Urgent - Short-circuited mainboard

Greetings from Australia!

6 years ago I had a kite-surfing accident that rendered me paraplegic. I had always wanted a real robot to help me move things/myself around the house. When I saw Magni 3 years ago I thought that could be it and bought one with the help of David Crawley.

When the Magni first arrived I was able to power it up and move it around. A few days later it started having power/charging issues and was unable to power up. Then I myself was having trouble “powering up” so the poor Magni had been completely neglected since, until recently when my wife threatened to throw away the dead Magni if I didn’t do anything about it.

Hence the mission to revive the robot… The following is what I’ve done so far.

With the Magni fully assembled:

  • Charger – Good, the output when unplugged is 27V
  • With the charger plugged in, its LED2 says charging but LEDs on the switchboard next to the charging port are off regardless of the state of the switches
  • The voltage across the battery terminals is only a few mV when “charging”, about 3V when disconnected from the mainboard.

To isolate the problem, I disassembled the robot, disconnecting the batteries and the PI from the mainboard (version 4.9) so I could look for sources of short-circuit like bent pins touching etc. I didn’t find any obvious ones.

However, the resistance between pins 1 & 2 of the charging port (with the switchboard connected) was only 0.2 ohm (the switches made no difference). The switchboard seems fine when disconnected from the mainboard (LEDs lights up)

I also found the resistance across the 470uf capacitor below the PI is only 0.1 ohm. I wish I was confident enough to use my cheap soldering iron to de-solder it so I’d know for sure if the capacitor was the issue.

Hope my findings will help your guys pinpoint the shortcircuit. If the board is faulty, could I get a new board asap?

Thank you for your help!

Hi Again,

Some updates about the MCB might help us troubleshooting:

Since I couldn’t charge the batteries via MCB I connected the charger to the batteries directly and left them on charge overnight. Because the batteries were left in a discharged state for a few years, they were unwilling to accept charges with only 10mA charging current. Eventually, they got up to 25.66V.

When I connected the batteries to the MCB (without the switchboard), the voltage quickly dropped to 1.6V with almost 2.5A being discharged.

Using a DMM, the resistance between the battery terminals on MCB (without the batteries) is 11Mohms, open when in reverse polarity.

Using the DMM’s diode function, the reading is 0.494V forward biased and open in reverse.

Hmm well if it’s a dead MCB it seems to have died in a way that we haven’t quite seen before, perhaps @mjstn2011 will have some more insight. In any case, we should be able to RMA the board it I think.

Btw just to be sure, have you confirmed that the SD card hasn’t been corrupted (can happen if the robot isn’t properly shutdown) and the Pi is working fine when running standalone and disconnected from the robot (these tend to also be easily destroyed when mounting/unmounting or any kind of shorting on the mcb)?

Thank you for replying. Pi works fine standalone, powered by USB.

Also as I mentioned previously, the 470uf capacitor below Pi on the front side appears short-circuited between its legs (still soldered on). Can I remove it from MCB to see what happens?

I’m more of a software guy that has limited insight into the MCB’s inner workings, so I can’t say anything for sure, but it may cause issues with supplying the Pi 4, as it’s far more demanding than the Pi 3 for which the board was designed and that cap may be the only thing keeping the voltage stable enough when it’s not shorted. I suppose you can try and see if it works, it can’t get much worse than it is now.

You’ll have to wait for Mark’s input for a more thorough assessment.

Hello, Mark here. I can take up this topic now. Because this is a serious issue specific to your Magni I would like this to move to an ‘official’ support issue. Can you please send an email to support@ubiquityrobotics.com Then in the subject put Attn Mark - Failed MCB Board I am in California, USA so am ‘across the pond’ but have been down to Australia and even New Zealand twice, amazing country!

Thanks for the MCB version, 4.9, that is important for me to know for the context. I would like a picture to show the the full white tag on the front of the board and also see the capacitor in question which on rev 4.9 was called C106_M2. Some early board may not have a serial number tag but check please. If you cannot arrange a picture, please put in email the full serial number.

C106_M2, if I understand your description, was on early boards a problem and had to be laid down and glued. That issue has since been resolved on boards for some time now.

I am sorry to say that once lead acid gel cells have sat far below 9V for a very long time they generally have been degraded to a point where they cannot be recovered due to damage to their plates.

Your problem seems to be on the board and the batteries are a side effect or collateral damage as one might say.

In all of MCBs shipped I have seen 2 times now where one of the many larger ceramic caps are the thing that shorts which usually it then blows up leaving a big large black soot spot. So look around the board and see if you notice any obvious other area of damage and take a picture of that too if you do see such an area.

Sorry for the trouble going on here, we will sort it out.


Thank you, Mark, for taking up the thorny issue. I’ve sent 2 portraits of the sleeping beauty (front & back) to support as advised.