There is a particular email we get roughly twice a month, and it always reads the same way. Someone has a machine that cuts fine for two hours and then, on one job, rips a corner. Not a crash - a clean, quiet loss of position that nobody notices until the part comes off the table. They have already replaced the driver, checked the wiring twice, and blamed the CAM post.
Open-loop steppers do that. They do not tell you when they have lost steps, because telling you would require feedback they do not have. A closed loop stepper motor kit does not make the motor stronger. It makes the failure visible, and in most cases it stops the failure happening at all.
What follows is what we would actually look at if you asked us to recommend a kit for a machine - the sizing question, what belongs in the box, and where closed loop stops making sense.
What closed loop fixes, and what it does not
An encoder on the back of the motor reports actual rotor position. The driver compares that with where it thinks the rotor should be. If they diverge beyond a tolerance, the driver raises a fault instead of silently carrying on.
That gives you three real benefits on a machine:
- Lost steps become a stopped machine, not a scrapped part. On a CNC router a position fault stops the job. Annoying, but cheaper than a finished part that is 0.3 mm out on one feature.
- Current is adjusted to load. A closed loop driver ramps current up when the axis is loaded and back down when it is not. The motor runs cooler and quieter than the same motor on an open-loop driver at fixed current.
- Better behaviour at the top of the speed range. This is the one people notice most. An open-loop stepper at speed has very little torque margin; a closed loop drive keeps the axis on track further into the speed range.
What it does not fix: mechanical backlash, a flexing gantry, a coupling that is slipping, or a badly tuned acceleration profile. We have seen a closed loop kit fitted to cure a "lost steps" problem that turned out to be a loose grub screw on the pinion. The kit made the fault appear earlier, which was genuinely useful - but the fix was the grub screw.
Pick the frame size first, the torque second
Frame size is a mounting question, and it is not negotiable. NEMA 23 has a 57 mm mounting face, NEMA 24 is 60 mm, NEMA 34 is 86 mm. If your bracket takes a 57 mm motor, that is the kit you need, and the torque options live inside that frame size.
Our NEMA 23 closed loop motors come in five torque steps: 1.2, 2.0, 2.2, 2.5 and 3.0 N·m, all with an 8 mm shaft and a speed range of 0 to 1000 rpm. For a gantry router X or Y axis, 2.0-2.5 N·m on NEMA 23 is the common landing point; a heavy Z axis or a machine driving a rack and pinion through a reduction usually goes up to the 3.0 N·m end of the range, or moves to NEMA 34 closed loop.
Two things worth doing before you choose torque:
- Estimate the moving mass and the acceleration you actually want. Torque is needed for acceleration far more than for steady feed. A machine that cruises happily at 8 m/min may still need a bigger motor to reach it in the time the CAM expects.
- Do not size for the stall condition. People add margin for the worst possible cut, end up two frame sizes up, and then have a motor whose rotor inertia does not suit the axis. Oversizing is a real failure mode, not a safe default.
What should actually be in the box
A kit that arrives as a motor and a driver is not a kit. It is two parts with a wiring problem waiting in the middle.
The encoder cable is the piece that matters and the piece most often missing. Ours ship with a 3 m coding cable, and lengths beyond that are made to order. This is not a convenience line item - an encoder cable is carrying a low-level differential signal, and a cable made up from three-core flex and a length of shielded audio lead is a fault generator. If you extend it yourself, use twisted pairs with an overall shield, keep it away from motor phase cables and VFD output wiring, and ground the shield at one end only.
Also check whether the kit includes the motor mounting hardware that matches your bracket, and whether the shaft has a flat or a keyway. An 8 mm plain shaft in a clamp coupling on an axis that reverses direction under load will slip eventually.
The driver is where cheap kits fall apart
Two kits with identical motors can behave completely differently, because the driver sets how the motor is driven.
