Fault protected RS485, constant short circuit to power supply

[abomin3v3l]

Hi all.

This is my first topic in this forum.

I am designing a product for use in the industry and it needs to include a fault protected RS485 interface.
In the industry the most normal voltage for the power supplies in the machinery is 24V, but we have seen power supplies delivering higher voltages, up to 36V in some cases.
The RS485 is differential with 2 signals, "A" and "B", but what do I mean with fault protected? To be specific, I want that if a 23V to 36V source, of a power supply for example, stay attached incorrectly to A or B signals, constinuosly, there will be no any damage to the RS485 circuit. And also for -23V to -36V, if possible.

Below is my current schematic.



- SN65HVD1781 is the transceiver from Texas Instruments that can withstand commom mode voltages of -70 to +70V on A and B pins.
- TBU-CA065-200-WH is a current surge supression IC, 065 means that 650V impulse is supported, 200 means a I.trigger of 200 to 400mA (min/max, typ is 300mA). It has a typical R-series of 8.6 Ohms and maximum R-series of 9.8 Ohms.
- SMAJ36CA is to protect the TBU IC, it has a minimum breakdown voltage of 40V.
- The SM712 TVS diode, the datasheet shows that the Clamp Voltage (max) is 19V @ 1A.
- The SMAJ7.0CA TVS diode has a Maximum Clamping Voltage of 12V.

SITUATION 1. If a 23V source is connected to A or B:
(23V-19V)/9.8 Ohms = 408mA, then the TBU device would trigger.
19V is the Clamp Voltage (max) of TVS SM712 @ 1A. But in the datasheet of it is only shown that 19V is the Clamp Voltage (max) @ 1A, and it is required only 400mA to trigger the TBU device.
The datasheet of SM712 shows as the picture below, it does not show the clamping voltage below 1A.



SITUATION 2. If a power supply (+ and - wires) of 23V is directly connected between A and B, independently of the polarity:
Then the SMAJ7.0CA would trigger the TBU device.
SMAJ7.0CA has a Maximum Clamping Voltage of 12V, then (23V-12V)/19.6 Ohms (two TBUs in series) equals to 561mA, what would trigger the TBU device also.

What can you say about this? Do you think this protection will work when A and B signals get shorted continuosly to a power supply of 23V to 36V? Or even with less than 23V? What are your observations/recommendations?

Regards,
Jeferson.

DATASHEETS:
TBU Devices: https://www.bourns.com/docs/Product-Datasheets/TBU-CA.pdf
SMAJ TVS diodes: https://www.bourns.com/docs/Product-Datasheets/SMAJ.pdf
SM712 TVS diode: https://www.littelfuse.com/~/media/electronics/datasheets/tvs_diode_arrays/littelfuse_tvs_diode_array_sm712_datasheet.pdf.pdf

[mikeselectricstuff]

Looks like overkill - polyfuse on each line and bidirectional TVS to ground is all you need.
You don't need differential protection as the common-mode protection will kick in.

[nctnico]

I'd use the 36V TVS diodes only and call it a day. Putting fuses in series isn't necessary; these might cause impedance mismatch problems in case you are using long wires. The pull-up / pull-down are also not necessary. The receiver has build-in pull-up & pull down resistors.

[abomin3v3l]

Hi mikeselectricstuff and nctnico, thanks for your replies, I will consider/check this.

But about the protection circuit proposed in the first post, even if it seems overkill, nobody commented if it would work or not. Anyway I would like to know if it would work.

Regards,

[nctnico]

Define 'will work'... more parts means more places where things can go wrong so less reliability. Active components like the TBU-CA065-200-WH and the 7V TVS diodes will have some wear when subjected to current surges so could be more prone to failing. In this case the RS485 transceiver can deal with the maximum voltage present in the system so I wouldn't add extra protection to clamp the input voltage to lower voltages.

[abomin3v3l]

Hi nctnico, thanks.

Define 'will work' is close to the end of the original post: "Do you think this protection will work when A or B signals get shorted continuosly to a power supply of 23V to 36V? Or even with less than 23V?"

I think it will work, but I wanted a second opinion.

Regards.

[chris_leyson]

For fault protection I've used LT2862 or similar they have current limited drivers and will withstand +/- 60V on the RS422/485 A B pins. We had an issue where equipment suffered from damaged cabling from time to time and the RS485 signal lines would get shorted to 48V taking out the RS485 drivers. Replaced broken parts with LT2862 drivers and only had to ship replacement cables.

[mikeselectricstuff]

For fault protection I've used LT2862 or similar they have current limited drivers and will withstand +/- 60V on the RS422/485 A B pins. We had an issue where equipment suffered from damaged cabling from time to time and the RS485 signal lines would get shorted to 48V taking out the RS485 drivers. Replaced broken parts with LT2862 drivers and only had to ship replacement cables.

Nice solution for a retrofit to an existing design, but way more expensive than a conventional driver plus polyfuse+TVS protection.
Of course the other issue with using fancy bulletproof drivers is when they are out of stock and you can only buy normal ones. At the very least you'd want to put a built option on your PCB for protection.

[boz]

Isolate the RS485 as the end points may be on separate branch circuits in a large factory

Basically so you need an isolated 5V supply and an isolated data path either use data-rated opto-isolators (normal ones not really any good over 30Kbit/s) or use an isolated transceiver like in the circuit below (one that has been in service with 1000's of units for 6+ years)

[langwadt]

Isolate the RS485 as the end points may be on separate branch circuits in a large factory

Basically so you need an isolated 5V supply and an isolated data path either use data-rated opto-isolators (normal ones not really any good over 30Kbit/s) or use an isolated transceiver like in the circuit below (one that has been in service with 1000's of units for 6+ years)

you can get it all in one, ADM2587E et.al.

