Protecting RS-232 lines

[Krotow]

Our customer has equipment with control over two RS-232 interfaces. One of RS232 ports manage to fry every RS-232 adapter I know on irregular basis. It has two powerful DC stepper motors, some power relays and huge transformer. The RS-232 port that cause adapter damage control one of internal motor controllers that drive the largest motor (48VDC stepper). Damage most likely is caused by back EMF on emergency shutdown or power loss (it happened). That power loss also fried MOVs at 230V input. Surprisingly nothing else in device wasn't damaged. Device and our equipment are powered from same 230V AC phase so load disbalance between phases is not a reason. RS-232 adapter dying rate become annoying, especially because we are planning to install RS-232 converter to Ethernet to get rid of two 20m long USB extender cables (transceiver in one of them sometimes fry together with adapter and these cables cost like made from platinum) and possibly get free from RS-232 dying problem thanks to Ethernet design nature, but we aren't sure. So title question remain actual. How to properly and seamlessly protect RS-232 from irregular short but huge voltage (and current) bursts at client side? The surge voltage is unknown, but I guess, it is not 48V, but at least twice more. I guess, installing bidirectional TVS diodes on all signal lines isn't enough. Double the protection with MOVs? Something else? Or hope that isolated RS-232 transceiver inside Ethernet converter will be enough?

[David Hess]

If power loss fried the line input MOVs, then there could have been considerable ground current, part of which went through the RS-232 ground blowing the ground trace on the other end so normal protection networks will not help.  This is something I have seen before.

A galvanically isolated RS-232 interface would be a good start.  A series resistance added to each RS-232 line helps but anything more than about 220 ohms will causes problems with the RS-232 itself.

[Krotow]

Customer office is located in district where buildings are built 110-150 years ago. These places usually are horror show for electricians due to bad grounding. Electric installation in property is pretty new (two 3-phase 380VAC 50A inputs and decent wiring in building itself, installed during renovation at 2000-ies). They still have irregular power delivery problems coming outside their installation though. It is old district after all and building renovation works around still happen. So ground current somewhere might be the reason. I will try to start with galvanically isolated industrial 2 x RS-232 to Ethernet adapter and see can it withstand mentioned sudden voltage bursts. Luckily speed is not very important. Data exchange works between 19200 to 57600 bps, but move around 5-20 characters long text strings for like 10 times at second.

[PCB.Wiz]

...I will try to start with galvanically isolated industrial 2 x RS-232 to Ethernet adapter and see can it withstand mentioned sudden voltage bursts. Luckily speed is not very important. Data exchange works between 19200 to 57600 bps, but move around 5-20 characters long text strings for like 10 times at second.

You could also check the isolation capacitance and isolation voltage rating of the power supplies side of this.
If they do not spec it, you can check the pF with a multimeter.

[Geoff-AU]

The back-EMF from shutdown might be pushing the supply rail higher than 48V, maybe some hefty clamping on the stepper is in order to stop it from going to the moon.

Galvanically isolated RS232 and Ethernet connections sound like a good idea in addition to TVS clamping of signal lines.

[Krotow]

...I will try to start with galvanically isolated industrial 2 x RS-232 to Ethernet adapter and see can it withstand mentioned sudden voltage bursts. Luckily speed is not very important. Data exchange works between 19200 to 57600 bps, but move around 5-20 characters long text strings for like 10 times at second.

You could also check the isolation capacitance and isolation voltage rating of the power supplies side of this.
If they do not spec it, you can check the pF with a multimeter.

I guess that there should be no issues with power supply isolation... but... Ok. Seems I must to that anyway. Now the fun part. The customer is a medical facility and the device I mentioned is a CT scanner. Quite new model (mid-2010s). Proudly made by US/NL based company. With starting price tag from 50 K$. With innovative scanning technology without ginormous magnets (that is correct, I seen the scanning and result images and videos, resolution is astounding, doctors which whom I spoke, confirmed that). Controlled by a computer, scans accessible over Ethernet... you got the idea. Look superb from outside. But it eat RS-232 adapters like popcorn. Ar first the ones that came from manufacturer and then every adapter we bought. You will laugh, but customer once bought all RS-232 adapters our nearby computer store had only to avoid rejecting patients due to fried serial adapter. The warranty ended, all manufacturer tech visits cost a sum with at least 3 zeros. Adapters are obviously cheaper. Of course, it is not a normal situation. I had a chance to get inside that scanner several times and that turned my imagination about superb medical tech upside down. The stories about cobbling together DIY modules from Aliexpress and selling the resulting contraption for 6 digit price definitely have a ground. From inside scanner look like Tesla interior panels in worst meaning, a lot of premade parts like power supplies, and filters, DC converters, servo controllers etc. Original USB to RS-232 adapters are the same you could purchase in local computer store at corner for $10 a decade ago (those white/blue things, I disassembled one to be sure). At least power supplies look like original Mean Well ones. The overall impression is a project where they wanted to test the new invented the scanner part with a profit. So they cobbled it together with reused modules from older project and something RPi with HDD as temporary image storage. With their customers as the beta testers. What I can't understand - if they made a device with built-in image server then why they didn't control the device over Ethernet from some cross-platform Web interface too? Now to control the device, it need not one, but TWO USB to RS-232 adapters with active extension cables (due to distance from a computer) who fry together with adapter if you're lucky. They could omit the fuss with external USB to RS-232 adapters at all.

[PCB.Wiz]

... Original USB to RS-232 adapters are the same you could purchase in local computer store at corner for $10 a decade ago (those white/blue things, I disassembled one to be sure).

If it uses USB-RS232 modules, you could upgrade those to proper isolated ones, and try to get ones with best-rated rs232 transceivers. ±15 kV is common these days.

[Terry Bites]

Be very wary of the headline isolation voltages quoted.
A deeper read of the spec is always warranted. Most off he shelf solutions only have a real galvanic stand off of 2-3kV for 60s!
We did a job for CNC shop recently and to be doubly sure we used some RS232 to fibre converters from MOXA.