RS-485 protection

[twam]

Hi everybody,

I'm still working on my flipdot project where I have several modules containing 16x14 flipdots running of 24 V. I want to daisy-chain several of those modules using RS-485 for the data connection. My question is now how much and what kind of protection I should put on the RS-485 lines. A few months back I watched all the awesome videos from Mike where he often mentions small hints on that topic, but is very hard to find all those tiny bits of informations again. So what kind of protection would you suggest? I guess the most common misuse would be mixing up power (24 V up to 8 A) and data lines.

Best regards
  Tobias

[T3sl4co1l]

A transceiver that's tolerant of those voltages might be worth a look, e.g.
https://www.ti.com/lit/ds/symlink/sn65hvd1785.pdf
If you want to cook something yourself, you might consider:
- Series resistors (significantly worsens line drive, though)
- Depletion MOS current limiters
- TVS (e.g. SMAJ12CA) or SIDAC (e.g. TISP4015L1) with polyfuse, or somewhat less resistance (but, the resistor should be relatively high power)
- Uhh, anything else that comes to mind; could do worse than building your own transmitter from scratch (with adequate voltage ratings and current limiting), but that would kind of suck.

Tim

[Brutte]

If it is not connected then you can kill a transceiver only with ESD but if you pick a version that has some ESD protection then it will be ok during handling.

Once you connect the transceivers with wires - I do not see the point to protect each device separately. Just a pair of TVS diodes somewhere near master would deal with EFT. If that is a super long bus then maybe at both ends is not a bad idea.

As for mis-wiring protection - that is a matter of personal preferences.

[floobydust]

There are special TVS designed for RS-485, low capacitance 75pF and lop-sided clamping voltages -7V, +12V. See SM712.
Those are a tiny SOT-23, convenient and what I would consider the absolute minimum to use.
They protect against differential and common-mode overload.

But not 24V on the line. Nothing protects against that without smoking the transceiver or using a lot of parts.
Most RS-485 driver failures I see are due to line shorts, one leg of the transmitter fails, or an open ground.
Larger TVS are high capacitance so careful if you upsize to SMA or SMB sized parts.

How tough the protection is depends on your budget and the distances between nodes, and speed.

[twam]

Thanks for your input.

I originally planned to use the SN65HVD7x as a transceiver, but the suggestion by T3sl4co1l pointed to me SN65HVD1780. This is a cheap upgrade and really seems to cover most of my fears.

[Doctorandus_P]

Somewhere in his video's Mike shows he puts PPTC's between the connectors and the electronics. (With probably TVS diodes behind them).

A few ohms of series resistance does not matter much fro RS485, and the PPTC's together with the TVS diodes protect against DC voltages.
I like the SM712 mentioned erlier, will look into them more.
Those are pretty small though, and I do not see a DC rating in that datasheet, so if you also want to protect against DC you probably need a somewhat bigger TVS.

Also:
Don't forget that RS485 is (at least) a 3 wire connection. A GND wire is pretty much mandatory to keep the signal lines within the common mode range of the transceivers.

A good reference for RS-485 is "10 ways to bulletproof RS485"
https://duckduckgo.com/?q="10+ways+to+bulletproof+RS485"&ia=web

[floobydust]

I put the TVS on the other of the 10R (or polyfuse), to lessen the currents it experiences. Contrary to TI's circuit with SMBJ43CA TVS, at ~500pF.

[Giaime]

I put the TVS on the other of the 10R (or polyfuse), to lessen the currents it experiences. Contrary to TI's circuit with SMBJ43CA TVS, at ~500pF.

There's a reason they put the TVS at the line side, not at the transceiver side of the resistors: that's because if you do it your way, the TVS appears in parallel to the internal protection diodes of the IC.
Who knows how the current will share between them? (hint: external TVS have higher voltage drop than built-in ESD diodes...) The 10R resistors make sure the most of the energy gets dissipated in the TVS, not the IC.

[Brutte]

These are 600W TVSes.
The clamping voltage is somewhere near 15V so we are talkig about 40A here (for 20us).

Once you decide on TVS rating, it is the TVS that has to die first.
You do not want to allow some weird EFT to save TVS and fry resistor or transceiver or both  .

[mikeselectricstuff]

1W 6.8V zener, and 100mA polyfuse to protect the zener. Note that if there is a chance of significant ground shift you may need two back-to-back zeners if the inout can go below -0.6v due to ground shift.