[langwadt]

I'd use the 36V TVS diodes only and call it a day. Putting fuses in series isn't necessary; these might cause impedance mismatch problems in case you are using long wires. The pull-up / pull-down are also not necessary. The receiver has build-in pull-up & pull down resistors.

is the build in pull up/down sufficient? Siemens sells an add-on fix for their some off their RS-485 stuff, without it noise on the idle line screws with communication

http://www.easycnc.org/media/catalog/product/cache/1/image/1024x/040ec09b1e35df139433887a97daa66f/A/5/A5E00151017-RS485.jpg

[chris_leyson]

Hi Mikeselectricstuff. The LT2862 was a drop in replacement for a competitors standard RS485 driver and was used on a 50m or longer cable run to interface a laser vision system to a welding control cabinet. 48V was used for the supply to avoid any voltage drop in the cable, sometimes 100m. It got expensive because a laser sensor, PCB from the control cabinet and cable had to be shipped back, repaired, and shipped to the customer. That meant a few weeks where your customer has a welding system out of commission. The additional cost of a robust RS485 driver is tiny when compared to all the shipping costs and loss to production, not to mention having to scrap some PCBs because of vaporised tracks.

I will see if I can dig out a schematic to check on the exact LT part number. TVS diodes were retro fitted to the laser vision system but didn't fix a short to 48V.

Would be nice to see someone reverse engineer the LT2862 because I couldn't see how the driver could survive a short to 48V.

[abomin3v3l]

I think I would like to have the option to have external-to-transceiver pull-up and pull-down resistors on the the bus, connected in one of the devices, in order the transceivers be less susceptible to noise, than just using the internal ones.

About PTC fuse, a PTC with I.hold of 200mA and I.trip of 400mA, in a part number I saw on Digikey, its normal/off resistance is around 0.8 ohms and the post trip resistance (R1,max) is around 5 ohms. If +24V connects to A or B, the TVS SM712 would burn, and maybe the termination resistor also. And I.hold and I.trip are very dependent on the ambient temperature. Really, I am not experienced with designs using PTCs.

Consider now a very wrong miswiring in the installation of the product, where the wires of the power supply were connected to A and B, and that the board is not powered (miswiring) and that then the GND of the board is floating, in this case the SM712 would not trigger the TBUs simply because there will be no current flow through it. That's the reason I have added the differential TVS diode SMAJ7.0CA.

Regards.

[chris_leyson]

Don't think the ADM2587E driver will toletate an A or B short to 48V. Nice part though, has good CMRR

[langwadt]

Don't think the ADM2587E driver will toletate an A or B short to 48V. Nice part though, has good CMRR

if the pair is floating it can't see 48V

[ajb]

I'd use the 36V TVS diodes only and call it a day. Putting fuses in series isn't necessary; these might cause impedance mismatch problems in case you are using long wires.

Without fuses a constant short circuit, which is one of OP's stated fault conditions, is going to just cook any TVS diode that trips.  A few ohms of series resistance from a polyswitch or these TBU dealies won't be a big deal for most systems.

The pull-up / pull-down are also not necessary. The receiver has build-in pull-up & pull down resistors.

  Those are line biasing resistors, which are most definitely NOT built into RS485 transceivers.  They should only be installed at one point in the line, and typically form part of the termination for one end.  There are plenty of 'fail safe' receivers available now that will treat an undriven line as idle, but with non-failsafe receivers you ideally want to bias the line to a definitively idle voltage state if it's ever going to be undriven.  Having those resistors available via jumper or switch is a nice convenience for the user. 

RS485 receivers don't really have pull-up or down resistors at their inputs.  The inputs do need to be divided down before they make it to the input comparator--how else can an IC running at 3-5V deal with the -7 to +12V common mode range the standard requires?--and those input dividers do pull on the line a little bit.  But because the valid input common mode range extends into the negative, they have to pull to somewhere between the IC's supply, and whether they pull 'up' or 'down' depends on the conditions on the line.  And in any case, they don't pull the line in the direction of any particular logical state, if anything they pull it towards zero differential voltage, which isn't a valid logical state at all, hence the need for a line biasing network. 

Don't think the ADM2587E driver will toletate an A or B short to 48V. Nice part though, has good CMRR

if the pair is floating it can't see 48V

Sure it could.  You still need to run the bus ground between devices (well, sometimes you can get away without it, but noise immunity suffers), and it's very common for that to be tied to a larger system/building ground somewhere in the system.  It's also very common for DC supplies to be tied to ground, possibly in the same cabinet as the RS485 bus.  Isolating the transceiver only breaks fault current/voltage reference paths through the device, you still have to deal with problematic voltages that appear within the interface. 

In the current market something fancy and specific like the all-in-one isolated transceivers are especially risky, but even in normal times they're awfully expensive compared to a standard transceiver, a digital isolator IC (can't see a good reason to use optocouplers anymore) and even a prepackaged isolated DC-DC converter, all of which can be multiply sourced. 

[abomin3v3l]

In my thinking, right at the moment of a short of +23V or more at A or B, there will be a current greater than 400mA, that is the maximum current required to guarantee that the TBU will trigger, and in about 1us it will trigger and limit the amount of current to a safe level, like few miliAmps (the datasheet shows 1mA @ 50V), and after triggered, when the voltage difference between its pins goes below a determinated value of voltage, that can be in the range from 12V to 20V according to the datasheet, then it untriggers. Maybe I'm thinking wrong.

Regards.