How-to
Sizing Servo Reduction for Belt-Driven Plasma Table Axes
Short answer
A 4.000 inch pitch-diameter pulley advances 12.566 inches per revolution, so 400 ipm turns it only 31.8 rpm and a 2500 rpm servo needs about 78.5:1 total reduction. Calculate travel per revolution from pitch diameter first, then pick ratio, motor speed and pulley size together; torque, voltage caps and software limits do not change the geometry.
On this page
- How many inches does a 4 inch timing pulley move per revolution?
- What reduction does a 2500 rpm servo need for 400 ipm?
- What speed will a 100:1 gearhead actually give?
- Can a 10:1 reduction with smaller pulleys do the same job?
- Does running a 90 volt motor on 75 volts set the cutting speed?
- Why won't a bigger motor or a software speed limit fix it?
- How much torque does the gantry axis actually need?
- Should both sides of the gantry share one reducer?
- Questions asked
Size the reduction from pulley geometry before you touch the motor catalog. A 4.000 inch pitch-diameter timing pulley advances the axis 12.566 inches per revolution, so a 2500 rpm servo needs about 78.5:1 total reduction to deliver 400 ipm — the travel geometry sets that number, not the motor's top speed. Torque, supply voltage and software velocity limits cannot change it; they only decide whether the motor you already own lives comfortably inside it. What follows is the arithmetic, the speeds that common ratios actually produce, a smaller-pulley alternative, and the torque and mass checks that close the decision.
How many inches does a 4 inch timing pulley move per revolution? #
One revolution of a 4.000 inch pitch diameter advances the axis π × 4.000 = 12.566 inches, so the pulley turns only 31.8 rpm at 400 ipm; use the pitch diameter, or tooth count times belt pitch, never the outside diameter.
Pitch diameter sits at the belt's neutral line, and it is the figure the pulley documentation publishes. If you would rather count teeth, travel per revolution equals drive-pulley tooth count multiplied by belt pitch — the same circumference by a second route. If the 4 inch dimension on your sketch is an outside diameter, the pitch diameter is smaller, so take the documented value rather than guessing the offset.
A belt axis obeys the same rule as a rack drive, with the pulley standing in for the pinion, so the geometry-first method used for checking servo speed margin on a rack and pinion axis applies here unchanged.
What reduction does a 2500 rpm servo need for 400 ipm? #
The ratio is travel per revolution times motor speed divided by axis speed: R = 12.566 in/rev × 2500 rpm ÷ 400 ipm = 78.5:1, and 200 ipm then falls out at half motor speed on the same ratio.
Work it in this order:
- Get the pitch diameter from the pulley documentation and compute travel per revolution: C = π × D = π × 4.000 in = 12.566 in/rev.
- Choose the motor speed you want at maximum axis speed — 2500 rpm here, a comfortable point for a 3000 rpm motor.
- Compute the total ratio: R = C × n ÷ v = 12.566 × 2500 ÷ 400 = 78.5:1.
- Check the slow end on the same ratio: at 1250 motor rpm the axis runs 12.566 × 1250 ÷ 78.5 = 200 ipm, so one fixed ratio covers the whole 200 to 400 ipm band.
- Round to a ratio you can actually build and re-check it: 80:1 delivers 12.566 × 2500 ÷ 80 = 393 ipm, still above target.
- Close with the torque and heating checks below, because the ratio fixes speed only.
These are calculated kinematics for the stated inputs, not measured machine speeds.
