0
\$\begingroup\$

After (mentally) banging my head at the wall wondering why my own circuit is glitchy after the PC communicates with it, I read some posts, and noticed others had very similar problems. All because the charge-pump action of the Max232 IC causes noise in the high millivolt range.

My source of info is here: https://e2e.ti.com/support/interface/f/138/t/151543#pi320995=1

I already considered the two-transistor circuits in place of Max232 but they aren't meant to be very reliable.

The speed I'm trying to work with is 38400bps and I want to be able to send and receive any set of characters over the serial port.

I thought of a replacement circuit but I don't know how well it works, but if it has hope, I'll add capacitors at the input and output of each regulator and between each VCC and GND pin.

Since much of my circuitry runs on 5V, the idea here is to use 2 pairs of inverters with 12VDC for VCC (might have to try CMOS IC) then the conversion takes place in the inverters themselves, then after conversion, the signals are sent to and from microcontroller (AT89C2051 in this example) via 5VDC only. The "SerialIn" and "SerialOut" are connected directly to the PC's serial port's TX and RX respectively.

Would such an idea work? If not, then what is a better alternative to Max232 that's commonly available and that doesn't make noise?

circuit

\$\endgroup\$
  • 1
    \$\begingroup\$ Perhaps a MAX3232 variant would be quieter? digikey.com/products/en/integrated-circuits-ics/… Follow the datasheet for the correct size caps to use, 0.1uF or 1uF for example. \$\endgroup\$ – CrossRoads Sep 2 '19 at 2:56
  • \$\begingroup\$ If you’re willing to go to that much trouble, why not provide a separate power regulator to the MAX323 instead? \$\endgroup\$ – Bob Jacobsen Sep 2 '19 at 3:40
  • 1
    \$\begingroup\$ @Mike RS232 interface needs single inversion, not double inversion. Which MAX232 chip it is, manufacturer and exact type? What kind of caps you have for the charge pump and bulk bypass? Any filtering on the supply line? Please post the orignal MAX232 schematic so it can be improved. Some clone manufacturers may perform the pump action a bit differently to combat noise. \$\endgroup\$ – Justme Sep 2 '19 at 4:38
  • 1
    \$\begingroup\$ No. You cannot run HCT chips from a 12V supply. Your solution also needs to swing to a negative voltage (RS232 goes +ve & -ve). \$\endgroup\$ – glen_geek Sep 2 '19 at 14:23
2
\$\begingroup\$

RS232 signals swing both 12V positive and negative. You can sometimes get away with swinging from positive to ground, without a negative level, but (a) it’s not standards compliant and (b) you can’t count on It working.

Hence the value of the MAX232, which provides its own supplies for that swing.

If it’s putting too much noise on the power rail, you should fix that problem instead of trying to reinvent a chip that’s worked well in lots of designs.

A brute force approach is a completely separate supply to regulate the power to the MAX232. Either a linear regulator or DC-to-DC regulator would do it.

Short of that, some low ESR capacitors and ferrites could be tried.

\$\endgroup\$
  • \$\begingroup\$ The caps I use are ceramic (mlcc) 1uF. Where exactly would the ferrites go? And so you're saying I should use two 5V voltage regulators: one for the Max232 only and the other one for the rest of the circuit? \$\endgroup\$ – Mike Sep 2 '19 at 6:18
  • \$\begingroup\$ Actually, the voltage swing of RS-232 can be as small as +/- 3V, and +/- 5V is very common. \$\endgroup\$ – Elliot Alderson Sep 2 '19 at 12:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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