On our NEMA 23 kit the driver is a CL57R. The parts of that spec worth understanding:
| Parameter | Value | Why it matters |
|---|---|---|
| Supply voltage | DC 24-48 V, 36 V recommended | Under-volting is the most common cause of "it stalls at speed" |
| Output current | Peak 6.0 A, load-dependent | Current tracks the load rather than being fixed |
| Control modes | Position, speed, homing, JOG, multi-stage position and speed | Homing and multi-stage modes save controller I/O on a retrofit |
| I/O | 7 inputs, 3 outputs, programmable | Enough to run limits and a tool sensor without extra hardware |
| Protection | Overcurrent, overvoltage, position tolerance | The position tolerance alarm is the whole point of closed loop |
Inside the driver there is also acceleration, deceleration and a smoothing filter. The smoothing filter is easy to overlook and genuinely useful on a machine with a resonant axis - it takes the hard edge off the step pulses and the axis stops ringing.
When something is wrong, the driver tells you rather than leaving you guessing: a single red blink within three seconds means overcurrent or an interphase short circuit, two blinks means overvoltage. Start there before you start swapping parts.
The wiring mistakes that come up again and again
Three of them, in the order we see them:
- One power supply for everything. Steppers and a spindle VFD on the same supply put switching noise straight onto the driver's power rails. Separate supplies, or at minimum separate filtering.
- Encoder cable run alongside motor phase cable. The encoder signal is small and the motor cable is switching tens of amps. Bundle them together and you get intermittent position faults that appear only under load.
- Driver mounted in the same air stream as plasma or metal dust. Heat is one problem, conductive dust finding its way onto a board is another. The driver is usually the first thing to fail on a dirty machine.
There is a fourth one that is less obvious. Step and direction signals come from the controller at a logic level the driver has to accept, and different controllers deliver different levels - 5 V from some, 24 V from others, and a fair number of breakout boards in between. Feeding a 24 V signal into a driver expecting 5 V damages the input; feeding 5 V into a driver that wants more gives you a motor that misses steps at speed and looks like a tuning problem. Check the two datasheets against each other before the first power-up, not after.
Grounding deserves the same care. The signal ground between controller and driver should be a single reference, not a loop through the machine frame and back via the earth conductor. On a machine with a spindle VFD, a ground loop turns the step signal into an antenna and you get steps that disappear only when the spindle is off.
When closed loop is the wrong answer
If you need continuous torque across a wide speed range, holding torque at standstill for long periods with no heat rise, or you are chasing high accuracy rather than repeatability, closed loop steppers are the wrong tool and a servo is the right one. The honest split is roughly this: closed loop steppers buy you reliable positioning at a stepper budget, and they are excellent at it. They do not become a servo by adding an encoder.
If you are picking a kit for a specific machine, the useful information is the moving mass, the acceleration you want, the top feed rate and the frame size your bracket takes. With those four numbers we can point at one torque step instead of a range.
Tell us the axis and what it has to do, and we will come back with a fit - send the details through here. If what you actually need is a bigger frame or a servo, we would rather say so than sell you the wrong kit.
Common questions
Will a closed loop kit drop straight onto a machine that had an open-loop stepper?
Mechanically yes, if the frame size matches. Electrically the driver needs a step and direction signal, which almost every controller already provides, plus the encoder cable back to the motor. Budget an hour for commissioning and homing setup.
Do I need the same driver brand as the motor?
The driver has to match the motor's encoder type and current rating, not its brand. Practically, kits are sold together for that reason - mixing a driver and motor from different sources works only if the encoder interface lines up, and that is usually where the time goes.
Can I run a closed loop kit without the encoder connected?
No. It will either refuse to enable or fault immediately. The encoder is the control loop, not an accessory.
Why does my axis fault only at high feed rates on hot days?
Heat raises winding resistance and reduces the torque margin. If the axis is marginal, the position error grows until it trips the tolerance alarm - and a warm shop is when that shows up. Measuring the actual acceleration requirement usually finds the answer, though raising the supply voltage from 24 V toward 36 V often helps first.
