solution of ground loop noise in 3 wires RS232 between two PCB board

Dear all,

I’m currently working on a project which requires RS232 communications
between two PCB boards. I wrote small script
to test the function of their communication and find out that the noise between them is really huge(3 wires RS232, no shield,baud rate 19200). The UART keeps of generating error interrupt(parity error) continuously and mess up the real signal. My code should work well, I verify my code by attaching the main board to another sensor(with UART interface )which has same baut rate, and there is no noise issue and I can send simply command to it and get respond for the sensor.

I powered the two boards with the same power supply(+12V), however, the two boards have their own voltage regulator which cut the voltage down to 5V. So the two boards don't have common ground, I guess.

I've bought a RS232 opto-isolator which should solve my problem. But I still want to post this thread to ask everyone, if there are some way to solve this problem without big surgery of PCB board.

Any suggestion would be appreciated, thanks in advance.
>
> Best regards,

anybody has an idea?

Solution: opto-isolated balanced line driver/receiver combo, such as IL3585. RS-232 is unbalanced and very prone to noise.

Beware that in the case of opto-isolation every board should have it's own power supply.

HDingmar said:

I powered the two boards with the same power supply(+12V), however, the two boards have their own voltage regulator which cut the voltage down to 5V. So the two boards don't have common ground, I guess.

Click to expand...

Just because they have their own voltage regulators doesn't mean they don't have a common ground. You'd have to show a schematic to be sure, but they probably do have a common ground (in the sense that the two grounds are connected to each other with a low resistance. But if that connection has high inductance, then there can be potential between them at high frequency).

Is one of the three RS232 wires a ground wire which connects to ground on both boards? Using a shielded cable with the shield as a ground connection would work well too.

And if it is a ground loop issue you can probably improve things by adding common mode chokes to the power supply feeds to each board.

zasto said:

Solution: opto-isolated balanced line driver/receiver combo, such as IL3585. RS-232 is unbalanced and very prone to noise.

Beware that in the case of opto-isolation every board should have it's own power supply.

Click to expand...

hi, zasto, thx for your remind.

But why should every board has its onw power supply? as I understand, the isolation will only isolate the three wires(TX,RX,GND),right?

---------- Post added at 10:12 ---------- Previous post was at 09:49 ----------

mtwieg said:

Just because they have their own voltage regulators doesn't mean they don't have a common ground. You'd have to show a schematic to be sure, but they probably do have a common ground (in the sense that the two grounds are connected to each other with a low resistance. But if that connection has high inductance, then there can be potential between them at high frequency).

Click to expand...

hi, mtwieg,
thank you for your suggestion.you are right, they do have the same ground. And there is no high freq noise.

mtwieg said:

Is one of the three RS232 wires a ground wire which connects to ground on both boards? Using a shielded cable with the shield as a ground connection would work well too.).

Click to expand...

Yes, the three wires(ribbon wire) are RX,TX and GND. then GND is signal ground which connects ground of both boards. Just let you know, when I power the two boards with different power supply, the error interrupt of rs232 occurs about several hundreds times per sec. After I use the same power supply, the error interrupt occurs only 7 to 8 times per second in a very regular frequency. The shielded cable sounds good, I'm going to try that.

As for the common mode chokes is to solve the high frequency noise, I assume that it would not help my problem, right?

Think once again:

as I understand, the isolation will only isolate the three wires(TX,RX,GND),right?

Click to expand...

Wrong! how can you isolate common ground :smile:

As you stated in your first post, your speed is 19200 bps. Higher baudrates are more prone to noise than lower ones. Ground loop noise takes some art to eliminate. PCB deesign, way of grounding, etc.

What is your distance between boards, anyway?

ok, yes, I said isolate GND, means isolate the signal GND(although they are common ground) at the UART connection, which just means that there is no electrical connection on the UART connection, is that right?


the distance is quite short, only about 10cm. Do you have some recommend design example, or design tutorial for a simple IL3585 design?

HDingmar said:

hi, mtwieg,
thank you for your suggestion.you are right, they do have the same ground. And there is no high freq noise.

Click to expand...

If there's no noise, then what is the problem? you said in your first post that you had "huge noise." What are you talking about?

hi,mtwieg
sorry for the vague expression , I mean there is no high frequency noise, but still some interrupt error (occurs 8 times per second, very regularly ).
Yes, the three wires(ribbon wire) are RX,TX and GND. then GND is signal ground which connects ground of both boards. When I power the two boards with different power supply, the error interrupt of rs232 occurs about several hundreds times per sec. After I use the same power supply, the error interrupt occurs only 7 to 8 times per second in a very regular frequency. but still quite noisy, but no high frequency. Hope what I explain makes sense.

As for the common mode chokes is to solve the high frequency noise, I assume that it would not help my problem, right?

mtwieg said:

If there's no noise, then what is the problem? you said in your first post that you had "huge noise." What are you talking about?

Click to expand...

---------- Post added at 12:31 ---------- Previous post was at 12:30 ----------

ok, yes, I said isolate GND, means isolate the signal GND(although they are common ground) at the UART connection, which just means that there is no electrical connection on the UART connection, is that right?


the distance is quite short, only about 10cm. Do you have some recommend design example, or design tutorial for a simple IL3585 design?

zasto said:

Think once again:



Wrong! how can you isolate common ground :smile:

As you stated in your first post, your speed is 19200 bps. Higher baudrates are more prone to noise than lower ones. Ground loop noise takes some art to eliminate. PCB deesign, way of grounding, etc.

What is your distance between boards, anyway?

Click to expand...

---------- Post added at 12:54 ---------- Previous post was at 12:31 ----------

mtwieg said:

If there's no noise, then what is the problem? you said in your first post that you had "huge noise." What are you talking about?

Click to expand...

hi, mtwieg

thanks, man, I have my problem solved. I used a shielded cable with one shield end connected to one ground. Then the noise is gone. So the issue is EMI problem, since the noise frequency is very close to 110hz, I would guess that it's the EM of this room of power. do you think so?

HDingmar said:

thanks, man, I have my problem solved. I used a shielded cable with one shield end connected to one ground. Then the noise is gone. So the issue is EMI problem, since the noise frequency is very close to 110hz, I would guess that it's the EM of this room of power. do you think so?

Click to expand...

It's extremely odd for any kind of EMI or noise to cause problems with just 10cm of RS232, even if the signals are jut over wires. RS232 is an old, robust feature which is really hard to corrupt, especially at low baud rates like 19.2k. And the problems with EMI are really only an issue at high frequencies, not at ~110Hz. A shield on the cable won't really do much to fields of that frequency.

Are you able to observe any actual commode noise with an oscilloscope?

What are your voltage levels? RS-232 needs to be MINIMUM +/- 3 volts. You said you regulate down to 5V. Maybe your drivers aren't producing the proper voltage levels.