Neato slow motion turn during dock & stop in front of dock

[TechGuy]

I am working on a Neato XV Signature which does not dock properly. Normally, Neato turns around at the dock and move slowly to park at the dock. But this Neato turns around in slow motion and then play the parked/charging happy tune and stop without touching the dock.

I found a youtube video. This Neato is doing the same thing.


- The batteries are about 9 months old and still run this Neato for an hour before charging.
- I found another video on the BotVac Neato doing the slow motion turn too. So, this fault is not limited to XV Neato.
- During charging on the dock, I move it a bit away from the dock and it does not move back to the dock as a normal working Neato does. This Neato is thinking that it is still charging - confirmed with my next test.
- Something is going on with this Neato. I removed it from dock and select Support --> Shutdown. It gave me an error "I can't shutdown while I am connected to power".
- I did my tiny screwdriver trick. But it did not fix it. This Neato has a rubber plug sealing the charging jack. Therefore, dirt is not an issue to keep the jack switch contact open. Is there any chance that this issue is due to capacitors? This is just a wild guess.
viewtopic.php?f=1&t=10350&hilit=charging+jack&start=40#p144005

Past repair: This Neato had a faulty charging mode while it was not charging. I replaced the diode(D??) 9 months ago. At the same time, I got a new set of batteries for it. After 9 months, this slow motion turn during docking procedure.

[glnc222]

The message on Shut Down that it is connected is the indicative fact. This probably means some component in the charging circuit on the system board is leaking voltage onto the charging contacts or that part where the dock voltage is sensed by the cpu. Some diode or mosfet switch defective, a short, whatever. The battery is supplying the voltage back into the charging input not properly switched confusing the software.
I encountered that when tinkering with an add on over-voltage charger cut off in "lithium ion revisited" thread.
There is a partial schematic there at least of the older system boards. There were some units where the charging plug was still there but not usable, until the whole thing was eliminated.

[TechGuy]

I am going over this thread. viewtopic.php?f=20&t=17617&hilit=flashing+charging+light+D14&start=20
After 9 months with the new D14, the D14 is slowly dying again. But not to the point showing constant flashing light.

[glnc222]

CPU voltage detection is a high impedence input like a powered volt meter, so very little current leakage back into the circuits is needed for the system to see a false dock voltage reading. A diode supposed to block such back currents would not have to be completely shorted to create a voltage detection where not wanted. The software also may not require a high voltage to believe it is on the dock. The software does use some analysis of the dock detection to pull away from bad connections which can heat up the contacts melting plastic, why it wiggles backing in etc. Older system board units have a half watt resistor across the charging contacts for a load to activate the power management IC inside the dock, signaling connection to the dock, but newer system boards with the newer chargers eliminated that hot resistor with some improved method.

The situation seems similar to the "locked in charging mode" from the diode D14 failure, and since that diode was replaced maybe try again. The very fact the problem was heating could imply wear on that part even when initially working. I think some work added a heat sink to the mounting, common with parts operating at high currents.

Another aspect is mosfet switch control where the switching input is into a capacitance, not a bipolar semiconductor junction, so very little current can be involved more easily distorted by some fault. The system has mosfets to manage the power sourcing between the battery, system, and dock.


USB reports the detected dock voltage which might be compared before and after component replacements.
I have not analyzed the charging circuit schematics traced by a member posted.

[TechGuy]

There is a large area of copper foil under the cathode of the MBRS540T3(5A 40V) SMD diode. But the underside of the SMD diode does not have any metal heat sink material that I can solder it onto the copper foil on the PCA.

I have axial type diode B560 (5A 60v). I can try the axial diode instead of the SMD diode.

[glnc222]

This is the heavier duty version posted, with some built on heat sinking and for bolting to more. TO-220 package.
Post http://www.robotreviews.com/chat/viewtopic.php?p=128251#p128251

MBR735 https://www.jaycar.com.au/diode-mbr735-schottky-35v-7a-to-220-2/p/ZR1029
from the usual suspects Digi-Key, ebay etc.
I don't know if the original is this shottky type, one would think so, different characteristics.

[glnc222]

This thread, incident, reveals yet another software detail missing in vacuum robots.

Apparently the Neato fails to dock, mysteriously giving up right in front of the dock, so close to finishing, because of a faulty indication the dock is already connected. If the robot bothered to integrate data from the lidar it would see that it is not docked, and should show an error message. Yet such a message does appear if trying to shut down the computer, finally revealing the faulty circuits, because not permitted when on the dock (apparently not handling the switching of power supply fully enough to shut down there, perhaps too expensive a complication in circuit boards for such an obscure situation).

The mysterious failure to dock when right in front has been reported several times and I observed it once here. Why give up when so close? That a hardware defect could cause it never comes to mind because one assumes the software is adequate. Another failure to utilize navigation info was when the Neato was stuck under an over-head pinch trap on carpet, where the wheels could slip, and it just keeps going (in one case tearing up treads on a kitchen floor caught under an open dishwasher door). I checked this on a possibly improved later Botvac by stepping on it, and it would not give up and shut down; no improvement there. The Lidar seems to be a Rodney Dangerfield -- gets no respect.

Yet be careful what you wish for. On the Samsung, the slightest error causes a shut down instead of allowing time to manually release the robot from a trap. It is hard to understand the thinking of these programmers, besides maybe they are not given enough time and resources to thoroughly do these jobs. It all seems a bit casual. Npt acceptable behavior in a major appliance.