Automower - motor drive help please?

[nzmikec]

Hi All,

Seasons Greetings.

Im attempting to repair the PCB in a Husquvarna Automower - 305. She is an oldie but a goodie.

I purchased this unit used and shortly after owning it discovered a fault where the unit would fault out stating motor drive failure. I opened the unit up to find the PCB had some corrosion from moisture. I cleaned this using isopropyl and it came up looking good however the problem persisted.

I then replaced the motor with a known good one but the problem was ongoing. I also tested the original motor and it is working fine in another mower.

FYI the motors are 3 poll brushless - using 3 FET packages (A P and a N mosfet in each package). The motor also has a feedback encoder.

Using the mowers advanced menu I tested the motors. At lower speed settings it "cogs" and has trouble starting however at full speed it runs fine. I guess this is a function of a faulty H bridge?

After cleaning the PCB some more and inspecting it under the microscope I couldn't find any faults so I replaced the 3 FETs  (FDD8424H) (https://www.aliexpress.com/item/1005002602611647.html?spm=a2g0s.9042311.0.0.1fc04c4d1qIVLq) that correspond to that motors drive. Excuse the poor soldering, I didn't have any solder paste. The motor behaviour is the same!

Can anyone else suggest any troubleshooting or components to replace?

Thanks for taking the time to read my post.

Regards,

Mike

[BrokenYugo]

Hard to tell in the pics but it doesn't look good to me, PCB corrosion should be cleaned down to bright copper, look closely and you may find the copper gone on one or more tracks.

[abdulbadii]

a fancy equipment, must learn its mechanism  https://www.manualslib.com/manual/621334/Husqvarna-Auto-Mower.html?page=19

[nzmikec]

Thanks, yea the photos are not great. The PCB has a heavy conformal coating so I was hopeful it wasn't damaged and have looked over the traces around where the corrosion was pretty closely with a microscope and couldn't see any damage. Im not really sure what else could be wrong tho, will have a look again!

[fzabkar]

Could there be an encoder fault on the PCB?

[nzmikec]

Thanks everyone,

For the life of me I cant find the fault on this PCB. Think it needs a look with an oscilloscope but I dont have one of these.

Does anyone know anywhere in NZ that could take a look at repairing this for me? Cost is obviously a factor but willing to pay for an expert as I understand this isn't simple stuff.

Thanks,

Mike

[BrokenYugo]

Could we get a close up of the corroded/worked in area?

To get more to the point, if you have corrosion under the conformal coating or solder mask, you need to remove the coating and scrape everything down to bright metal to make an assessment of the traces and install patch wires if the traces are too far gone.

[nzmikec]

Thanks BrokenYugo,

Ive spent some more time cleaning conformal coating from that area and photographed. dropbox link to full size pics.

Also have a short video displaying the behaviour. I kinda think because it is running at 90 and 100% at least the mosfet driver part must be ok?

https://youtu.be/-ayUQKdCo2o
https://www.dropbox.com/s/tcr5yxvt1wu5dg4/IMG_6399.jpeg?dl=0
https://www.dropbox.com/s/iwf1rnyprsxdgd7/IMG_6400.jpeg?dl=0
https://www.dropbox.com/s/l19hvg3qy94x3qf/IMG_6401.jpeg?dl=0
https://www.dropbox.com/s/fm3nr4i7ttmw1qo/IMG_6402.jpeg?dl=0


Thanks for taking your time to look into my problem.

Regards,

Mike

[nzmikec]

Hi All,

Im yet to work out the issues here. I purchased my first oscilloscope to try and work out what is going wrong. It looks like im only getting occasional blocks of PWM power across all 3 phases at a time. When comparing this to the working sides phases, it has pwm signals throughout.

This is the non working side


This is the working side


And the test locations. Phase outputs were compared to battery negative.


Very happy to have any input on ideas of what might be causing this. Is it a motor feedback / encoder problem or a power driver problem?
Also very happy to have feedback on how to best test this type of circuit. This is literally the first time I have used a scope.

Thanks and regards

[cncjerry]

motors like that will cog like you said at low speed when one of the drive phases is defective.  They run smoother at high speed as they can use their momentum to get through the bad phase degrees of rotation.  You should be able to scope the phases and find the culprit.  Start with the drive signals.  I can't tell from the pictures if all the caps are fine.  Do all the caps have flat tops (not swollen)?  What is on the underside of that board?  Sometimes there are electrolytic caps on one side that leak and run thru the vias to the other side.

