Day Two of a Five-minute Job

Yesterday I began a task which led to several others - today I got back on track with what I started.

Reassembling the wheel unit was the first task.  This proved to be difficult - removal of the axle from its bearing to take the wheel off yesterday had left the face that accepts the motor shaft deformed.  Again, one of the lab staff was a great help, and had the mounting hole redrilled to an excellent fit in no time at all.

Wheel Sensors

Back to the original task, I verified the sensor output levels.  I had to reduce the LED current a bit to prevent the sensor saturating.  Applying the other wheel's new disc was much simpler - I noted that the plastic wheels are a tight push-fit, and can be removed much more easily than the axles from the bearing.  After fitting this, I tuned the left wheel sensors, and managed to get a very good quadrature output.

I copied the circuit for the right wheel sensors, and attempted to tune them.  I found that the load resistors needed to be much higher for this wheel - initially I thought it could be ambient light interference, but a simple wave of the hand disproved this.  To account for this difference, I'll put variable load resistances on the final board.

After tuning the right wheel, I redrew the schematic in my lab book as a record.

LEDs and Waste Energy

It's worth noting that LEDs can amount to a large power drain on a system, particularly when it involves a linear regulator.  The original connections involved the four sensor LEDs (seven if you include the floor sensors) drawing about 10mA from the +5V rail (via a regulator from the +12V rail), via resistive current limiters.  Noting that the LEDs drop about 1.3V at 10mA, and that the remaining 10.7V at 10mA must be dropped by the resistors (3.7V at 10mA), and the regulator (7V at 10mA), it's not hard to see that the system is quite inefficient, and will contribute a large proportion of the heat dumped from the regulator.

To counteract this problem, it's possible to wire at least all of the wheel sensor LEDs in series - 1.3V x 4 gives 5.2V - still at 10mA.  By driving the LEDs from the 12V rail via a limit resistor, the losses are reduced, and the regulator doesn't need to dump any waste energy on behalf of the LEDs.  Theoretically this can be done with the floor sensors too, if they require the same current - the total LED voltage would be about 9.1V, which actually only leads to an increase in efficiency.  Note, however, that when the LEDs operate near the minimum battery voltage, the current will not be very stable due to the lack of voltage on the resistor.

Ideally, the LEDs would be driven by a switching regulator, which should have next to no losses.  An improvement on a simple load resistor would be to use a linear current regulation circuit, which would guarantee stability even when low voltage exists across the current limiting element.