What speed will a 100:1 gearhead actually give? #
A 100:1 unit delivers 12.566 × 2500 ÷ 100 = 314 ipm at 2500 motor rpm, and reaching 400 ipm would take 400 × 100 ÷ 12.566 = 3183 rpm, past a 3000 rpm motor's rated point.
| Total reduction | ipm at 2500 motor rpm | ipm at 1250 motor rpm | Motor rpm for 400 ipm |
|---|---|---|---|
| 3:1 | 10,472 | 5,236 | 95.5 |
| 10:1 | 3,142 | 1,571 | 318.3 |
| 50:1 | 628 | 314 | 1,591.5 |
| 78.5:1 | 400 | 200 | 2,500 |
| 100:1 | 314 | 157 | 3,183 |
Calculated from 12.566 in/rev; kinematic results, not measured speeds. The decision: a 100:1 gearhead is a workable build if you accept 314 ipm as your rapid ceiling — do not chase 3183 rpm past the motor's rating. Before committing to any high-ratio unit, read its documented backlash, efficiency, rated input speed and rated output torque, because those, not the ratio badge, decide whether it belongs on a gantry. If nothing near 78.5:1 exists in your motor frame, the two-stage belt route below is the standard answer.
Can a 10:1 reduction with smaller pulleys do the same job? #
Yes: with 10:1 at 1250 motor rpm, 400 ipm needs a pulley of 400 × 10 ÷ (1250 × π) = 1.02 inch pitch diameter, and 800 ipm rapids need about 2.04 inches.
Build the 10:1 as two belt stages on a jackshaft — 2:1 into 5:1 is one illustrative split — and the small pulley also cuts the torque demand: a 1.02 inch pitch diameter has a 0.51 in radius, about a quarter of the 2.00 in radius of the 4 inch pulley, so the same axis thrust needs about a quarter of the output torque. When you get to choosing the timing pulleys, match tooth count and belt pitch as a pair, because belts have a minimum pulley tooth count and the belt's documentation sets the allowable tension and the tooth engagement you must hold.
The decision forks on what you already own: if the belts are bought and cut for 4 inch pulleys, keep them and fit the high-ratio reduction; if you can still change pulleys, 10:1 with small pulleys is the lighter build with far less carried mass.
Does running a 90 volt motor on 75 volts set the cutting speed? #
No: 75 V on a 90 V, 3000 rpm DC servo only caps the speed ceiling near 3000 × 75/90 = 2500 rpm; the motion controller commands the actual cutting speed, and load current erodes even that estimate.
For a permanent-magnet DC servo the terminal voltage covers back EMF plus the armature drop: V = Ke × ω + I × Ra. At constant load, speed scales with voltage, which is where 2500 rpm at 75 V and 1500 rpm at 45 V come from as first-order estimates. During acceleration the current term grows and less voltage is left for speed, and a PWM drive makes supply voltage and average motor-terminal voltage different quantities.
So fix the speed range with the transmission, and treat supply voltage as a ceiling held inside the drive's ratings, never as the speed control. One caution belongs here: a 75 to 90 V DC bus stores charge after power-off, so verify zero volts with a properly rated meter before touching drive terminals.
Why won't a bigger motor or a software speed limit fix it? #
Neither changes the geometry: a bigger motor adds torque but not travel, since the pulley still advances 12.566 inches per revolution, and a software velocity limit only stops over-speed commands without touching mechanical advantage, reflected inertia or belt tooth load.
At 3:1 the motor would loaf at 95.5 rpm to give 400 ipm, and the load inertia reflected to the motor shaft grows by roughly (78.5 ÷ 3)² ≈ 685 times compared with the 78.5:1 build — enough to wreck the inertia match the servo was bought for. A software cap hides the mismatch until someone raises the limit or loads a different profile, and the belt mesh sees whatever the wrong ratio dictates. Spend the money on the ratio, not on a larger frame.
How much torque does the gantry axis actually need? #
Far less than a 1300 oz·in class motor delivers: a 160 lb gantry reaching 400 ipm in 0.10 s needs about 27.6 lbf of thrust, or roughly 0.78 lbf·in (12.5 oz·in) at the motor through a 78.5:1, 90 percent efficient reduction.
