Samsung EcoBubble washing machines

[shakalnokturn]

They seem to fail young and in numbers.
A lot of the failures seem to be about electronic control board that goes bad, of course these cost 1/2 the price of the appliance to replace and I guess to be no more reliable than the original.

But what actually fails on the control board?
Most of the reports are of stopping in the middle of the cycle or motor not spinning at all.
The one that brought me here a WW80J5555DA won't spin the motor, it's a universal motor based model, the problem is not the motor.
Any washing cycle can run forever, the motor will not spin but no error code is triggered...

I went as far a de-potting the control board to check the things I could think of.

680nF AC input noise filter capacitor was a little low.
Power supplies are good including ripple.
Zero-cross detection works.
All >100k resistors test good.
All relay contacts test good.

When testing in the machine there is no output to fire the triac.

The strange thing is that I cannot enter diagnostics mode, when I do the display shows "-:" dial and buttons are unresponsive except for on/off.
If I create an out of range situation on the water level sensor it will trigger an "0C" fault, still entering diagnostics mode from there will not allow to run any tests as it should.

Defective CPU or simply a corrupt EEPROM?
If anyone has a working dump for a similar machine (control board DC41-00230B) I'll give it a try... It seems a shame that all these machines should end in parts when there may be an easy cure.

Edit:
Added schematic. Although it isn't for my model it looks very similar, component numbering differs.

[Gyro]

...
I went as far a de-potting the control board to check the things I could think of.

680nF AC input noise filter capacitor was a little low.
Power supplies are good including ripple.
Zero-cross detection works.
All >100k resistors test good.
All relay contacts test good.
...

I'd lay odds on that 680nF capaciitor being the series dropper in a capacitive PSU rather than a noise filter. 'A little low' would be a classic sign of self-healing failure, resulting in less controller supply current being available. This could possibly cause the processor to hang when the supply is loaded by certain combinations of relay activation.

Just a guess.


EDIT: Looking at the PCB image, there could well be an SMPS in there instead of a capacitive dropper. I'm on the wrong track!

[carlitos]

I found in some Samsung boards dry joints on coil terminals of relay, one did strange things.
Ressoldering all components related with motor could be first step.

[shakalnokturn]

karpouzi9:

They could make the electronics a lot simpler, a little cheaper and still be more reliable.
They have indeed done very well in making a hard to repair piece of junk that barely lasts past the warranty. In fact you almost get the feel they've been counting the run time before deciding on failure.
I could give it a quick peek with the thermal cam although I don't really believe it is hardware related just because it really feels like a software problem: No access to diagnostics mode, no fault detection for motor not spinning which seems like the basics as motor speed control has to be closed-loop anyway.
The only thing between the CPU and triac that hasn't been tested is a ULN2003, still that isn't an excuse for no fault detection.

Gyro:

At my first glance that was also my exact expectation and hope, unfortunately the 680nF capacitor isn't used as a capacitive dropper it's as shown for C01 on the previously attached schematic.
Yes there's a real SMPS based on an ICE2QR2280Z, in my various past repairs I've seen some of the ICE family IC's dropping out once warm, I wonder if it browned out in the past a few times causing write errors to the 24C04 on the way.

carlitos:

Thanks for the hint, the motor speed and direction relays did have weak solders on the coil terminals as well as a couple other components did on initial inspection, reflowing didn't change anything.

[fzabkar]

The EEPROM appears to consist of variable length data records beginning with 0x5A and terminating in a checksum byte. The last byte of each record is the sum of all the previous bytes in that record. All checksums are OK.

There are several gaps filled with 0x00 or 0xFF, but I can't verify whether those are just placeholders.

These are the records:

Code: [Select]

5A 30 35 44 45 35 45 53 4D 38 30 30 30 38 38 42 DC
5A 07 00 61
5A (5A 00 00 00 26 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01 81) 5C   <-- record within a record
5A 00 02 01 04 02 01 00 00 F0 30 03 00 00 87
5A 00 00 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 74 1D 02 00 00 EF
5A 04 02 60
5A 00 02 01 04 02 01 00 00 F0 30 03 00 00 87
5A 00 00 00 00 25 00 02 00 01 00 00 00 00 00 00 00 82
5A 20 74 43 31 0B 0D 00 00 E8 00 00 00 00 00 00 00 00 62
5A 20 20 31 43 0A 0D 00 00 E8 00 00 00 00 00 00 00 00 0D
5A 20 20 4F 43 03 0D 00 00 E8 00 00 00 00 00 00 00 00 24
5A 20 20 64 43 07 04 16 63 17 00 00 00 00 00 00 00 00 DC
5A 20 20 55 62 06 04 42 63 1E 00 00 00 00 0A 00 00 00 28
5A 20 62 43 32 14 04 8A 5D 1D 00 00 03 00 08 06 00 00 7E
5A 20 20 64 43 07 04 3F 61 15 00 00 00 00 00 00 00 00 01
5A 0A 13 77
5A 00 02 01 04 02 01 00 00 F0 30 03 00 00 87
5A 00 09 00 00 00 06 00 00 20 0C 00 00 00 95
5A 00 02 01 04 02 06 00 00 F0 3C 03 00 00 98
5A 00 02 01 04 02 01 00 00 F0 30 03 00 00 87
5A 00 02 01 04 02 01 00 00 F0 30 03 00 00 87

