The next question in Panterla’s design is how to turn a loaded processing drum quietly, reliably, and with as few parts as practical. The target is 50 and 75 rpm, with repeated stops and reversals. That sounds like a small motor problem. It becomes more interesting once the drum contains liquid and the drive has to fit into a machine that someone will actually assemble and maintain.
My starting choice is a NEMA 17 stepper, the TMC2209 driver already available for the project, a short direct connection, and a sensor that confirms the drum itself is moving. An integrated brushless servo is the next candidate if the stepper fails the load or sound test. A small geared serial servo is useful as a comparison, but its impressive stall-torque figure does not establish continuous drum duty.
This is a selection study, not a completed motor experiment. The loaded torque, comparative noise, and endurance measurements are still ahead. The aim here is to explain which prototype makes sense first and what evidence would justify changing it.
Start with the motion, then count the parts
The drum needs slow, continuous, bidirectional rotation. Its motor and electronics belong on the dry side, while the drum or intermediate shaft needs its own supports. The motor bearings should not carry the drum’s weight.
The complete drive matters more than the motor alone. A compact motor that requires a gearbox, two adapters, a separate controller board, another supply rail, and several adjustment points may create a larger assembly than a less integrated alternative.
For the first prototype, I want one motor on a local bracket, one coupling or drive adapter, and the existing controller. A 1:1 connection is the starting point. A belt is an option when measured torque, alignment, packaging, or vibration demands it; it also adds pulleys, tension adjustment, and service work.
The same logic applies to feedback. Start with the information the machine needs: confirmation of actual drum motion. A high-resolution encoder becomes worthwhile when faster fault detection, direction confirmation, or detailed speed measurements require it.
What a servo actually adds
A servo is a system: motor, sensor, power driver, and a regulator that compares commanded motion with measured motion. It can use a brushed motor, a brushless motor, or even a stepper.
The ordinary stepper plus TMC2209 arrangement is open loop mechanically. The controller sends STEP/DIR commands, and the driver regulates winding current. If the motor loses synchronism under load, the command count no longer proves that the shaft completed the movement.
The ST3215-HS and DDSM315 contain their own motion feedback and regulator. When load increases, the regulator tries to correct the speed error within its limits. A servo can still stall; feedback gives it information about the failure rather than unlimited torque.
An internal motor sensor also cannot prove that a downstream coupling is intact. For every candidate, Panterla still benefits from measuring movement at the actual drum.
Three torque figures that cannot be ranked together
The headline numbers describe different operating conditions:
| Candidate | Published torque | What the number describes |
|---|---|---|
| LDO-42STH47-1684A stepper | 4.4 kgf·cm, approximately 0.431 N·m holding | An energized, stationary rotor at the specified current |
| Waveshare ST3215-HS | 20 kgf·cm, approximately 1.96 N·m stall at 12 V | A locked output at zero speed |
| Waveshare DDSM315 | 0.55 N·m rated at 200 rpm | A stated operating point, with an 18 V rated supply |
The figures come from the LDO 42 mm series specifications, ST3215-HS documentation, and DDSM315 documentation.
Holding torque does not establish a stepper’s running torque at 75 rpm. Stall torque does not establish how much torque a geared servo can supply throughout a development stage. A rated operating point is more useful for transmission calculations, but it still does not describe every lower-speed or thermal condition in the finished machine.
Choosing the largest number in this table would therefore be a sizing mistake. The relevant number is available continuous torque at our operating speed, with enough reserve for friction, startup, and reversal.
Why the stepper is the first prototype
A stepper at these drum speeds has no inherent need for an internal gearbox. It also has no brushes. That removes two possible sources of noise and wear before any acoustic tuning begins.
For a 1.8-degree motor, the command rates at 1:1 are modest:
| Drum speed | Full steps per second | STEP pulses per second at 16 microsteps |
|---|---|---|
| 50 rpm | 166.7 | 2,667 |
| 75 rpm | 250 | 4,000 |
The limiting questions are mechanical torque, resonance, acceleration, and temperature. Generating those pulse rates is a straightforward task for the ESP32-S3.
The TMC2209 datasheet, revision 1.09, describes StealthChop2 for quiet operation and SpreadCycle for more dynamic operation (PDF pages 4, 38, and 44). I will compare both on the loaded mechanism. StallGuard load information is useful, but it is not a mechanical position encoder.
The documented reference motor is LDO-42STH47-1684A: four leads, 1.68 A rated phase current, 2.8 mH inductance, and a 5 mm shaft. The available LDO sheet does not provide a running torque curve for our operating points.
