Supplier for 15 KV protection for RS485

Hello Chris,

That is one of the situations I am facing regularly. Every single time I ended up having to design it from scratch, even the transformers. My cases were usually not RS-485 but analog signal and power transfer across safety barriers. The bandwidths were similar though for the signals (>20MHz).

The main issue are the signal transformers. For IEC601 (medical) we need around 5KV, sometimes more. Completely contactless transfer is one option but becomes difficult in EMC cert. Closed loop transformers require the qualification of a reliable source for the wire or insulated core material, very stringent production procedures and then tests. I suspect you might not have a choice but go through that scenario.

In other words it is not easy but it is feasible.

If it's higher volume or you want to have a company supply custom transformers I'd contact companies such as Pulse Engineering. Also, consider other options such as fiber transmission.

Regards, Joerg

Try

If you put one of their SL2DIN-485 Signal Line protectors at each end of your cable, you won't get anything higher than 10 Volts through it. And if those particular units won't do what you want, I'm sure they could make something up for you.

Cameron:-)

Hello Cameron,

These look nice, especially since they are rail mounted. What still needs to be considered is whether any data loss is acceptable during an inductive spike. Could be ok with RS-485.

If much of the spikes appear in common mode there is another great way to muffle these: Run each pair (or the whole cable if it happens to be the shielded kind) through a toroid as many times as it comfortably fits. #43 nickel zinc material may be good for higher frequencies, #77 manganese zinc for the lower end of the spectrum. This could be mounted in addition to other measures, it is a rather cheap method.

Regards, Joerg

Sounds like a good idea - as long as the added inductance doesn't interfere with the data (the OP didn't specify a baud rate).

Personally, I would do both... and expect that the protocol's parity checking / error correction would deal with infrequent spikes in the wrong places.

Cameron:-)

Hello Cameron,

It really doesn't interfere. The largest common mode choke I ever used on a data line was 1mH. Not that I needed this much but it was available at reasonable cost, initially intended for the ISDN market. It won't make a difference as long as both wires go through together and, in the case of RS-485, the twist isn't messed with too much. At least the line impedance should not change and stray capacitances between wires and core should be kept low.

Regards, Joerg

...

Common-mode chokes add little (in theory, no) inductance to the differential path that carries the signal, but block common-mode voltages that could interfere with proper operation or produce unwanted EM interference. Those bumps on computer cables are common-mode chokes.

Jerry

Hello Jerry,

They are often more like Aspirin. Probably along these lines: EMC engineer looks at screen and says "oh, oh!". Design engineer looks and needs to go to the bathroom. Comes back and says he needs to call the boss. Some screaming is heard from the receiver, something about first shipments, trade show, quarter end results. A mad dash to the clamp-on ferrite box and, whew, just scraped by again.

Regards, Joerg