This is the complete dump:

Code: [Select]

Offset(h) 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F

00000000  5A 30 35 44 45 35 45 53 4D 38 30 30 30 38 38 42  Z05DE5ESM800088B
00000010  DC 5A 07 00 61 00 00 00 00 00 00 00 00 00 00 00
00000020  00 00 00 00 00 00 5A 5A 00 00 00 26 00 00 00 00
00000030  00 00 00 00 00 00 00 00 00 00 00 01 81 5C 5A 00
00000040  02 01 04 02 01 00 00 F0 30 03 00 00 87 5A 00 00
00000050  02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 74
00000060  1D 02 00 00 EF 5A 04 02 60 5A 00 02 01 04 02 01
00000070  00 00 F0 30 03 00 00 87 FF FF FF FF FF FF FF FF
00000080  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
00000090  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000A0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000B0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000C0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000D0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000E0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
000000F0  FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF FF
00000100  5A 00 00 00 00 25 00 02 00 01 00 00 00 00 00 00
00000110  00 82 5A 20 74 43 31 0B 0D 00 00 E8 00 00 00 00
00000120  00 00 00 00 62 5A 20 20 31 43 0A 0D 00 00 E8 00
00000130  00 00 00 00 00 00 00 0D 5A 20 20 4F 43 03 0D 00
00000140  00 E8 00 00 00 00 00 00 00 00 24 5A 20 20 64 43
00000150  07 04 16 63 17 00 00 00 00 00 00 00 00 DC 5A 20
00000160  20 55 62 06 04 42 63 1E 00 00 00 00 0A 00 00 00
00000170  28 5A 20 62 43 32 14 04 8A 5D 1D 00 00 03 00 08
00000180  06 00 00 7E 5A 20 20 64 43 07 04 3F 61 15 00 00
00000190  00 00 00 00 00 00 01 00 00 00 00 00 00 00 00 00
000001A0  00 00 00 00 00 5A 0A 13 77 00 00 00 5A 00 02 01
000001B0  04 02 01 00 00 F0 30 03 00 00 87 5A 00 09 00 00
000001C0  00 06 00 00 20 0C 00 00 00 95 5A 00 02 01 04 02
000001D0  06 00 00 F0 3C 03 00 00 98 5A 00 02 01 04 02 01
000001E0  00 00 F0 30 03 00 00 87 5A 00 02 01 04 02 01 00
000001F0  00 F0 30 03 00 00 87 FF FF FF FF FF FF FF FF FF

Edit:

The 3rd record is peculiar:

Code: [Select]

5A 5A 00 00 00 26 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01 81 5C

It appears to consist of a record within a record, and this sub-record also has a valid checksum.

Code: [Select]

5A 00 00 00 26 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01 81

[coppice]

If the consumers were paying a monthly fee to lease the washers, Samsung would have an incentive to make them last as long as possible. They'd be able to use higher quality manufacturing processes, they'd conformal coat everything, they'd maximize repairability.

Possibly, but I'll give you a counter example. In the 60s TVs were unreliable, so most people in the UK rented them. That way there was always someone to call when a TV failed, and there were incentives in the rental agreements to ensure quick repairs. Then electronics became pretty reliable, apart from connectors. UK manufacturers were so locked into the rental model they didn't re-engineer their designs to use larger boards with fewer dodgy interconnects. People started buying TVs when they were reliable, but bought Japanese models, and the UK TV industry descended from have a series of world firsts to oblivion. Both makers and customers can react to a lease or rental model in a number of ways, depending what they prioritise, the details of their incentives, and how adaptable they are.

[shakalnokturn]

fzabkar:

Thanks for the EEPROM analysis.
So you're saying it mostly looks sound and I'm on the wrong track suspecting data corruption?

[fzabkar]

fzabkar:

Thanks for the EEPROM analysis.
So you're saying it mostly looks sound and I'm on the wrong track suspecting data corruption?

All I can say is that all the records have valid checksums. There is empty space between offsets 0x15 - 0x25, so I don't know whether that should be of any concern. Basically, I can't see any obvious problem. To me, it looks like the MCU has gone stupid, so I would suspect its internal flash, possibly some strange firmware bug. But I'm glad I'm not the one working on this machine. :-?

