Re: automatic pool cleaners.. in my opinion, are overpriced plastic toys designed to capitalize on a vertical market "keep my pool" sparkling clean wish. Nevertheless I fell to the notion and became a proud owner of a "dead in the water" Blue Diamond automatic pool cleaner. It had a sticker Water Tech on it and claimed to be manufactured in late 2003. So with the warranty long gone. I cracked it open.
Upon closer examination it revealed that one of the plastic cap retainers from a drive train assembly snapped, letting chlorinated water (grade 5 oxidizer) into the drive train and a controller compartment.
To get to the part, you need to open up and extract the filter and a protective plastic net around the pump and the drive train modules. Remove the top screws to free the pump motor. (do not open the pump motor housing, it is filled with oil and it will be messy. Should you do that by accident, close the housing up open up a rubber plug under the impeller and refill with non-detergent SAE20 or 30 motor oil, wipe off the residual spill clean from surrounding plastic parts and the motor with paper towels). Note the rubber plug under the impeller.
Once the controller board dried up and plugged into power the driver motor remained motionless while the pump motor would be spinning with no problem. The smoke, touch test revealed that one of the ICs was overheating to a burning point. Looking closely it appeared that the board contained several familiar elements. There was PIC16F84 and a number of voltage regulators for 5 and 12 volts, an opamp that was probably feeding signals from tilt sensors to the PIC, and the MC33033 brush-less 3-phase motor controller close looping throu. MC33039.. The controller would be driving powerful FETS in the half-bridge mode for each of the 3 phases. Suspecting soare me of the FETS blown (and this can happen if both of them remain open foencodedodod of time), I decided to measure the resistance between Gate, Source and Drain on all of them. Two of them showed less than 30 or so Ohm in each and every direction and they were in the same H-Bridge configuration.
The FETs are MTP50N06 and the corresponding P series MTP23P06, these are old but you still can find them on eBay. Mouser and DigiKey sell newer versions of it, I found RFP50N06 fits quite well. I got MTP23P06 cheap on eBay.
After lots of flux:
And wiped clean.
Upon closer examination it revealed that one of the plastic cap retainers from a drive train assembly snapped, letting chlorinated water (grade 5 oxidizer) into the drive train and a controller compartment.
To get to the part, you need to open up and extract the filter and a protective plastic net around the pump and the drive train modules. Remove the top screws to free the pump motor. (do not open the pump motor housing, it is filled with oil and it will be messy. Should you do that by accident, close the housing up open up a rubber plug under the impeller and refill with non-detergent SAE20 or 30 motor oil, wipe off the residual spill clean from surrounding plastic parts and the motor with paper towels). Note the rubber plug under the impeller.
Once the controller board dried up and plugged into power the driver motor remained motionless while the pump motor would be spinning with no problem. The smoke, touch test revealed that one of the ICs was overheating to a burning point. Looking closely it appeared that the board contained several familiar elements. There was PIC16F84 and a number of voltage regulators for 5 and 12 volts, an opamp that was probably feeding signals from tilt sensors to the PIC, and the MC33033 brush-less 3-phase motor controller close looping throu. MC33039.. The controller would be driving powerful FETS in the half-bridge mode for each of the 3 phases. Suspecting soare me of the FETS blown (and this can happen if both of them remain open foencodedodod of time), I decided to measure the resistance between Gate, Source and Drain on all of them. Two of them showed less than 30 or so Ohm in each and every direction and they were in the same H-Bridge configuration.
This is the board with the two taken out. You can see the tilt sensors in the white frame.
I
The FETs are MTP50N06 and the corresponding P series MTP23P06, these are old but you still can find them on eBay. Mouser and DigiKey sell newer versions of it, I found RFP50N06 fits quite well. I got MTP23P06 cheap on eBay.
The MC33033DW controller was KIA, so it too had to go. I used a hot air soldering gun at 280 Celcius plus a pre-heater made out of an electric skillet to de-soldBefore:
After lots of flux:
And wiped clean.
Soldered new MC33033DW back to place and replaced some of the charred resistors which connected the motor control IC to the blown MOSFET gates.
It also turned out that the rover uses mechanical tilt sensors. These are based on a metal ball rolling inside a cylindrical enclousre and shorting contacts depending on an angle of the unit. It looked like one of them failed, likely due to oxidation. I replaced those weith the newer units.