Common mode choke for RS232 signals going onto cable?

[cupoftea]

The following eRIC9 transceiver module is sending and receiving RS232 data from its pins 3 and 4 respectively. These signals are going to a connector on the PCB which connects to a cable, and goes to another PCB. We need to put a common mode choke(s) on these signal lines so as to stop the cable acting like an antenna and giving us an EMC fail. (the PCB that comprises the eRIC has four SMPS's on it.)
Do you know what woudl be the best type of common mdoe chokes for this purpose? Presumbaly it depends on the data rate (baud)?

eRIC9 Transceiver

https://www.farnell.com/datasheets/1755796.pdf

 

[FvM]

It's TTL level UART rather than RS-232 interface.

If you want to stop RF propagation through data lines, the "IO-protection" series resistors suggested in data sheet are probably the best means. If ground is included in the RF propagation scenario, you also need to care for power supply lines. In other words, you arrive at a completely different problem.

 

[cupoftea]

Thanks, i see the I/O protection resistors on page 6 of the eRIC datasheet....between eRIC and a micro thats on the same board as the eRIC.
Actually, in our case here, this is a connection going off board, on a cable.

 

[FvM]

Going off-board with TTL level? 3-wire connection? If so, think of a triple CM choke.

 

[cupoftea]

Thanks, yes, i see your great point, the thing is, is it really worth converting TTL to some differential signalling mode, and having to put a converter chip in there to do that...just because its going off the board?....i mean, can we not get away with just sticking to TTL for the off board connection?
Triple line common mode choke we cant find commonly on the web, so we would put ferrite beads in each line, do you think we will get away with it?

--- Updated Jul 22, 2021 ---

We have a system in a plastic enclosure comprising multiple PCBs interconnected by cables and looms. One PCB comprises the 4 SMPS's. There is an offline 20W SMPS on this PCB also. Our circuit ground is directly connected to earth. We need cable screens for all the cables. Do you agree that we must connect our cable screens to circuit ground at each end of the cable? The thing is, do you also agree that we need to put ferrite beads in the connection leading to the cable screen, in every case?
You see, the situation is that we do not have a separate chassis ground, because our enclosure is plastic...so therefore we cannot connect our cable screens to any chassis ground.

 

[KlausST]

Hi,

As so often: it depends..on
* expected noise
* ground signal current (caused by loops)
* receiver input threshold levels
* receiver input filter

You will have your reasons for expecting common mode noise.

Basically there is
1) a transmitter. It sends out pulses (voltage signals). The voltage levels refer to it's GND_reference (Tx_GND).
Also the rise time (dV/dt) of the transmitter plays a role when considering ringing, overshot...

2) Then there is a cable...with it's length and characteristic impedance....the communication signal....and the unwanted GND_currents that don't have any relation to the communication signals at all but cause a shift in (Tx_GND - Rx_GND)

3) The receiver circuit. Maybe with an LPF and an receiver with it's input threshold levels referring to it's receiver_GND (Rx_GND).

You get wrong receiver signals when the (Tx_signal + noise + overshot + GND_bounce) leads to a wrong Rx level.

So one might be able to calculate the overshoot caused by the impedance (mismatch) of the cable.
But for us it's impossible to decide whether there is high or low GND_bounce caused by GND currents, and how big the expectable (induced) noise will be.

I'd say:
A suitable wiring (supply_GND, signal_GND, communication_signal ....) and a suitable LPF at the receiver_input..will reduce the chance of erroneous signal significantly. For short distance in a not_very_noisy environment even TTL signaling should work reliably.

Klaus

 

[cupoftea]

Thanks, as well as reducing the bit error rate, we want to stop our cables from radiating and failing EN55032 radiated EMC....but data line CM chokes have very low impedance at 100MHz....so we're a bit stuck...do you know the way round this?

 

[KlausST]

Hi,

a series R will reduce dV/dt and thus reduce emmissions.
Mind: Nowadays "TTL like" logic is rather fast with high dV/dt. Reducing your baud rate will not reduce dV/dt ... but you may do it with the series resistors (maybe add capacitors)

Klaus

 

[betwixt]

I'm really not sure you are on the right track here. In my opinion the far better solution is to simply screen the cable. Use slew rate limited drivers if necessary but adding chokes will not reduce BER unless your EMC problem is already VERY severe (like you wrapped the cable around your SMPS transformer core!).

All single wire signalling will radiate some interference and be prone to picking it up, there is no avoiding that. I think the better strategy is to concentrate on the escape path rather than pre-conditioning the signal. Use a screened communication cable, carry ground (usually pin 5) by wire along the cable and connect the screen to the plus shell at each end. That gives you a completely screened end-to-end, chassis-to-chassis connection, even if the real signal ground is carrying noise.

For better results, use differential signalling, it is no more complicated electrically and has many advantages, especially in regards to signal leakage. I have a run of RS422 along CAT-5 cable here that runs parallel to and spaced 3m away from an antenna for some 30 metres. The cable carries continuous data at 38.4Kb/s and the antenna is connected to a transceiver that can pump out 100W across 1.5MHz to 60MHz. Its receiver is sensitive to 160nV for 10dB SNR and they don't interfere with each other. Another system only a few metres further away can feed 180W at 144MHz and yet another 20W at 1250MHz. I'm not sure what that equates to in EMC numbers but I'm sure the figures would be impressive.

Brian.

 

[cupoftea]

Connect the screen to the plus shell at each end.

Thanks, i am sorry though i dont recognise "plus shell".?

 

[betwixt]

Sorry, typing with my brain in neutral gear. Solder or clamp the shield in the cable to the metal body of the plug at each end. Effectively linking the chassis together through the shield while still using pin 5 as the signal ground.

Brian.

 

[cupoftea]

Thanks, i should be the one apologising not yourself ..i should have realised that.