Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations pierreick on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

RS485 "thermal incident" protection 6

Status
Not open for further replies.

RyreInc

Electrical
Apr 7, 2011
205
We are using a USB to RS485 interface, currently a Gearmo USA485422. We recently had a "thermal incident" i.e. fire on this device when one of the 485 lines was shorted to 24V for a sustained period. For the other devices on this network, the RS485 chips are resilient against such assaults, but this USB interface is not.

What is a simple way to protect against such an incident? Ideally the solution would keep the USB interface operational after the fault condition is removed, but I really don't care; I just don't want a fire! Would a simple slow blow fuse do the job?

Baud rate is currently 1Mbps, but a solution that allows up to 3Mbps would be nice. Would e.g. a fuse cause impedance issues, or otherwise interfere with transmission or max node count?

I should also mention that there is no PCB involved, so any solution would need to be mounted and wired in with screw terminals or something. I recognize that wire routing will be important with any solution to prevent reflections and whatnot.

Thanks!
 
Replies continue below

Recommended for you

Maybe a PTC resettable fuse i.e. thermistor?
 
Maybe a fused terminal for each concerned wire?



Just google it. Most automation houses have different manufacturers brands. Fuse size is the problem. I think Littlefuse or allied electronics would have the fuse size of below 1 amp. see below pictures



fusedTermiinal_mhuz9l.png


fuse_ez3dpf.png
 
Rather more expensive than the fuse would be an RS-485 isolator / repeater. That would effectively isolate your USB device from whatever happens on the main RS-485 bus.

We used these at each drop on a large gas detection system network after someone thoughtfully used one of our cable glands as a welding earth return and wiped out the whole lot.
 
Another solution is to use a surge suppressor rated for RS-485 (5 volts). Phoenix Contact, Citel etc make ones.
 
I would second the suggestion of an isolator. A RS485 network is much more robust with isolated ports.
 
Bear in mind that the converter in question has built-in optical ("photoelectric") isolation; whether it actually works or not is a different question. If it were working, then the most likely scenario is that the gnd terminal was "one of the 485 lines was shorted to 24V" but the interface doesn't specifically require gnd, although, in the bigger scheme of the "system" a common ground is required to protect the inputs from getting zapped. If that was indeed the case, then the gnd terminal should be disconnected and potted to prevent future incidents.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
Thanks for the replies so far! From a quick internet search it looks like the isolator idea is cost prohibitive. I'll look into that a little further though.

Regarding IRstuff's comment, unfortunately we do not know which line was shorted to 24V. That being said, it is probably most likely that it was the ground line. We are using a shielded twisted pair and grounding one end, at the ground terminal of this USB interface.

Maybe part of the answer then is to not use a shielded wire? I understand RS485 does not need the shield, but early on we were having communications issues (however most likely unrelated to noise on the line), and I basically defaulted to a shielded wire just to avoid introducing any additional troubles.
 
They show a PCB picture of that converter and there is no port isolation there. They must be referring to the LED indicators with that wording.

There would be no issue shorting any of the 3 conductors to 24V with 1500V isolated ports at each device. Yes, RS485 needs 3 conductors since RS485 uses ground referenced signal wires.
 
Barring a situation with a lot of noise, you could possibly float the shield at the converter end and only terminate at the other end. This way, a short to the shield would only affect the distal end equipment.

The "sustained" time is troubling, though; how did that happen?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
LionelHutz, that is not true. The only requirement is that the common-mode voltage is between -7 and +12V. See bottom of page 4: [URL unfurl="true"]https://www.renesas.com/us/en/doc/tech-brief/tb506.pdf[/url]

IRstuff, an integrator of our equipment wired the 485 cable into a general-purpose connector (a bad idea anyways and against our recommendation - but out of our control) that also had a 24V supply line going through it. This is why I say it was most likely the ground/shield wire that was shorted: they did not add insulation to the shield wire after stripping back the outer shield, and loose strands were observed.
 
Yes it is true.

You are wrongly construing that the RS-485 method of encoding the signal as the difference between the 2 signal wires means that RS-485 is a differential circuit. The circuitry for each of the signal wires is referenced to the circuit common/ground conductor. This can be clearly seen in a number of the figures in your link.

The transceiver can use earth ground as the 3rd conductor when it's not isolated, hence no need for a 3rd conductor.

Read AN005 from here. Read some of the others too if you want more help.

 
That's an excellent set of application notes Lionel. Thanks.
 
One thing I learned at a previous job is that some of the differential receivers, ala 26LS32, may need to have the circuit commons directly tied together, particularly if the two sides are not powered from the same AC main. AC main power-on transients can be large, and the differential in commons can wind up being equalized through the input pins of the receivers, for which they are not remotely designed for. This can result in shorted emitter-base junctions on the inputs, which can be verified by looking for >5 mA input grounded current.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
If a faulty installation is the problem don't look for protection! Fix the faulty installation! Then leave the system to operate the same as millions of other systems.

Keith Cress
kcress -
 
LionelHutz, thank you for that informative article. I was aware that a return path is needed, and what I failed to communicate was that a 3rd conductor is not necessarily needed in the same cable as the twisted-pair; the ground return can be incorporated into the device's grounding, as you and the article state. I definitely get why this is inferior to an isolated system!

But, an isolated system seems to be off the table for a few reasons, including cost. In this application all of the nodes will be within a dozen meters of each other at most (typically much less), and all be powered from the same 24VDC source, so I am pretty confident that a non-isolated 485 system will suffice. Some day there may come along a unique application that has the nodes more spread out and/or powered separately - in that case we may need a custom isolated 485 solution.

itsmoked, I couldn't agree more regarding fixing the installation! Unfortunately this is a product we are selling to external integrators so we don't always have control over things like that. Even though this "thermal incident" was very likely the integrator's fault, I'd still prefer our product to at least fail gracefully. That's why I'm looking for a cheap and easy fix, rather than an expensive one such as an isolator.

If no one thinks fuses will cause an impedance issue then I will probably just head down that road. Thank you for your responses!
 
IRstuff, yes a standard fusing solution would untwist the pair for a good inch or two... What about a thermistor? I would be worried that these have some inductance to them.

JG2828, are you trying to trick me into an isolating converter? [wink][wink]
I hadn't seen that before, seemed like a decent option, but after looking into it (you have to go awfully deep to find this!), that model only goes up to 128kbps...
 
A thermistor is probably even worse. You need to just do the proper termination and protection of the shielding; you cannot protect against any and all possible methods that a determined idiot could use to destroy a system.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor