3
\$\begingroup\$

I am using an Analog Devices ADM483E RS485-chip to build a device with RS485-capability. Datasheet here: http://www.analog.com/media/en/technical-documentation/data-sheets/ADM483E.pdf .

The finished apparatus is supplied with 24V DC, and now I'm looking to design it so that it isn't destroyed if a technician accidentally connects the supply 24V +/- to the RS485 inputs (which are connected to the RS485 input/outputs on the ADM483E).

The ADM483E is specified for +/- 14V applied to the RS485 inputs/output.

It would be perfectly adequate for the apparatus to simply short-circuit the power supply in order to protect itself. The power supply is 24V DC and I can arrange for it to have a fuse of 2A.

I can't use reverse biased diodes to ground, since -5V is perfectly acceptable RS485 signal level.

I've considered using transient suppression diodes, but the typical ones are only rated for 1-2 joules of transient energy, and I haven't found any 2A fuses that will reliably break at such low energy.

Do I have to build my own crowbar device, or is there some best practice for what I'm trying to do?

\$\endgroup\$
  • 1
    \$\begingroup\$ Have you thought about buying delibrately robust tranciever chips ? Like ti.com/lit/ds/symlink/sn65hvd82.pdf \$\endgroup\$ – Robin Oct 1 '18 at 11:02
  • \$\begingroup\$ That product won't help. The datasheet says maximum 18V to A/B-inputs. So it would also fry if accidentally connected to 24V. But some similar product which could endure say +-30V would certainly solve my problem! \$\endgroup\$ – avl_sweden Oct 2 '18 at 10:33
  • 1
    \$\begingroup\$ Yes, there are a couple of very good CAN TXRs that cope with malevolent wiring. The MAX3053 is \pm 80V. \$\endgroup\$ – Robin Oct 2 '18 at 10:35
  • \$\begingroup\$ Ah, cool! Thanks for the tip. Not a RS485-tranceiver though. \$\endgroup\$ – avl_sweden Oct 2 '18 at 10:36
  • 1
    \$\begingroup\$ Oh wait, the datasheet for the first chip you linked to says +-100V with 100 ohm in series. So maybe it would do the job in this case. Thanks! \$\endgroup\$ – avl_sweden Oct 2 '18 at 10:42
3
\$\begingroup\$

Rather than the brute-force method which is fraught with challenges & compromise, I'd put a couple of PTC resistors (aka polyfuse) in series with the 2 RS485 lines, & probably place them the last thing before they reach the connector to go off-board.

PTCs aka polyfuses have a near-zero resistance normally, but when you exceed their hold current & approach their trip current, they go fairly high resistance, and stay that way until the 'fault current' is removed. So if you apply 24V to the 485 pair, current will go through your termination resistor (120 ohms?) & fairly quickly put them into trip.

For RS485, depending on what source & line & termination resistors & impedance you've got, something around 50-100mA 'hold' current should be adequate for them to stay low-resistance at normal operating conditions.

PTC/polyfuse examples: http://www.littelfuse.com/products/resettable-ptcs/radial-leaded.aspx

\$\endgroup\$
  • \$\begingroup\$ What if the 24V is applied to GND and one of the RS485 inputs? And are the PTCs really fast enough to save the RS485 chip? \$\endgroup\$ – avl_sweden May 17 '15 at 19:10
  • \$\begingroup\$ Do you have line biasing resistors (as well as terminators)? \$\endgroup\$ – Techydude May 17 '15 at 23:34
  • \$\begingroup\$ I have no resistors. According to the datasheet for the ADM483E, these aren't needed, and I like to save the components. I still doubt that the PTC will respond fast enough to save the gates of the ADM483E. \$\endgroup\$ – avl_sweden May 18 '15 at 6:42
  • \$\begingroup\$ It's not up to the datasheet to tell you whether your specific application needs biasing resistors. but I was asking in case they could form part of the protection mechanism. So you're doubtful about the response time of PTCs, but want to depend on fuse protection? Then you've already lost :) Perhaps: PTCs in series with the differential pair inputs, and then a pair of ~12V zener diodes back-to-back to ground on each of the pairs? \$\endgroup\$ – Techydude May 19 '15 at 12:49
  • 1
    \$\begingroup\$ I've now realized that I could easily use zener's back-to-back, as per Techydudes response. I was just a little slow in realizing how it would work :-). With the right PTC the current could be held below 50mA. This would require a 2W Zener or so for the Zener to survive, a component which is readily available. \$\endgroup\$ – avl_sweden Feb 1 '16 at 12:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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