[Uup]

fzabkar:

Thanks for the EEPROM analysis.
So you're saying it mostly looks sound and I'm on the wrong track suspecting data corruption?

I would erase the EEPROM and see if that corrects the issue. It could be that the data in the EEPROM has become corrupted or erroneous, which is causing the program to not behave as expected. I have seen this occur more times than I can count on other types of equipment. It is likely that the data stored in the EEPROM is not critical and that when the program detects that it is blank it will write default values for anything necessary.

You should do the weight calibration afterwards but entering the diagnostic mode should be possible after clearing the EEPROM.

Also, while you're erasing the EEPROM, I would check it for bad cells. Samsung probably doesn't use a circular buffer for the EEPROM and it may have worn out cells. I have seen this occur too many times in cheap commercial equipment, consumer equipment is probably no different.

[Bicurico]

Last month I retired my few years old Samsung EcoBubble.
I bought a new no brand washing machine, but kept the Samsung one in the garage.
Why? Because as everyone else says, mine isn't really broken. It just stops in the middle of a program. Pressing Play it will resume, on a bad day it will stop again a few minutes later. But guess what, on a good day, week or month, it will will work just ok.
I read that the culprit are bad solder points on the PCB. I wanted to just resolder them, but was confronted with a layer of some resin, which supposedly is applied to protect the PCB against water and humidity. I don't know how to remove it or what happens if you leave the PCB without this protection after repair.
Will be following this thread and I don't recommend to ever buy a Samsung watching machine. Ours already had two previous repairs. It was the most faulty one until now.
And the new machine cost about the same as a new PCB...

[PA0PBZ]

Half a year ago I fixed the Samsung of my son by removing all the potting and resoldering all the heavy components. It's a very common problem with these machines, the internet is full of it and yes, the problem is randomly stopping in the middle of a cycle. OP's pcb looks different though.

After the potting massacre:



Problems are easy to spot:



Machine has been running fine since.

[Bicurico]

How did you remove the resin?

That is my main obstacle to try a repair!

Please post how to get rid of it in a reasonable and safe way.

Did you then place a new protection on the PCB afterwards?

[PA0PBZ]

The first problem is to get it out of the plastic housing, the resin is very sticky! I made a cut in the plastic where the side meets the bottom so I could carefully push the pcb including the resin up and out of the plastic. Next you need to remove the resin by pulling and lifting it with some non-metal tool to avoid damaging any tracks, I think I used some of the plastic tools you get to open phones. Just start at a side and carefully start peeling and scratching until you have something like my first picture above... there's no shortcut I think so you have to invest like 15-20 minutes. After soldering all the suspicious joints (mostly on the heavy components, I guess due to the shaking) fill the hole you made in the plastic, put some self-adhesive feet under the pcb (corners and middle), put it back in the plastic and use some kit or whatever to close the gap between the pcb edge and the plastic housing, I used a glue pistol. I did not put anything on the bottom of the pcb apart from the feet and I never removed the coating on the top so I guess it should be fine. As I said, it's still running fine after 80-100 cycles. YMMV of course...

[Bicurico]

My PCB is not inside a plastic enclosure. It can be accessed directly.
It is the film that looks very hard to remove...

[Zero999]

Is it within  warranty? If so, dont' bother, just send it back. Even if it's past the warranty, consumer law in many countries mandates minimum warranties, or at least obliges the seller/manufacture to foot the bill for parts.

[Bicurico]

It is of course out of warranty. In the 5-7 years it lasted, it was already repaired twice. One time it was the rubber around the door, the other time I don't remember.

For some time I decided to buy the cheapest machines without any warranty extension (which can go to up to 5 years), as my experience tells me that no matter the brand, quality or price, househodl appliances will break around the 7th year.

This particular machine is a shame, because I am pretty sure that resoldering the contacts will repair it. The day the new machine arrived, my wife successfully washed with the Samsung one, which did not show any symptom. The days before, it was constantly stopping, which lead to the purchase of the new machine.

[shakalnokturn]

There are 2 ULN2003 on the PCB, I figured it was the last thing that I hadn't tested between triac and CPU so I swapped them just in case. No change.

I also checked how well the 24C04 was doing for write/read consistency on the universal programmer, looks good. Blanked it as suggested by Uup, true that this method has been successful in the past for me mainly on CRT TV's, as time goes by and complexity increases success rate is falling.

No joy, the display reacts very differently, which i guess is because the PCB is used on a whole range from 7 to 9kg machines with various types of display, key presses seem to have similar action as with original memory contents. After a few seconds everything goes into blinking so that must mean that CPU has detected wrong data.
The CPU did write some data to the blank EEPROM, I'm attaching it for fzabkar to peek at.