Worked example, calculated for these inputs and not a measured result. Inputs: moving weight W = 160 lb, top speed 400 ipm = 6.67 in/s reached in 0.10 s, pitch radius r = 2.00 in, ratio R = 78.5, efficiency η = 0.90 assumed for the example — take your reducer's documented value.
- Mass: m = W ÷ g = 160 lb ÷ 386 in/s² = 0.414 lb·s²/in
- Acceleration: a = 6.67 in/s ÷ 0.10 s = 66.7 in/s²
- Thrust: F = m × a = 0.414 × 66.7 = 27.6 lbf, before friction and cable drag, so add your measured resistance to it
- Output torque: T = F × r = 27.6 lbf × 2.00 in = 55.2 lbf·in
- Motor torque: T ÷ (R × η) = 55.2 ÷ (78.5 × 0.90) = 0.78 lbf·in ≈ 12.5 oz·in
Two cautions close the check. First, read the motor datasheet before trusting the nameplate figure: continuous versus peak torque, the units, and the continuous current rating, because copper loss P = I² × Ra means sustained low-speed torque still heats the winding even when speed is easy. Second, count every carried kilogram: a 50 lb reducer riding a 150 to 170 lb gantry adds 29 to 33 percent moving mass, and the thrust you just calculated rises by the same percentage. Once the ratio pins the speed, the deciding quantity becomes torque available at that speed — the trade-off that drives rack-and-pinion servo choice by torque at speed.
Should both sides of the gantry share one reducer? #
Use one reducer and a jackshaft when your controller cannot home the two sides independently; use dual drives when it can, because per-side homing squares the gantry at every start and removes any shared belt stretch from the alignment.
A jackshaft gives one backlash path and locks both sides in step mechanically; square the gantry with the tensioners and idlers at assembly, verify with a square and indicator, and re-check after the belt settles, because a skipped tooth is the only thing that can pull it out. Dual drives give two backlash paths that can rack the gantry at reversal if tuning differs, but homing each side to its own switch restores square every cycle. Commit this way: dual drives if your controller supports two coordinated axes with per-side homing; otherwise the jackshaft, and keep the spare servo and drive on the shelf. The belt-stage-versus-gearhead trade-offs behind belt or gearbox reduction for a stepper robot arm apply to the jackshaft stages the same way.
Questions asked
How do I figure travel per revolution without the pitch diameter?
Multiply drive-pulley tooth count by belt pitch. As a calculated example, a 24-tooth pulley on a belt with 0.500 inch pitch advances 24 × 0.500 = 12.000 inches per revolution, close to the 4 inch pitch-diameter case. Take the exact pitch from the datasheet of the belt you are using, because tooth count and pitch together decide which pitch diameters actually exist.
Will dropping the supply to 45 volts halve my axis speed?
First order, yes: 3000 rpm × 45/90 = 1500 rpm as a ceiling at constant load. Real load current and armature resistance push the number lower, and you give up rapid-traverse headroom. Set cutting speed as a commanded velocity in the controller instead, and keep the supply within the drive's ratings.
Would a stepper suit this low-rpm belt drive better than a servo?
Partly. Steppers hold torque at low rpm and lose torque as speed rises, so they tolerate small ratios, while a 3000 rpm servo needs the large reduction to stay in its efficient band. Once 78.5:1 is in place the servo cruises near rated speed with torque margin, at the cost of reducer mass, backlash and efficiency.
How do I keep a jackshaft-driven gantry square?
Square it mechanically at assembly with the belt tensioners or idler positions, then re-check after the belt settles. A single loop cannot rack the gantry the way two backlash paths can, but a skipped tooth will; size tooth engagement and tension for the full axis thrust, and home both sides from one shared switch.
How much acceleration do I lose by carrying a heavy gearhead?
Directly in moving mass: 50 lb added to a 150 to 170 lb gantry is 29 to 33 percent more mass, so the same acceleration demands that much more thrust and the frame sees it too. Move the ratio into two belt stages, or a smaller unit, whenever the mass lands on the moving axis.