RS-485 holds the headline figure of 1200 m (4000 ft) between driver and receiver at low signaling rates, but that distance collapses as baud climbs: 10 Mbps is only guaranteed across roughly 12 m of twisted pair [S1][S2][S3].
CAN and 4-20 mA are physically the same copper, but their protocol budgets, voltage swing, and allowable loop impedance all push the practical maximum run length into a much tighter band than RS-485's marketing-friendly 1.2 km number. Selecting the right shielded cable for each bus is a numbers exercise, not a preference.
RS-485: 1200 m is the ceiling, not the default
The RS-485 standard from 1998 caps the recommended data rate at 10 Mbps, achievable across roughly 12 m (40 ft) of cable, and the TIA/EIA-485 spec itself sets 1200 m (4000 ft) as the maximum driver-to-receiver span at the low end of the rate table [S2][S3]. At 90 kbps, eye-diagram lab work still shows a usable signal after 1200 m of 22 AWG twisted pair, while 2 Mbps and 4 Mbps traces are visibly degraded over the same reel [S1].
Working tables from cable vendors and integrator guides put the practical run length against common baud rates at: 9600 bps up to 1200 m, 19200 bps up to 600 m, 38400 bps up to 300 m, and 57600 bps up to 200 m, with the 1200 m figure reserved for the slowest 9600 bps and similar low-rate links [S8][S3]. Modbus RTU trunk guidance on industrial sites further constrains the main bus to 700 m maximum, with branches kept short, even though the underlying RS-485 electrical spec would tolerate more [S9].
CAN bus: 40 m at 1 Mbps, 500 m at the slow end
CAN (Controller Area Network) is a different protocol layer with a tighter bit-time budget, so the same shielded twisted pair that pushes RS-485 to 1200 m at 9600 bps falls short of CAN at high rate. The commonly cited CiA-recommended numbers, repeated across industrial cable catalogs, are 40 m at 1 Mbps, 100 m at 500 kbps, 250 m at 250 kbps, and up to 500 m at 125 kbps and below. [S4]
CAN cable is essentially the same 120-ohm characteristic-impedance shielded twisted pair as RS-485, with 22 AWG or 24 AWG conductors and characteristic capacitance below 30 pF/ft (100 pF/m) between conductors; the difference is that the bit timing window at 1 Mbps is only 1 microsecond, and any extra cable capacitance or stub length eats directly into that budget. The article on shielded cable selection for load cell and strain gauge signal runs covers the same impedance and capacitance rules that govern CAN, since both rely on differential 120-ohm signalling.
4-20 mA: the loop, not the wire, is the limit
A 4-20 mA current loop has no baud-rate problem because the analog signal is essentially DC, but the maximum cable length is set by the total loop resistance against the transmitter's compliance voltage. With a typical 24 VDC supply and a 3.6 mA threshold at the low end, the available loop budget is roughly 20 V at the bottom of the span, so at 250 ohm total loop resistance (a common worst case including the receiver) the run can be hundreds of metres on 24 AWG. [S3]
RS-485 shielded twisted pair is acceptable to carry a 4-20 mA loop because the same twisted pair physically meets the noise-rejection need, and 4-20 mA loops can suffer from grounding issues, improper wiring, or unstable readings when the shield is not handled correctly [S7][S10]. The pair-vs-loop trade-off mirrors what is described in the shielded cable reference: conductor gauge and shield termination, not protocol, decide whether the run is stable.
Selection criteria: impedance, capacitance, gauge, shield
Across all three buses the four cable numbers that matter are characteristic impedance (100-130 ohm, with 120 ohm as the terminator target), conductor-to-conductor capacitance (under 30 pF/ft or 100 pF/m), conductor-to-shield capacitance (under 60 pF/ft or 200 pF/m), and conductor gauge matched to distance [S3][S4].
