RS-485 protection

You could check an industrial RS-485 driver like the 75LBC184 (TI), which has a lot of built in protection. It won't survive 240 VAC supplied directly to its inputs but besides that it is in my experience almost indestructible.

Regards, Klaus

Isolating the RS-485 from ground will buy you a lot more reliability per dollar than worrying about differential voltages.

Best regards, Spehro Pefhany

I looked at some of the fully-isolated RS485 transceivers, e.g. the ADM2484E.

Seems like that requires a fairly beefy isolated DC-DC converter, though, since I doubt the interface spec includes 5V to power the isolated side of the transceiver.

The MOSFET trick ought to handle both line-to-ground and differential problems without needing that isolated supply. Or am I missing something ultra-important?

Thanks

Phil Hobbs

Won't that blow out the gates?

The depletion fets are great, but they still deserve ESD protection. That can be done with small capacitors.

LT has the widest selection of RS-485:

Not sure /anyone/ protects against a differential 240VAC across the inputs.

Don't think so. The FETs turn off at about -2V, which limits the voltage across the polyswitch, and hence V_GS. Almost all the voltage will be dropped by the normally-biased FET. Of course that makes the dissipation much higher in a sustained fault condition.

The polyswitch I had in mind is the TE RXEF010, which is nominally 2.5 to 4.5 ohms, and the FET gates are rated +-20V. I suppose inductance might be an issue though, so some small caps from G to S would be useful to slow the edges down. Check.

Cheers

Phil Hobbs

Well, check out better-grade multimeters. ;0

Cheers

Phil Hobbs

Take a look at these notes

That's potentially a nice part, thanks.

Cheers

Phil Hobbs

I have used the LTC2862 (60V protection) in combination with galvanically isolation. It's expensive, but from the sound of your application you cannot afford to save pennies, risking failures

Cheers

Klaus

I agree. A small, isolated, DC--DC converter and some optocouplers will give you more peace of mind than all the poly-fuses, MOV, or ESD clamps on the planet.

PPTC in series and bidirectional TVS diodes on the load side.

for twin pairs, use a PPTC on both inputs with the TVS across them on the other side. You could also put TVS from those points to common.

We had an issue where a device with a ADC input would take a whack now and then, it got tedious to replace it. So we installed a analog isolator which took the hit and simply plugged in a new card when the isolator crapped out..

After awhile, even that got ridicules. So we experimented with some 50mA PPTCs and a TVS diode across them on the load side. That reduced the need to replace the isolator for 95% of the faults.

Jamie

Good stuff, thanks.

Cheers

Phil Hobbs

OK, as long as the poly doesn't absolutely open, the gates are OK.

Of course that makes the

The poly doesn't help all that much. It needs about 400 mW to heat it to its trip point, which means it will still pull a lot of current if less then 2 volts is available. So the current in the fets will stay high, as will the power dissipated in the fets.

The Idss of the CPC5603 might never trip that polyfuse. So the fets could fry, which might technically "blow the gates" after all.

What we need here is a negative-resistance fet circuit thingie. I can think of ways, but they aren't pretty.

Hey, here's an idea from some past thread...

I was thinking about caps to ground, at the i/o connector. 1, 10 nF maybe, to divide down any ESD zaps. Kills RF, too.

This input protection thing can get messy. Polyfuses and big transzorbs are one way, up to the polyfuse voltage limit. There are some high-voltage polyfuses around.

We've used SSRs and transzorbs, with logic to sense the overload and turn off the SSRs before they fry. Works, but it's complex.

I'd say, if someone connects line voltage to an RS-485 gadget, they blow it up.

It's easy for a non-terminating, receive-direction RS485 line.

The ammeter inputs of the better DVMs have a huge cartrige fuse inside.

It might be interesting to use a SMT polyfuse thermally coupled to the FETs, sort of a homebrew PolyZen.

Kind of tough on the bandwidth, though.

I may make that call too, but I wanted to kick all the tires first. I'm getting a few of the FETs and Transzorbs to see how they work.

Cheers

Phil Hobbs

ADI makes isolated drivers that don't require a separate supply. If you need more than one channel, it's probably cheaper to use a separate chip to do the isolation, which can also supply the RS-422 driver. They aren't cheap but they work really well.

Years ago I once asked why the cost for the replacement fuses on the flukes were so expensive. The answer was that each one is calibrated and fully tested before packaging.

Tell me how you calibrate and test a fuse?

Jamie

Take a batch of perhaps 100 fuses and destructive test one or two fuses. If they pass, assume that the other 98 fuses are acceptable and ship them.

If the fuses are for a high reliabilty application, destructive test