A STEPPERONLINE NEMA 17 has already been ordered. Its exact model and manufacturer data still need to be confirmed, so I will inspect that unit before buying the LDO reference motor. The order title’s “59 Ncm, 2 A” is identification information, not a verified operating specification.
The present conservative prototype target is approximately 1.2–1.4 A RMS per phase, subject to the actual motor, carrier board, cooling, and measurements. It is a project starting range, not a universal TMC2209 module rating. RMS and peak current must be distinguished when setting the driver.
A nominal 24 V bus can supply the motor through the current-regulating driver even though its winding voltage is much lower. The motor must not be connected directly to that bus. Current settings, supply transients, and the final enclosure temperature all need verification.
The integrated brushless alternative
DDSM315 combines a brushless motor, sensing, and a servo drive without an internal reduction gearbox. It uses RS485 and provides motion and diagnostic feedback. Its published rated point makes it a useful alternative when the tested stepper has insufficient margin.
The first DDSM layout should also be direct 1:1 if the measured load permits it. Its 200 rpm rated speed does not require us to run it at 200 rpm. However, the published 0.55 N·m point is not a guarantee of continuous performance at 50 or 75 rpm; low-speed regulation and temperature need testing.
The documented operating range is 12–24 V, with an 18 V rated point. A suitable existing 24 V branch may avoid another supply, provided the installed configuration is validated. Calculations based on the rated point must retain its stated conditions.
When reduction earns its place
If direct drive lacks torque or the layout needs a belt, an 18-tooth motor pulley and a 48-tooth drum pulley give:
Reduction = 48 / 18 = 2.667
Motor speed at 50 rpm drum = 133 rpm
Motor speed at 75 rpm drum = 200 rpm
Estimated drum torque = 0.55 × 2.667 × 0.90 = 1.32 N·m
The 90% transmission efficiency is an assumption. The approximately 1.32 N·m result is a screening calculation at the published motor operating point, reached at 75 rpm on the drum. It does not prove that torque across the full speed range or in a warm enclosure.
The larger DDSM115 is another possible direct-drive candidate if size permits. It belongs in the next comparison only if the first prototype leaves a real torque or packaging problem unresolved. This is a sequence of conditional tests, not a shopping list for every motor discussed here.
Why the small geared servo remains an experiment
ST3215-HS is attractive because its controller, encoder, and gearbox are inside one compact module. Its approximately 106 rpm no-load speed at 12 V makes a 1:1 drum prototype plausible, although no-load speed does not establish loaded performance. It supports continuous rotation with speed feedback and uses a half-duplex TTL bus.
Its gearbox brings potential gear whine, backlash, and reversal impacts. The reviewed manufacturer material does not publish a continuous-torque rating, gearbox life, or a torque-speed curve. The 20 kgf·cm headline cannot answer those questions.
A long, loaded test could show that it is perfectly adequate. Until then, it is an integrated comparison candidate rather than a proven final drive. Its supply range is 6–12.6 V; it cannot share the stepper’s 24 V branch directly.
Quietness belongs to the whole assembly
The audible result includes the motor, drive mode, bearings, transmission, mounting panel, liquid movement, and motion profile. A quiet motor bolted to a resonant enclosure wall can produce an unpleasant machine.
I will start with a stiff local bracket, correct alignment, tuned current, microstepping, and smooth ramps. Isolation or a belt becomes useful when a repeatable measurement shows which vibration path it needs to address.
Every candidate needs a controlled reversal:
- Ramp down to zero.
- Allow a short pause if the process permits it.
- Ramp up in the opposite direction.
An immediate command from +75 to −75 rpm asks the drive to absorb an avoidable liquid and transmission shock. Closed-loop control does not remove that inertia.
There is currently no controlled acoustic ranking of these drives in Panterla. Manufacturer “silent” claims and sound figures without comparable mounting, distance, speed, and load cannot establish a winner. The saved DDSM315 product and wiki material also disagree on rated current and noise figures; those discrepancies remain open and must be resolved before final power sizing or acoustic claims.
Measure the drum, not just the motor
A magnet or optical mark on the drum and one stationary pickup can provide a compact motion monitor. With one pulse per revolution:
- at 50 rpm, pulses arrive every 1.2 seconds;
- at 75 rpm, pulses arrive every 0.8 seconds.
That can confirm sustained rotation and estimate average rpm with a suitable startup grace period and timeout. It cannot provide rapid sub-revolution fault detection, fine speed-ripple measurements, or independent direction confirmation.
