• Deduction about the Maxon motor that we're using on the RotationalFrictionDisplay (we found the motors "around the lab").

The Motor is probably 118458 (stall torque 1.54 mN-m) or 118443(stall torque 1.54 mN-m), as they both have the same terminal resistance of 32.6 ohms and a max continuous current of 158 mA.

These are part of the RE13 13mm Precious Metal Brushes series.(0.75 Watt)

The other possibility is 268348, from the RE-max 13 series. The terminal resistance and max continuous current for this is similar.

The encoder is a standard magnetic encoder from Maxon with 16 counts per turn, is what the data sheet says. (Im assuming this means 16 lines, and hence 64 counts). No. 110778 http://www.maxonmotor.com/docsx/Download/catalog_2004/Pdf/04_245_e.pdf

The gear head is about 20mm in length which makes me think its 17:1 reduction. Part Number 110314 http://www.maxonmotor.com/docsx/Download/catalog_2004/Pdf/04_194_e.pdf

We are now using a E4-300-184-T 300 CPR encoder from US Digital and a 16mm Maxon motor with 5.4:1 planetary gearhead.

To control the motors we used a Sensoray 626 PCI control card and Copley Motion Systems model 403 amplifier for more information on other older Copley amps see the following Legacy Amplifier page on the Copley site.

On this device we are using US Digital E4 encoders with either 120 or 300 ticks (x4) per revolution.

-- WillP? - 16 Nov 2004

• Sensoray_626InstructionManual_F.pdf: Sensoray PCI control card manual

Sensoray_626 Control Card ($500) Six versatile 24-bit counters for encoders/timers with interrupt generation Battery backup of counters Four 14-bit D/A outputs, 20 kHz update rate Sixteen 14-bit differential A/D inputs, 50 kHz rate Watchdog timer may be used to reset PCI bus 48 bi-directional digital signals for solid state relay boards, 20 with edge detection & interrupts QNX driver PCI bus

In controlling the RFD, we've run into a problem while tuning a PID controller to take a step input or track a sine wave. With lower gains, the system is stable; however, tracking is poor. This is not acceptable considering that part of the point of our initial experiments is to see how sensitive we are to actual magnitude of rotational friction/skin stretch. Therefore, the controller for these experiments must command the desired position with very tight accuracy and must also exhibit critically or overdamped behavior, since it would feel unnatural to have the rotaional friction forces oscillate when this is not expected.

The RFD was initially driven by a RE13 motor with 17:1? gearhead and 20:1 worm gear. The motor used a 300 tick E4 encoder from US Digital. This design was quite sluggish and was replaced with a RE16 motor with 5.4:1 planetary gearhead (same encoder). This design had a better step response, but we ran into a problem when increasing the gains -- the device went unstable. When this occurred, the motor would spin wildly and the on-screen angle output (based on encoder and reductions) for the friction disk would hardly increment. It seems that at very high motor velocities, that the encoder count was not being incremented. We believe that this is a hardware limitation with the encoder register on our control card. The proposed solution is to get rid of the planetary gearbox on the motor, so that the motor speeds can be reduced for the same output speed. In addition, we will also switch to a 120 tick E4 encoder. The combination of these will reduce the pulse frequency about 15x.

* This actually turned out to be a limitation on the E4 encoders, not the Sensoray card!!!!

But, in looking at the "limit" specs. on the Sensoray 626 card (p.27), I find that the minimum captured width pulse is 250_ns (or 4MHz). Specifically for the encoder, it specifies a max rate of 2_MHz in "internal timer mode" and 500_KHz in external driven mode. So if we are to take the slowest of these we should be able to count 500,000 pulses/sec / 1200 pulses/rev * 60 sec/min = 25,000 RPM. So 25,000 RPM should have been acceptable with the 300 tick encoder in quadrature, which exceeds the motors maximum speed, so something isn't quite right. Perhaps something else is causing us to loose counts at high speeds, like perhaps high capacitance due to long encoder cable runs or inductive pickup from the amplifier lines running next to the encoder lines.

New motor
A-max 16 (110052)
Stall Torque: 3.62 mNm (about half of the 1524 from Micromo)
No load speed: 9,700 RPM

• Amax16_110052_1.2watt.pdf: A-max16_motor_datasheet