Direct answer: A servo cable has no single maximum length. Three independent limits decide it — voltage drop on the power conductors, voltage drop on the 24 V brake, and signal integrity on the encoder — and the shortest of the three is your real ceiling. In most machines the brake or the encoder binds first, not the power cores. As a working rule: up to about 20–30 m almost any correctly sized assembly works; between 30 and 70 m you must compute the drop and check the drive manual; beyond about 100 m, run the numbers, use the qualified long-length cable from your drive maker, or move the drive closer to the motor.
Why “maximum length” is really three questions
People ask for one number because the datasheet gives one number. Physically there are three separate failure mechanisms with three different answers:
| What runs through the cable | What goes wrong as it gets longer | What usually sets the limit |
|---|---|---|
| 400 V power conductors | Resistive voltage drop, cable heating, higher earth-leakage capacitance | The drive manual, plus a drop budget of roughly 2–5 % |
| 24 V brake conductors | Brake releases late, or does not release at all | Voltage drop against the brake’s ±10 % tolerance |
| Encoder / feedback signal | Attenuation, phase shift, crosstalk, comms errors | The drive manual for the specific encoder and cable type |
Because the three limits are independent, a cable that is electrically fine for the motor can still fail on the brake, and a cable that measures perfect on a multimeter can still throw encoder errors at 60 m. Solve all three, then take the shortest answer.
Limit 1 — Voltage drop on the power conductors
Voltage drop is simple resistance. For a run of length L (one way, in metres), current I (amps) and conductor cross-section A (mm²), the current travels out and back, so:
ΔU = 2 × I × L × ρ / A, with ρ ≈ 0.0175 Ω·mm²/m for copper at 20 °C.
Worked values for a 400 V servo axis drawing 8 A:
| Run length | 1.5 mm² | 2.5 mm² | 4 mm² |
|---|---|---|---|
| 20 m | 3.7 V (0.9 %) | 2.2 V (0.6 %) | 1.4 V (0.4 %) |
| 40 m | 7.5 V (1.9 %) | 4.5 V (1.1 %) | 2.8 V (0.7 %) |
| 60 m | 11.2 V (2.8 %) | 6.7 V (1.7 %) | 4.2 V (1.1 %) |
| 80 m | 14.9 V (3.7 %) | 9.0 V (2.2 %) | 5.6 V (1.4 %) |
On a 400 V drive this rarely bites first: even 80 m on 1.5 mm² stays under 4 %. Two caveats matter, though. First, peak current during acceleration is typically 1.5–2× the rated current, so double the number for a worst case. Second, the drop is wasted heat inside the cable: in a drag chain with no free air movement, a long run at high duty cycle runs hotter than the same cable on a machine frame. That is a thermal question, not just a voltage question.
Limit 2 — The 24 V brake is usually the binding constraint
Brakes are unforgiving. A spring-applied, electrically-released brake needs enough voltage to pull the armature fully open; at roughly 10 % below nominal it may buzz, drag, or fail to release, and the drive reports an overcurrent or a following error instead of a brake fault.
The same formula, applied to a 24 V brake drawing 0.7 A:
| Run length | 0.5 mm² | 0.75 mm² | 1.0 mm² |
|---|---|---|---|
| 20 m | 0.98 V (4.1 %) | 0.65 V (2.7 %) | 0.49 V (2.0 %) |
| 40 m | 1.96 V (8.2 %) | 1.31 V (5.4 %) | 0.98 V (4.1 %) |
| 60 m | 2.94 V (12.3 %) ✗ | 1.96 V (8.2 %) | 1.47 V (6.1 %) |
| 80 m | 3.92 V (16.3 %) ✗ | 2.61 V (10.9 %) ✗ | 1.96 V (8.2 %) |
At 0.5 mm² the brake is already outside tolerance at 60 m. This is why long-axis machines so often specify an oversized brake pair — and why “it works on the bench” is not a test: the brake releases at 24 V with a 2 m cable and stops releasing at 80 m on a cold morning, when supply voltage sags and coil resistance rises.
If you cannot change the cable, raise the supply to the top of the brake’s tolerance band, or use a brake with a lower holding current — but check that the drive’s brake output can supply it.