More marks, a second sensing phase, or a full encoder should follow the required detection time and process information. A temporary high-resolution encoder may be useful for qualification even if the final machine uses the simpler pickup.
For the stepper, the initial sensor can be a monitoring channel: disagreement between command and measured movement stops the process and reports a fault. It does not need to become an aggressive position-correction loop.
The loaded test that decides the drive
The THD Lift inspection showed a motor attempting to turn while the failed transmission stopped motion from reaching the drum. The continued teardown adds the wet/dry boundary and service-access requirements. These are reasons to measure actual drum motion and protect the complete drive, rather than rely on commands or motor feedback alone.
The JOBO CPE2 test also showed why mechanical layout is a thermal decision: the drum initially rotated above the bath surface before the water level was raised. A replacement coupling and supported shaft must preserve the intended bath contact, lift path, and preheat geometry. That temperature test did not log loaded rpm or validate any of the candidate motors.
First, install the actual supports and the heaviest intended drum, filled with water to the maximum intended process load. Measure breakaway force at a known radius in both directions:
Torque [N·m] = tangential force [N] × radius [m]
That gives a breakaway measurement. Steady running torque needs a controlled measurement while the assembly moves at the target speeds; a static pull alone cannot establish it. Repeat after warm-up and in the least favourable drum position.
The motor then needs to pass a complete worst-case process cycle, including repeated reversals. Record:
- actual drum rpm and agreement with the command;
- startup, stopping, and reversal behaviour;
- supply current and phase current where measurable;
- motor, driver, and connector temperatures;
- background sound and operating sound at a fixed operator position;
- tonal noise, clicking, and vibration after warm-up;
- response to safely simulated increased friction or obstruction.
Use the same drum, load, supports, enclosure, and microphone position when comparing drives. Phone recordings can help compare tonal character, but uncalibrated phone readings should not be presented as absolute sound-pressure measurements.
The provisional project sizing rule is at least twice the measured steady drum torque in verified continuous capability. Breakaway and reversal peaks need their own short-term margin. Neither holding torque nor stall torque can satisfy that continuous criterion by itself.
Motion loss needs a timeout and a means to remove drive power. Leaving any candidate energized against a jam converts a motion fault into a heating problem.
The lift can use a different motor
The drum rotates throughout a process stage. The lift moves briefly between known positions. For the existing lift pivot, an integrated serial servo such as ST3020, one output arm or short link, and actual endpoint sensing may create fewer parts than a new screw-driven axis.
The resting mechanism should support the load quietly. A suitable counterbalance, latch, or other mechanical restraint must be selected for the real geometry; active servo torque disappears when power fails. The servo output should move the lift without carrying its structural weight.
A stepper remains sensible if the lift already uses a suitable screw mechanism. Sharing a motor type is useful only when it simplifies the complete machine.
Current decision
The first build is the inspected NEMA 17, existing TMC2209, shortest practical direct drive, and actual drum-motion pickup. I will tune it and test it under load before adding another transmission stage or a larger motor.
If that assembly fails the torque, temperature, or sound requirements, the next comparison is an integrated DDSM at 1:1. The calculated 2.67:1 belt arrangement remains an option when measured demand justifies it. ST3215-HS is worth testing when its integration could resolve a remaining tradeoff, with its continuous duty still to be established.
The final choice will be the simplest complete assembly that passes those tests. A quiet direct stepper would be a good result; a brushless servo that removes a demonstrated limitation would also be a good result. The loaded machine will decide.
Sources and related work
This article combines the project’s drum-drive comparison with the revised actuator-selection report dated 1 October 2026. The revised report sets the current direction: reuse available hardware, test direct drive first, and add reduction or more detailed sensing when a demonstrated requirement calls for it.
- LDO Motors: 42 mm hybrid stepper series — electrical, mechanical, and holding-torque data; the archived information sheet identifies the LDO reference variant.
- Analog Devices: TMC2209 datasheet, revision 1.09 — operating modes on PDF pages 4, 38, and 44; current and thermal requirements on pages 75–80.
- Waveshare: ST3215-HS — speed, supply range, feedback, and locked-rotor specifications.
- Waveshare: DDSM315 and product page — operating point, control interface, and the specifications to reconcile before final selection.
- Waveshare: DDSM115 — the larger integrated direct-drive alternative.
- Waveshare: ST3020 — the serial-servo candidate for intermittent lift motion.
For the thermal side of the same machine, see the direct-heating experiment. The THD Lift assessment documents the existing mechanism that informs this rebuild.