issue with terminating each stub is excessive loading.

This page has some nice rules of thumb: https://www.felser.ch/profibus-manual/der_einsatz_von_stichleitungen.html

Both resources utter bullshit, of course. Stubs cannot be terminated, as you say, because of the loading it causes. Yet the source you refer to just says it can be done, no conditions.

The next resource says that stub length depends on baud rate, which is totally made-up idea. With slower edge rates, longer stubs can be allowed, but that has absolutely nothing to do with baud rate and everything to do with slew rate control of the transceivers.

(The latter one could be excused by that it's documentation for one specific product only. They know the transceivers they use. They could even dynamically control slew rate based on baud rate setting (doubt it, but theoretically possible). But definitely not generic "rules of thumb"!)

My eyes start bleeding every time I see an URL in an RS485 related discussion. Even the best resources are just poorly written. The average ones are total made up bullshit. RS485 is very hard to work with simply because of so much disinformation with 40-year legacy.

Probably the only paper which deals with a very limited subset of things properly is the official modbus standard (see https://www.modbus.org/docs/Modbus_over_serial_line_V1_02.pdf ) which gives some correct advice in authoritative "must" form, to render appnote solutions recommended by the rest of the industry non-standard-compliant, rectifying confusing naming like "fail-safe biasing", and so on. Kudos for them! But it only scratches the surface.

For the OP, 2-meter stub is a bit awkward in the sense that it can't be unconditionally recommended without knowing what transceivers the bus participants use (and how they are configured e.g. with respect to slew rate control pin); if high-speed transceivers are used, it will definitely cause signal integrity problems. Then again, it is also quite likely it will work just fine. Testing if it works or not is easy, but if it works, then you don't know by what margin. One ghetto idea without signal integrity measurements is to test if it works reliably with e.g. a 10-meter stub. If it does, chances are better it works at 2 meters with some margin to spare.

[tooki]

In real world it is not a problem if not too long.
Definitely no extra termination.

Usually.

Some sources say you “can” add a terminator to each stub to prevent reflections, e.g. section 7 on page 4 of https://www.renesas.com/en/document/apn/rs-485-design-guide-application-note#page4

The issue with terminating each stub is excessive loading.

This page has some nice rules of thumb: https://www.felser.ch/profibus-manual/der_einsatz_von_stichleitungen.html

Both resources utter bullshit, of course. Stubs cannot be terminated, as you say, because of the loading it causes. Yet the source you refer to just says it can be done, no conditions.

The next resource says that stub length depends on baud rate, which is totally made-up idea. With slower edge rates, longer stubs can be allowed, but that has absolutely nothing to do with baud rate and everything to do with slew rate control of the transceivers.

(The latter one could be excused by that it's documentation for one specific product only. They know the transceivers they use. They could even dynamically control slew rate based on baud rate setting (doubt it, but theoretically possible). But definitely not generic "rules of thumb"!)

My eyes start bleeding every time I see an URL in an RS485 related discussion. Even the best resources are just poorly written. The average ones are total made up bullshit. RS485 is very hard to work with simply because of so much disinformation with 40-year legacy.

I ask that you tone down your wording, please.

“Can” does not mean “can under all circumstances without restrictions or consequences”.

As for the rules of thumb: the author is a retired expert in industrial communication who ran a university’s industrial communications lab for 30 years. So it’s not product-specific. And remember that a “rule of thumb” is a starting point only, an approximation, not a thorough, flawless formula covering all aspects.

I do agree with you about the dire state of RS485 documentation, though. The fact that the industry can’t even agree on how the two data lines are named (and how the polarity is defined) are solid evidence of this…