For receive-only, then a couple of series ressistors may suffice.

[T3sl4co1l]

I put the TVS on the other of the 10R (or polyfuse), to lessen the currents it experiences. Contrary to TI's circuit with SMBJ43CA TVS, at ~500pF.

There's a reason they put the TVS at the line side, not at the transceiver side of the resistors: that's because if you do it your way, the TVS appears in parallel to the internal protection diodes of the IC.
Who knows how the current will share between them? (hint: external TVS have higher voltage drop than built-in ESD diodes...) The 10R resistors make sure the most of the energy gets dissipated in the TVS, not the IC.

Reminder that RS-485 devices do not have ESD clamp diodes.  They have a wide line voltage range: -8 to +12V or thereabouts.  The question is, which diode pulls in first: the internal ESD zener, or the external one?

The internal one will most likely* have a comparable V(I) curve to any other TVS, so it's not so bad putting another in parallel as long as it's lower -- or separated by the resistor, which indeed is to isolate the current somewhat.

*Unless it's a snapback device, not uncommon in ICs as far as I know, but hard to find isolated in the wild; one of the few examples.  Anyone know if transceivers use these internally, or how to tell?

For transient protection, personally, I just toss on a SMAJ6.0CA from ground to each line.  Or a higher voltage (9 or 12 say) if more range is required (e.g. industrial EMC spec -- combined with using a wider-range transceiver like the ones mentioned above, to tolerate the higher Vpk).

Bidirectional is better than a unidirectional device, because under normal conditions, we can expect the signal to bounce negative, even with common grounds, and such a condition will cause bit errors.

(I didn't discuss this earlier as the subject was sustained overvoltage.)

I'm a bit "ehh" about protecting a TVS with a PPTC -- even the small chips let through a lot of energy, and even as robust as TVSs are, they can only take so much (10s, maybe 100s mJ for an SMA size part).  And at the ratings needed here (~100mA fuse rating, >=32V maximum allowed drop), you need a fair size (read, slow) chip, offhand I think 1206 or so?

Another excellent use case for hybrid TVS-PPTCs -- these heat up the PTC with a TVS, giving actually reliable protection.  They're hard to find stocked, though.

Tim

[Brutte]

1W 6.8V zener, and 100mA polyfuse to protect the zener.

Anyone has access to IEC 61000-4-4 gear to watch how raw 1W zener dies when EFT is clamped to RS485 cabling? There is a reason the TVSes are there. These are rated for high instantaneous powers. Raw zener is not designed for such actions.
And even the smallest 0603 100mA polyfuse will pass 40A for 10ms without dissipating anything.

This could be a kind of protection against installers that ground the transceiver and apply 24V to data line (from 5A supply of course). But if you apply 100mA to 1W 6.8V zener (15V), it will be killed anyway after 10s so I would not bother.

For transient protection, personally, I just toss on a SMAJ6.0CA from ground to each line.

Is that enough? I think the negative absolute minimum voltage is -10V for many transceivers. I'd use 5.0CA. That would limit the common mode voltage but I cannot see any other good alternative.

[T3sl4co1l]

I wouldn't worry about it much, TVSs are just big zeners, spec'd for transients instead of regulation.

More than you ever wanted to know, but AN784 is the particular highlight here:
http://www.icbase.com/File/News/download/ON_Reference_3.PDF
I wouldn't recommend using just anyone's zeners for such an "undocumented" purpose (i.e., no one puts peak power ratings on their zeners), but I wouldn't feel bad about it with Motorola/On Semi's.  Well, I wouldn't at the time, but who knows if the appnote is still applicable?...

Tim

[mikeselectricstuff]

For receive-only applications, I've found these murata 100r PTC thermistors are brilliant

https://www.digikey.co.uk/product-detail/en/murata-electronics-north-america/PRG18BB101MB1RB/490-2469-1-ND/588683

They trip so fast that you don't even need any additional clamping, even for a +48V or -48V direct short.

[T3sl4co1l]

Ah yes, ceramic PTCs -- rarely seen, but interesting.  Their performance is worse (relatively high Rcold) but they're available in much higher voltages (450V+?) so can be useful in just the right place.

And yeah, an 0603 chip is small enough, and Rcold is large enough, that something on the other side stands a good chance of survival!

Tim