Add feedback to a gearmotor
Four different requirements all get called feedback. Three of them do not need an encoder.
Decide which question you are answering before choosing a device, because the answer changes the resolution, the placement and whether you need a device at all.
What feedback is actually for
In order of how often the requirement turns out to be the real one.
- Fault detection
- Did it move, did it stall, did it jam, did it reach the end. This is the most common real requirement and the cheapest to satisfy. A single index pulse per revolution, a limit sensor at the position that matters, or a current threshold on the drive will answer it. Many programs specify a full encoder and then use one bit of it.
- Speed regulation
- Holding a speed against a varying load. A tachometer signal or a low-resolution incremental output is enough, because the loop is integrating over many counts. Resolution matters far less than update rate and signal quality.
- Position control
- Knowing where the output is, and putting it somewhere specific. This is the requirement that justifies real resolution, and it is also the one where the placement of the device decides whether the number means anything.
- Commutation
- A brushless motor needs rotor position to energize the right phase. This is not optional and it is not the same device as a position encoder, although one device can sometimes serve both. If the motor is brushless, this feedback exists whether or not the application asked for any.
Resolution, and the resolution you actually get
An encoder on the motor shaft measures the motor. The gearhead sits between it and the load.
counts per output revolution = 4 × ppr × R
- ppr
- encoder pulses per revolution, per channel
- 4
- quadrature decoding of two channels, counting both edges of both
- R
- gear ratio
Worked from the published Micro Motion feedback record: MM-EM08-012 is a 12 ppr, 2-channel incremental unit. Decoded in quadrature that is 48 counts per motor revolution. Behind a 100:1 gearhead it becomes 4800 counts per output revolution, or about 0.075 degrees per count. That is a calculated figure, not a published accuracy.
The distinction matters. Resolution is how finely the count divides a revolution. Accuracy is how close the output actually is to where the count says it is, and everything on the backlash list sits between the two: gear clearance, transmission error, torsional windup and coupling lost motion. A control loop reading a motor-mounted encoder can be perfect and the load can still be a degree away. If the number has to be true at the output, the measuring device belongs at the output.
f_signal = ppr × n / 60
- f_signal
- output frequency per channel, Hz
- ppr
- pulses per revolution, per channel
- n
- motor speed, rpm
Signals, cable and connectors
The electrical detail that decides whether the feedback works in a machine rather than on a bench.
| Characteristic | Value | What it constrains |
|---|---|---|
| Output signal | TTL compatible | Single-ended logic. It has no differential noise rejection, so routing and length matter. |
| Supply | 5 V ±5 % | A tight tolerance. Do not share a rail that sags when the motor starts. |
| Supply current, maximum | 20 mA | Small. Easy to budget, and easy to disturb with a noisy rail. |
| Output current, maximum | 3 mA | This is the real limit. It cannot drive a terminated line or a long cable. It wants a nearby high-impedance receiver. |
| Maximum frequency | 20 kHz | Sets the speed ceiling for the count. Rarely binding at low pulse counts. |
| Cable | AWG28 flat band cable | Fine gauge and unshielded. It is a signal cable in a machine that also carries motor current. |
| Operating temperature | -10 to 60 °C | Narrower than most motors. In a hot enclosure this can be the component that sets the limit. |
- Keep the signal cable away from the motor leads
- A brushed motor is a broadband noise source and a PWM drive is a narrowband one. Running the encoder cable alongside the motor leads, or through the same connector body without separation, is the most common cause of phantom counts. Separate them physically, keep the return close to its signal, and if the run is long enough to worry about, buffer it near the motor rather than at the controller.
- The connector is the failure point
- In a vibrating mechanism the board connector on a fine flat cable outlasts nothing else on the assembly. Strain-relieve at both ends, and support the cable within a short distance of the motor exit so the flex happens in the cable and not at the solder joint.
- Budget the added length before the layout is fixed
- A rear-mounted encoder adds axial length. The published record gives 3.5 mm added to its fitting platform, for a combined 27.8 mm overall. Find that space at the concept stage, not after the housing is tooled.
- The rear shaft is now occupied
- A rear-mounted device uses the rear shaft extension. If the design also wants a manual override knob, a brake, or a second output there, those requirements are in conflict and it is better to discover that now.
The compatibility question
Feedback is not a universal accessory. It fits specific platforms, in a specific way.
Bring the control question, not just the encoder question
Which of the four purposes you actually need, what the loop runs on, and what accuracy has to be true at the output. That is what decides the device and where it belongs.