De-potting the PCB is not that hard, the risk is breaking an SMD or PCB by levering too hard in one place, once it starts giving way it's easy.

After doing quite a bit of searching on the EcoBubble machines I'm aware that there are solder problems but these seem to affect the IPM module on the "Digital Inverter" models more than mine.

I wouldn't care if it was for the couple of machines I've had to look at but they're failing by shit loads...

[fzabkar]

All the new records have valid checksums, and they are located in the same place as the originals. However, many of the original records are now empty (0xFF).

The first record appears to identify the machine in some way.

I suspect that the following area is assigned to 7 user defined wash cycles, in which case it would make sense for it now to be empty.

Code: [Select]

Offset(h) 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12

0000010A                          5A 20 74 43 31 0B 0D 00 00 E8 00
0000011D  00 00 00 00 00 00 00 62 5A 20 20 31 43 0A 0D 00 00 E8 00
00000130  00 00 00 00 00 00 00 0D 5A 20 20 4F 43 03 0D 00 00 E8 00
00000143  00 00 00 00 00 00 00 24 5A 20 20 64 43 07 04 16 63 17 00
00000156  00 00 00 00 00 00 00 DC 5A 20 20 55 62 06 04 42 63 1E 00
00000169  00 00 00 0A 00 00 00 28 5A 20 62 43 32 14 04 8A 5D 1D 00
0000017C  00 03 00 08 06 00 00 7E 5A 20 20 64 43 07 04 3F 61 15 00
0000018F  00 00 00 00 00 00 00 01

I notice that offsets 0x15 - 0x25 have not been rewritten, so I'm wondering what kind of record goes here, if any at all. It may be a crazy idea, but I'm wondering what would happen if we were to write a dummy record at this location in the original dump.

Code: [Select]

Offset(h) 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F

00000010                 5A 00 00 00 00 00 00 00 00 00 00
00000020  00 00 00 00 00 5A

[shakalnokturn]

fzabkar:

I think you're right about the machine identification, I also suspect that this will set the correct addressing of the display.

With the blank EEPROM I did a little fooling around trying to run different wash cycles and enter diagnostic mode.

Funnily before the front panel goes into blinking there is a different display for each cycle selected suggesting that at least the basic times are stored in CPU memory.

At first I thought that the records you noticed actually defined the sequence of default run times or timeout limits depending on the cycle.
I'd have to set the logic analyser to a working machine...

I've started looking for a cheap replacement PCB if I find one I'll read the EEPROM and compare, if not I'll give the dummy record a try just for the sake of it but I'm not going to be spending for ever on this, besides there is no trace of the firmware anywhere on internet.

[fzabkar]

78K0R/KE3 16-bit Single-Chip Microcontrollers User's Manual (μPD78F1146A, 256KB flash, 12KB RAM):

https://www.renesas.com/us/en/document/mah/78k0rke3-users-manual

[jchw4]

Most of the reports are of stopping in the middle of the cycle or motor not spinning at all.
The one that brought me here a WW80J5555DA won't spin the motor, it's a universal motor based model, the problem is not the motor.
Any washing cycle can run forever, the motor will not spin but no error code is triggered...

You have probably already checked that, but just to double check: a very common failure is a heater rusting through and leaking to the ground. Surprisingly this commonly leads to the water level sensor to report false values and then electronics does weird things.

Did you try runnig it with the heater disconnected?

[shakalnokturn]

No I haven't checked on that, the machine looks new but I should know better...
I'll check heater and NTC for earth leakage in the daylight, it's out in the garden  and it's night now.

[carlitos]

Check also for leakage to earth on all coil of motor, resistance more than 100M ohm.

[jfiresto]

I suspect the problem with Samsung is that they are a huge conglomerate and have folks with different levels of experience designing their machines. I got 20 years (3000~4000 operating hours) out of my last Samsung washing machine, but had to fix it twice: first to reenforce a too-flimsy door and then to replace a not-quite-durable-enough hose. After the motor electronics failed and would not stop spinning the drum, I replaced the machine with a Miele made in Gütersloh.

I could write something about a Miele if someone wants to start a thread asking: what's a reliable washing machine these days?

[shakalnokturn]

Just done some insulation resistance tests @1kV:

Input filter >2G
NTC >2G
Heater >2G
Motor frame to tacho > 2G
Motor frame to stator 1.5G
Motor frame to rotor 80M (Not fantastic but the motor frame isn't grounded anyway! Motor is attached to plastic, it can't even leak through the drive belt as the drum wheel is plastic... Cutting costs on ground wires? At least you avoid leaky motor faults...)

The other odd detail is the earth continuity is good up to the heater spade tab but between there and the fastening nut there's 13 . I'm not investigating on that unless I get the machine running.