A practical gauge ladder for RS-485, echoed in Sungrow's installation note, is: 24 AWG up to 10 m, 22-20 AWG up to 300 m, and 20-16 AWG up to 1200 m, with unshielded twisted pair acceptable only under 10 m and away from VFDs or other EM sources [S4]. The same gauge rules apply to CAN and 4-20 mA since the only variable is ohmic drop and capacitance, which are physical properties of the cable rather than the protocol on top. For Belden-style 22 AWG PVC reference parts, Belden 9744 is one of the common RS-485 reference designs [S6].
Comparison table: RS-485 vs CAN vs 4-20 mA on run length
Side by side, the three buses differ less on cable type and more on what limits the run: RS-485 is limited by baud-vs-length, CAN is limited by bit-time budget, and 4-20 mA is limited by loop resistance and supply voltage. A 1200 m RS-485 run at 9600 bps, a 500 m CAN run at 125 kbps, and a 500 m 4-20 mA loop on 16 AWG with a 24 V supply are all feasible, but the engineering margin on each is set by a different failure mode. [S3]
Where the runs overlap, the cable is interchangeable: 120-ohm shielded twisted pair, 22 AWG, with characteristic capacitance under 100 pF/m between conductors, will carry RS-485, CAN, or 4-20 mA. Where they diverge, it is the protocol overhead (bit timing for CAN, AC signalling for RS-485, DC loop drop for 4-20 mA) that pulls the maximum run back from the 1200 m headline figure. checking and testing apparatus for field verification typically includes a 120-ohm terminator, a capacitance meter, and a loop-resistance check before commissioning a long run.
Failure modes and what kills a long run
On RS-485, the three killers are: DC parasitic resistance on undersized gauge at long runs (especially above 1000 m without 20 AWG or heavier), unterminated or double-terminated buses at 120 ohm that reflect the edges, and missing or multi-pointed shield grounding that lets common-mode noise into the pair [S1][S3][S4]. On CAN, the killer is usually stubs longer than the bit-time allows, which is why daisy-chain trunk with short stubs is mandatory above 250 kbps. On 4-20 mA, the killer is voltage drop on long home-run loops, particularly on 24 AWG or thinner wire, and ground loops where the shield is bonded at both ends.
For RS-485, Sungrow requires shielded twisted pair with 100-130 ohm impedance and capacitance limits of 100 pF/m between conductors and 200 pF/m between conductor and shield, and specifically excludes Cat6 as a guaranteed RS-485 cable for solar inverter chains [S4]. The general rule of thumb, repeated across vendor material, is that higher drive strength transceivers above the 1.5 V minimum help counter the DC parasitic resistance of long cables and stretch the 1200 m ceiling at low data rates [S1].
Who this is for, and where the standard numbers do not apply
This distance table is for industrial automation, BMS, solar inverter daisy-chains, and process plants running Modbus RTU, BACnet MS/TP, CANopen, or DeviceNet on copper. It is not for high-speed CAN-FD at 2 Mbps and above, which collapses the run to roughly 20-40 m, and it is not for FOUNDATION Fieldbus H1 segments that share the same 120-ohm cable but have their own 1900 m trunk specification under IEC 61158. [S5]
The 1200 m RS-485 number is also contingent on the cable spec: 22 AWG or heavier, 120-ohm characteristic impedance, low capacitance, and proper termination, and on the 32-unit load limit of the RS-485 standard, which forces repeaters above 32 nodes. The 4-20 mA loop distance is not a fixed number, it is a function of supply voltage, receiver input resistance, wire gauge, and the loop must be calculated against the transmitter's compliance; do not use 500 m as a universal ceiling without checking the math.
Field-verification handoff should always include: a 120-ohm termination check at both ends, a capacitance reading under 30 pF/ft between conductors, a shield-ground bond at one end only on RS-485 and 4-20 mA (both ends on CAN in noisy plants), and a loop-resistance calculation against the transmitter compliance on 4-20 mA. A proper cable gland at the panel entry, bonded to the enclosure ground, is what keeps the shield from becoming a radiating antenna at the chassis transition.