Neato XV11 Button problems (acting like crazy)

[snoozy]

Hi all,

Well, I have a neato with Rev 64 board. I repaired the LIDAR motor before and now I have a different problem.
When the battery gets low the buttons on the LCD board doesn't work as expected.
It's like when you press "menu" it does a "return" or sometimes it just does button presses by its own

I did some measurements on the LCD board and made I a pin out plan for the cable:

1__brown__buttons
2__red_____3,3V
3__orange__GND
4__yellow__SCLK
5__green___SDATA
6__blue____5V
7__violet___LED D1/D2
8__green___LED D1/D2
9__white___90mV (not used?!)
10_black___GND

It seems like they use different voltage divider to separate the buttons. I measured this levels:

SW1__1,9V
SW2__1,8V
SW3__1,6V
SW4__1,25V
not pressed__2,7V

Now comes the BUT:
The button signal is very ugly. It hase some noise and sometimes a lot cross talk coming from the I2C.
Damn, what a silly Idea to do a voltage divider on this buttons combined with this kind of layout.
They have a single layer PCB only and routed this signal all around the PCB...

I already added a 100µF cap parallel to the C10 capacitor (to improve the regulator voltage).

Any suggestions to fix this signals are welcome

[vic7767]

Without a valid schematic available you're not going to get any suggestions other than keep the voltage up. Evidently there is either a component installed that is out of spec or choosing the lowest bidder for parts will get you sometimes.

[snoozy]

Yeah, a schematic would be awesome...
I think I will track down, where the 2,7V are from (maybe µC?!) and I'm thinking about replacing all the caps.

[glnc222]

Just a guess, but the voltages you see on the buttons is typical of a digital logic input, such as on a cpu chip or gate chip, when not connected to an input signal, a certain transistor circuit used in the chip. Some switches work by grounding the input, sinking current from the logic input. Others might be driven high to the power supply rail. I believe the bumper switches, for example, ground the input. There can be pull up resistors on the input sometimes, and small debouncing capacitors. The cpu operates at some lower voltage than 5v, used for sensors, and maybe the LCD display power (you could check similar components in catalogs).
Noise in the power supply, caps and voltage regulator IC's, could be the issue. Unfortunately, the easiest way to tell is to have another working system with which to compare.
The tact switches themselves can be at fault. Some such switches have been found defective on bumpers and even opened up and cleaned by some users, instead of replacing.
Re resistor network, could be for current limiting to the logic input.

[glnc222]

Resistors on switches can be part of an R-C delay circuit used for debouncing the switch contacts -- suppressing oscillations at the switch closing and opening with a millisecond delay or something. You could look at how switches are interfaced to Arduino's, never read myself. Some keyboards have it all built in to specialized keyboard handling IC's. Don't know if built into the cpu equipped with various input types.

[glnc222]

A resistor network could be used to make a single analog input to a cpu serve multiple switches, by arranging different voltages to be presented for different key presses. The cpu reads the voltages and matches them to some table in software. No idea whether this technique was used. Sounds sensitive to noise in the power lines as it is not the usual wide division between logic signals high and low, with a gap, for zero and one representation. A lower than normal power supply would obviously shift all the switched voltage levels mixing up the switch outputs. Could be associated with some weakness in the voltage regulator. I think there have been some defects there.

[snoozy]

A resistor network could be used to make a single analog input to a cpu serve multiple switches, by arranging different voltages to be presented for different key presses. The cpu reads the voltages and matches them to some table in software. No idea whether this technique was used. Sounds sensitive to noise in the power lines as it is not the usual wide division between logic signals high and low, with a gap, for zero and one representation. A lower than normal power supply would obviously shift all the switched voltage levels mixing up the switch outputs. Could be associated with some weakness in the voltage regulator. I think there have been some defects there.

I think you are right. This are my thoughts on the function and the issue. Because the level are only 200 mV separate and created by a voltage divider is not very accurate, when the input voltage is not stable.

Actually I have a second neato here, but disassemble would be a lot work
So, first of all, I will check, where the 2,7V are created.

Thanks for your interest in this topic, I will keep you updated.

[robocleaner]

Could be associated with some weakness in the voltage regulator...

If Neato is dependent on such tight voltage control, this issue could easily be associated with one or more of the other 6 100uf/25v caps failing too (the OP having already replaced C10), as discussed in the "Behaving strangely as if confused thread" here: http://www.robotreviews.com/chat/viewto ... 6&start=20

[snoozy]

Ok, finally I found the problem: One of the buttons is really defect.

Well I should have seen that before, but the signal seems like to have cross talk on it, cause there is really something going on. I had never expected that a button can do something like this.
I attached you a screenshot. I didn't touch the button and it gave me a signal like that, continuously.

I ordered new switches and will tell you if it fixed the problem at all.
Thank you all for helping me out.

[robocleaner]

I had never expected that a button can do something like this.

Interesting. I have seen this before, albeit with higher voltage circuits/switches: Due to an unrelated defect elsewhere in a power supply, the voltage being supplied to the switch was considerably higher than the expected/rated voltage of the switch, and current was jumping the contacts. I'm not saying that's happening here at this low voltage, just that it can happen... and that the problem may still originate elsewhere.

[snoozy]

Replacing the return button helped. Its working a gain fine

Btw, I found the switches on ebay.de "Mikrotaster Mikroschalter AUS-(EIN) 4,5x4,5x3,8 mm 0,05A-12V Drucktaster mini"

Thank you all for helping!

[TechGuy]

Thanks for the update and the switch info.
Over the year, I collected/harvested a few of those mini switches from old mice before taking them to recycle.

[glnc222]

Hello snoozy,
What model oscilloscope was used to get the display in your post?