2
\$\begingroup\$

Is there any "simple" way to get (1500V) galvanic isolation on RS232 (just RX and TX)? By isolating, I mean than I have 2 different grounds on each side of the insulation barrier.

On side A, I have 1.8V, 3.3V, 4.5V, and 5V rails, and can accept voltage ranges between (-4, +4) and (-15, +15). On side B, I only have 5V rail, and it should be something rather standard (the equipment on side B is for future use).

For logic level signals, it's quite easy to use a digital isolator (or even an opto-coupler).

However, I'm quite stuck with RS232, which goes bellow 0V.

The solution I found so far is convert RS232 to TTL, then pass the isolation, and convert back to RS232 : nothing difficult, but that makes 3 ICs and lots of PCB space. I even found some ICs doing RS232 to TTL and isolation in one step.

But for now, I haven't found a nice way to directly isolate RS232 signals.

Any idea?

\$\endgroup\$
15
  • 1
    \$\begingroup\$ @Justme I retracted that because I realized I wasn't sure if the OP needed galvanic isolation or just to be able to withstand some difference in the ground potential at both nodes. \$\endgroup\$
    – DKNguyen
    May 16, 2023 at 19:02
  • \$\begingroup\$ Even if you only need TX and RX for data transfer, do you have the handshake pins available for positive and negative voltages? \$\endgroup\$
    – Justme
    May 16, 2023 at 19:04
  • \$\begingroup\$ @DKNguyen : I need galvanic isolation \$\endgroup\$
    – Sandro
    May 16, 2023 at 19:07
  • \$\begingroup\$ @Justme : on the A side, I have an IMU, with just TX and RX pins (RS232) for an external sensor (so no handshake pins). Side B is the "outside" side, with a connector to add an external aiding sensor to the IMU (magnetometer, GNSS, odometer, ...) \$\endgroup\$
    – Sandro
    May 16, 2023 at 19:11
  • 1
    \$\begingroup\$ Disclosure: I work at Analog Devices - we make a solution for this exact problem. Link: analog.com/en/product-category/isolated-rs-232.html \$\endgroup\$
    – MarkU
    May 16, 2023 at 19:33

4 Answers 4

4
\$\begingroup\$

I haven't seen any solution simpler than RS232 chips to TTL, I think that is the way you'll have to go with an isolation design.

\$\endgroup\$
4
\$\begingroup\$

You just need 2 optocouplers; one for Tx and one for Rx.

RS232 drivers should have enough current to drive an opto LED, it is just a question of where you connect the opto transistor. Generally you can get away with an signal that just swings between ground and 5V, so it is worth trying that. If you want a wider voltage swing, then you need a negative voltage rail, which obviously can be created using a switchmode IC.

If you really must save on components, you could instead create the poor man's negative supply by charging up a capacitor to that voltage from an unused handshake line - this works fine on non-isolated interfaces, but might be a bit tricky when called upon to drive an opto LED - it depends how much current is needed.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Sure you can power the optocoupler LED from TXD pin. But how do you power the optocoupler receive side to make data wire have RS-232 level +/- voltages, if you don't have them? If there were these voltages available, then your suggestion would work. \$\endgroup\$
    – Justme
    May 16, 2023 at 20:07
4
\$\begingroup\$

You could get the 3 chip solution down to a two chip solution by using a chip that has the isolation and RS232 transceiver in one chip.

If you put one isolated chip and one regular TTL to RS232 transciever back to back you can do the whole thing with two chips.

The ADM3251E made by Analog Devices is one example. The chip also includes isolated power generation so you don't need a separate chip for that.

enter image description here

\$\endgroup\$
3
\$\begingroup\$

assuming you need RS232 on both sides of the barrier There's basically four ways to do this.

A: use RS232 interface chips like MAX232 to convert to and from TTL level signals and opto-isolate those signals

B: double up the opto-isolators to allow them to pull both up and down.

C: ignore the fact that RS232 goes negative - if there is not much interferance the negative voltage typically isn't needed.

D: as C but with pull-down resistors so the output can go negative.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ A: sounds reasonable, B: Why, single should be enough at RS-232 speeds, and you get ones with digital pushpull output too, the problem is getting positive and negative voltages C: It stops being RS-232 at that point, you might as well just use a 5V CMOS inverter D: better idea than C. \$\endgroup\$
    – Justme
    May 17, 2023 at 5:55
  • 2
    \$\begingroup\$ digital push-pull output will struggle with the minimal +/-3V swing required to meet the RS232 spec, so you're left with single-ended analogue parts. \$\endgroup\$ May 17, 2023 at 13:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.