I assume you rung-out the leads going to the motor?

[nzmikec]

Hey thanks for the reply,

Re the motor, ive swapped it for a known good motor and same behaviour. also swapping side to side the problem stays with the driver.

The motor runs fine when the mower is put in test mode with wheels running 100%, its only at slower settings that it cogs. I was expecting to see one phase causing the issue but I see the same patterns on the scope with all 3 phases (I can only measure 2 at a time). This is making me think that the problem is with the feedback rather than the phase power drive parts.

At this stage im kinda just scoping different parts of the drive comparing it to the working side. Not getting anywhere fast unfortunately.

Thanks

Mike

[DavidAlfa]

The corrosion compounds can conduct electricity, and will be specially bad on higher voltage areas.
Thus, the mosfet gate and its driver could have been exposed to the motor voltages.

When there's corrosion on the pcb, no amount of alcohol will fully remove it, IPA it's a solvent, solvents don't dissolve metals, neither oxides, yeah you'll give a nice brushing, it's the mechanic action what removed it, but still you'll never remove all of it that way.

I always remove (desolder) the parts, then fully renew the solder on the traces so the flux makes some work on the corrosion.
That will help, but won't remove all of it, so pass a desoldering wick to remove all of the new solder, clean with ipa, then give a light brushing or fine sanding (I said light, with a small brass or glass fiber brush, or a grain 1000 sandpaper) to fully clean the area until you see shiny copper or tin.
Another IPA cleaning, now make some connectivity tests ensuring all traces are ok.

Take care of any color change on the solder mask (darker/lighter colour), expose the copper on those areas with the brush, as that's usually a sign of corrosion, some traces might look ok but be completely corroded under it.

Fix any damaged tracks, re-tin the traces/pads.
Also do the same with the removed parts if still in good enough condition.

Then put everything back together, again make continuity tests before applying any power.
If you got it working, apply some coating to protect the repair areas. Nail laquer is usually a cheapo and simple way.
If often exposed to humidity, better to apply some type of silicone.

[cncjerry]

does the motor stop completely if you disconnect the encoder?  is the encoder quadrature (able to tell forward vs backward)?  How many leads are on the enoder?  3 leads usually not quad, 4 leads or more quadrature.  Some encoders isolate the output from the input power.  There are all kinds of funky encoders out there. Do you have a signal generator that you can feed into the processor instead of the encoder?  You setup the signal generator for a square wave at approximately the correct frequency for slow speed operation and see if it impacts the motor cogging.  You should be able to start the motor at say 10% of test speed and feed in a dummy encoder signal of the approximate speed.  Use a resistors on the encoder lines, say 1k to keep from overdriving the processor.  Dial it around a little and see if it clears up the cogging.  Cogging goes away then your problem is the encoder.

If it is a quadrature encoder, and encoders are notorious for going bad, usually some oil migrates from the shaft into the optics, you can setup some switches maybe or a simple dual hall effect encoder with some small magnets on the wheels held on with tape or glue.  Depends on how badly you want to fix this thing, the board looks really bad but it can be repaired once you find the bad drive signal.  The processor moves the motor and then looks for a pulse.  I doubt it cares of the timing of the pulse within limits, that's why I think putting a signal into the encoder lines might help determine motor drive vs encoder.

I thyink you missed my point.  High speed motor operation of a switched phase motor or stepper is usually not how they are tested.  Slow speed is where the problems will show (within limits).  At high speeds, the enocder pulse might look better depending on how it is generated, especially if it is generated using hall effects.  Some of these setups just have a sensor that reads the points of a drive gear.  The faster the better for a cleaner pulse, again within limits.  Cogging goes away at higher speeds because the missed drive signal can be carried by the motor momentum.  I'm not suggesting the motor is bad, the drive circuit is bad or the encoder.  you can solder little wires on the drive signal pins and attach your scope to the wires.

one other point, an open drive line to the motor usually will have a higher voltage than the ones that aren't open.  So look for a signal from the drive that is higher than the others.

[Fasse]

I have a similar problem but with the cutter motor.  Did you ever find the fault?