Limit 3 — Encoder signal integrity
Encoder signals do not fail because of resistance; they fail because of capacitance and noise. Every metre of shielded twisted pair adds roughly 50–100 pF, and that capacitance slows the edges and attenuates high-frequency content. Drive electronics have a limited budget for both, which is why the drive manual — not a formula — gives the maximum length for each encoder type.
| Feedback type | What degrades with length | Practical consequence |
|---|---|---|
| Analog 1 Vpp (sin/cos) | Amplitude and phase between channels | The most length-sensitive type; interpolation accuracy drops before the drive faults |
| TTL / HTL incremental | Edge rise time vs. pulse frequency | Works at low frequency on long runs, fails at high count rates |
| Resolver | Phase shift from cable inductance and capacitance | Tolerates longer runs than analog encoders, but needs the drive’s resolver cable compensation |
| Serial / digital encoder (Drive-CLiQ class) | Protocol timing and power for the encoder electronics | Governed by a published maximum in the drive manual; exceeding it gives intermittent comms alarms |
Two practical rules carry most of the weight. Keep the encoder pair away from the power cores — separately shielded pairs inside one jacket, or separate cables with at least ~200 mm spacing — and bond the shield at 360° to the connector shell at both ends, not with a pigtail. A pigtail drain wire raises the shield’s impedance at high frequency, which is exactly where encoder noise lives.
Worked example: a 60 m axis
Take a 3 kW servo motor, 8 A rated (16 A peak), 400 V supply, a 24 V / 0.7 A brake and a 1 Vpp encoder, on a 60 m run.
- Power: 1.5 mm² gives 11.2 V at rated current (2.8 %), 22.4 V at peak (5.6 %). Acceptable electrically, but step up to 2.5 mm² (1.7 % / 3.4 %) so the peak case stays inside a 5 % budget with margin.
- Brake: 0.75 mm² gives 8.2 % — inside the ±10 % tolerance but with no margin for supply sag. Use 1.0 mm² (6.1 %) or verify the actual supply at the cabinet terminals under load.
- Encoder: no calculation substitutes for the manual. Confirm the published maximum for this encoder with this cable type at 60 m; if the manual allows less, the run is what has to change, not the cable.
Result: 2.5 mm² power, 1.0 mm² brake, and an encoder type qualified for the distance. Note that the expensive part of the answer came from the two small conductors, not the big ones.
Practical rules that save long runs
- No joints. A terminal block or splice in a servo run adds contact resistance and an impedance discontinuity. Use a single continuous assembly built to length.
- Do not coil the excess. A coil of cable carrying current is a heater and an inductor. If a cable is far too long, replace it rather than looping it.
- Derate for heat. In a drag chain, in a cabinet, or at ambient above ~40 °C, the same cross-section carries less current than on an open frame.
- Respect the bend radius. A long run usually means more rollers, more chain, and a tighter layout — length problems are often really installation problems.
- One continuous shield. If the run is split across a connector mid-way, the shield must carry through the shell, not through a pin.
- When in doubt, move the drive. Relocating a drive 20 m closer is often cheaper than a qualified 100 m cable assembly, and it removes the encoder limit entirely.
Frequently Asked Questions
Is there a maximum length for a servo cable?
No single number. The limit is the shortest of three: power voltage drop, 24 V brake drop, and the encoder’s published maximum in the drive manual. For most machines the brake or encoder binds first.
How do I calculate voltage drop on a servo cable?
ΔU = 2 × I × L × ρ / A, where L is the one-way length in metres, A the cross-section in mm², I the current in amps and ρ ≈ 0.0175 Ω·mm²/m for copper. Double the current for the acceleration peak, and compare the result with the tolerance of what the conductor feeds — a few percent for power, about 10 % for a brake.
Why does the brake cable limit my length before the power conductors do?
Because it starts from 24 V instead of 400 V. A 2 V drop is 0.5 % of 400 V but 8 % of 24 V, so the same absolute drop consumes far more of the brake’s tolerance.
Can I extend a servo cable with a junction box?
Not on a moving or EMI-sensitive axis. A splice adds contact resistance, an impedance break for the shield, and a new failure point. Order a continuous assembly to the required length.
How far can a PROFINET cable run?
100 m per copper segment, which is a standard limit rather than a vendor preference. Longer distances need switches, repeaters or fibre.
Does a longer cable reduce motor torque?
Only through voltage drop at the motor terminals, and only measurably at high current and small cross-section. At 400 V with correctly sized conductors the effect is a few percent; the brake and encoder limits arrive long before torque does.
What cross-section should I use for a 60 m axis?
Size the power cores for peak current at no more than about 5 % drop, then size the brake pair for no more than 10 % at holding current. In the worked example that meant 2.5 mm² power and 1.0 mm² brake.
About TENSOlink
TENSOlink (GUANGDONG TENSOLINK ELECTRIC CO., LTD.) supplies OEM-compatible servo cable assemblies for Siemens V90, S120 and S210 systems, built to the length you specify, in high-flex, fixed-installation and oil-resistant jacket grades and matched to the original part number and pinout. If you are sizing a long axis, send us the drive model, motor current, brake data and run length and we will confirm the cross-section and the feedback-cable limit before you order. Visit our contact page to start.