In short: RS-232 is a single-ended, point-to-point link good for a few metres, while RS-485 is a differential, multi-drop bus that carries data over 1,200 m or reaches 10 Mbps, and shrugs off the electrical noise that corrupts RS-232.
Two standards, two different problems
Engineers often treat RS-232 and RS-485 as two versions of "the serial port". They are not. They are both electrical layers only: neither defines a protocol, framing or baud rate. Your UART still produces the same start bit, 8 data bits, parity and stop bit in both cases. What changes is how those bits are turned into voltages on the wire, and that choice decides distance, noise immunity and how many devices can share a cable.
RS-232 (TIA/EIA-232) was written in the 1960s to join a terminal to a modem: one DTE, one DCE, a few metres apart. RS-485 (TIA/EIA-485) was written in the 1980s for factory floors, where dozens of instruments hang off one long cable running past motors and welders. Keep that origin in mind and every difference below becomes obvious.
Single-ended versus differential signalling
RS-232 sends each signal as a voltage on one wire, measured against a common ground. A logic 1 (mark) is a voltage between -3 V and -15 V; a logic 0 (space) is +3 V to +15 V. Note the inversion: negative means 1. The band between -3 V and +3 V is undefined, which gives a generous 2 V of noise margin on a typical ±5 V to ±12 V driver. Chips such as the MAX232 create these rails from a single 5 V supply using charge pumps.
The weakness is the shared ground. If two boxes sit on different mains outlets, their grounds can differ by several volts, and any noise picked up by the signal wire is added straight to the signal. The receiver cannot tell noise from data.
RS-485 uses a twisted pair, called A and B, and the receiver looks only at the difference VA - VB. A difference above +200 mV is one state; below -200 mV is the other. A compliant driver delivers at least 1.5 V differential into a 54 Ω load, so the margin is large. Noise couples almost equally onto both wires of a twisted pair, so it cancels in the subtraction. This is called common-mode rejection, and the receiver tolerates a common-mode window of -7 V to +12 V.
Key fact: RS-485 does not win because its voltages are bigger. It wins because the receiver ignores anything that appears on both wires at once.
Point-to-point versus multi-drop
An RS-232 driver is designed to see one receiver (about 3 kΩ to 7 kΩ load). Hang two receivers on one line and you violate the driver's load spec; two drivers on the same line fight each other. Real RS-232 is therefore strictly a link between two devices, with extra handshake lines (RTS, CTS, DTR, DSR, DCD) when needed.
RS-485 drivers have a tri-state output, so many can share a bus provided only one talks at a time. The standard allows 32 unit loads; modern 1/8-unit-load transceivers allow up to 256 nodes. Two-wire RS-485 is half-duplex: every node shares one pair, and software or hardware must flip a driver-enable (DE) pin before and after each transmission. Four-wire RS-485 gives full duplex but is usually master/slave.
Distance, speed and why they trade off
For both standards, distance and speed are linked by cable capacitance and reflections. The original RS-232 spec limits load capacitance to 2,500 pF, which works out to about 15 m at 20 kbps. In practice, with low-capacitance cable, 115.2 kbps over 10 m to 15 m is routine, and 1 Mbps works over a metre or two. Long RS-232 runs are possible only at low baud rates, such as 9,600 bps over 50 m of decent cable.
RS-485 pairs a stronger drive with differential reception: about 10 Mbps at 12 m, 1 Mbps at roughly 100 m, and 100 kbps out to 1,200 m (4,000 ft). The rule of thumb is that speed × distance stays under about 108 bps·m. Beyond that, the cable's loss turns bits into mush.
| Feature | RS-232 | RS-485 |
|---|---|---|
| Signalling | Single-ended, ground referenced | Differential, twisted pair |
| Logic levels | ±3 to ±15 V (1 = negative) | ±1.5 V min drive, ±200 mV sense |
| Topology | Point-to-point | Multi-drop, up to 32/256 nodes |
| Duplex | Full (separate Tx/Rx) | Half (2-wire) or full (4-wire) |
| Typical max distance | ~15 m | 1,200 m |
| Max speed | ~1 Mbps (short) | 10 Mbps+ (short) |
| Common-mode range | Essentially none | -7 V to +12 V |
| Termination | None | 120 Ω at both ends |
Noise immunity in motion
The animation below shows a bit stream travelling along a noisy line. The upper trace is a single-ended wire: the noise bursts ride on the signal and push the high level toward the receiver threshold. The lower trace is the differential output: the same disturbance is present on both wires and the subtraction leaves a clean waveform.
Key fact: Twisting the pair matters as much as the differential receiver. Twisting makes the two wires see the same magnetic field, so the noise really is common-mode.
Practical design details that interviews love
- Termination: place 120 Ω (the cable's characteristic impedance) only at the two physical ends of the bus, never at every node. Unterminated fast edges reflect and cause ringing.
- Biasing and fail-safe: when no driver is active, A and B float and the receiver can output random bits. Pull-up on A and pull-down on B (typically about 560 Ω) keeps the idle state at logic 1. Many modern transceivers include a built-in fail-safe.
- Direction control: in half-duplex the firmware must release the bus right after the last stop bit. Releasing a byte too early truncates a frame; too late, and it collides with the reply. Modbus RTU also needs a 3.5-character silent gap between frames.
- Isolation: industrial RS-485 usually adds galvanic isolation (for example digital isolators plus an isolated DC-DC) because ground differences exceed the -7 V to +12 V window.
- Level shifting: a microcontroller UART runs at 3.3 V or 5 V logic (TTL), so both standards need a transceiver chip. Never connect an RS-232 line directly to a microcontroller pin; the ±12 V swing will destroy it.
For a deeper look at bus arbitration and frame timing, read our guide to RS-485 communication, and compare it with the clocked alternative in UART vs SPI vs I2C.
Which one should you pick?
Choose RS-232 for a console port, a bench instrument, a GPS module or a bootloader link where the cable is short and the device pair is fixed. Choose RS-485 for sensors, PLCs, energy meters, lighting (DMX512) and building automation, where cables are long, noisy and shared. Many boards carry both: RS-232 for a service port, RS-485 for the field bus.
From a verification viewpoint, the analog differences sit in the transceiver, but the digital UART behind it is the same RTL. Teams write the UART in Verilog, simulate it in Synopsys VCS, Cadence Xcelium or Siemens Questa, and leave the voltage levels to a separate PHY chip. If you want hands-on practice building and verifying such interfaces, the engineers at ChipXpert's VLSI training programme run lab-based batches.
Ready to build real communication-interface skills? Explore the hands-on embedded and VLSI modules in our CourseTron courses.
Frequently asked questions
Can I connect an RS-232 device directly to an RS-485 bus?
No. The voltages, wiring and driver behaviour differ. Use an RS-232 to RS-485 converter, which also manages the direction-control pin for half-duplex operation.
Why does RS-485 need twisted pair cable?
Twisting makes both conductors pick up the same interference, so it appears as common-mode noise that the differential receiver cancels. Untwisted wire loses much of that benefit.
Is RS-485 faster than RS-232?
Yes, at useful distances. RS-232 is typically limited to about 1 Mbps over a metre or two, whereas RS-485 reaches 10 Mbps or more at short range and still manages 100 kbps at 1,200 m.
Do I always need 120 Ω termination on RS-485?
Not always. At low baud rates over short cables, reflections die out before the receiver samples, so termination is optional. At high speed or long distance, terminate both ends.