Dock-Based Run Length Clock

[glnc222]

GetLifeStatLog and GetSysLog gone from HELP list in Botvac.

[glnc222]

Indeed Statlog missing form Help list on Botvav.

Circuit Revisions

The Neato Wall Sensor proximity detector burned out once, and then another did so while adjusting the circuit parts dealing only with the clock, changing a resistor. They seem to be very delicate. So abandoned the IR sensor in favor of a simpler micro switch, easy to use on the Samsun Powerbot. ABS plastic bends with a medium clothes iron for a right angle bracket.
Another problem was that from time to time when arriving after a run, the clock would reset and run again instead of stopping. The button controlled is separate from the reset button on the clock, so there presumably is some internal state difference in the clock's circuitry when running. The pulse delivered may be over-voltage, so the drive voltage on arrival reduced 5 per cent 0.1v with a parallel resistor (lower does not trigger the clock, delicate balance). It is hard to reproduce the problem so only time will tell. More adjustments may be tried when the problem occurs again.

With the microswitch the filtering comparator component may not be necessary, replaced by simple RC delays on the flip-flop clock inputs. But it was already present to provide the delays and debouncing so kept.

To mount the clock 3M Command adhesive Picture Frame mounting strips, stronger than Velcro are used so removable. Angle bracket on the back, the clock is on a flap of duct tape over the back so can be lifted up for lifting the robot. Layered strips of foam mounting tape equalize the opposite side of the back to the right angle bracket fastening. The cardboard shown is to grip carpet pile with side edges.

Samsung Powerbot Charger

[edit] Actually two such switches are needed, one on each side, for when the vacuum parks at an angle (so at least one switch is pressed). Normally open, when departed, switches in parallel so either one can close. Another possibility is a photodetector placed over the Powerbot dock connected indicator light; see later posts.

[glnc222]

May 2021: a new version of this interface using a reed relay instead of transistor to operate the clock buttons is posted in the Hacking forum. http://www.robotreviews.com/chat/viewto ... =4&t=23011
Older experiments below.

It is proving difficult to set a voltage which makes the second simulated clock button push work. A graph of the pulse was finally made:

clock start/stop button pulse capacitive coupling

The actual button delivers the 1.5v battery voltage to the internal circuit. The pulse used has to be above that voltage to be at 1.5v very long, and is a triangle instead of a square shape. It works on the first pulse when starting, but when the peak is lowered near 1.5v, the second pulse does not always work the clock, and higher resets the clock.
Digital logic is not normally supposed to be exposed to inputs higher than the power supply, so that could be the problem. Moving forward I would try rebuilding the circuit using half of the comparator with an RC delay to make a nice square pulse at a constant voltage, more easily adjusted, instead of the crude capacitor coupling. Since the weird filter needed for the noisy IR detector is no longer needed. I might need to replace the comparator with a 555 timer/one shot. Big job will take a while.
[edit] Actually the existing pulse just needs to be fed through a comparator to straighten the edges.

[glnc222]

Revised version reusing comparator for clock pulse control. Seems to be working now. Not the simplest way but avoids rebuilding from scratch. Trick seems to be keeping voltage driving the coupling diode below the clock power 1.5v, despite any voltage drop across the diode (which varies with load).

[edit] Problem traced to 60 cycle line noise in mini A/C adapter, worn over time. Revised to use battery power only.
Refinements to clock pulse previously made, might not be needed.

The 7805 regulator is over-powered and bulky but was on hand. Smaller transistor like packages are available to fit more neatly.

Layout approximate from modifying previous versions.

Simplified version for Powerbot dock direct connection to LED indicator.

[edit] Late 2017, Continued problems with the second, stop clock button pulse resetting the clock intermittently seems to be solved by adding a bypass filter capacitor to the 1.5v button press signal, run off the clock's 1.5v battery with a PNP transistor simulating the button. Something in the internal state of the clock when running changes its sensitivity. Hard to tell, with difficulty reproducing, but one possibility is sensitivity to rapid onset of the pulse, needing a little delay, as some digital parts specify such caps to slow the onset in battery powered devices -- something in the internal junctions mentioned vulnerable. There is no A/C noise with battery supply as far as I can tell, and none seen on scope.
[edit] 12/17/17. Found 2nd, stop pulse problem: clock button 8K ohms, delivering 18 micro amps to contact from 1.5v battery. Reproduced seems to solve problem finally. Transistor not enough on 1.5v to use, diode isolation caused problem. See revised schematics above.

[glnc222]

The only way I could get the second, stopping pulse to work was use a relay, applying the clock power voltage, replicating the push button. The comparator sinks 20ma and operates small relays taking only 10ma or so. Contact if the final schematic is wanted. The older Trisonic clock seemed easier to control than the kitchen timer.

[glnc222]

Problem traced to 60 cycle line noise in mini A/C adapter used a long time, maybe worse recently. Revised circuit using 9v battery and 5v regulator, 1.5v AAA cell in clock. 500mah 9v battery 5ma load only when running, should last a year.
Revised diagrams edited in post above. Some more parts could be eliminated but already there,
[edit] small capacitor added to slow onset of button pulse, in case of sensitivity in some microcircuits noted in datasheets. Still had one unwanted reset.
Am waiting on a phototransistor to try over the Powerbot dock indicator light as another type of sensor. Too messy to tap the led inside the dock and risk any damage.
[edit]Final correction added pull up resistor to output transistor base since driven by open collector comparator, erratic without pull up (essentially disconnected in High state). Transistor alone insufficient to stabilize.

[edit] Seems the erroneous clock reset on the clock-stop pulse due to insufficient button limiting resistance, raised to 1K ohms to fix.

[edit] Late 2017, Continued problems with the second, stop clock button pulse resetting the clock intermittently seems to be solved by adding a bypass filter capacitor to the 1.5v button press signal, run off the clock's 1.5v battery with a PNP transistor simulating the button. Something in the internal state of the clock when running changes its sensitivity. Hard to tell, with difficulty reproducing, but one possibility is sensitivity to rapid onset of the pulse, needing a little delay, as some digital parts specify such caps to slow the onset in battery powered devices -- something in the internal junctions mentioned vulnerable. There is no A/C noise with battery supply as far as I can tell, and none seen on scope.

[edit] End 2017: seems to work rounding the button signal and 10K resistance in line with clock, simulating the button contact. Perhaps some debouncing circuit in the clock for the buttons plays some role, unknown. I previously had another type of clock, full function like a watch, that was easier to interface than the kitchen timer.
Revised schematics posted above.