Understanding communication circuitry of split system A/C

[Nyaarla]

I am currently attempting to diagnose an issue with an LG Split System air conditioner (Model LSK244V-3) from circa 2003.

Background

Over the past few months, the air conditioner has been randomly turning off a few minutes after startup. If it did not, it would run for as long as requested. This became more and more common until it never starts successfully. The indoor unit will run fine fan-only mode (I would assume because it does not engage the outdoor unit).

The indoor unit shuts down with an error code of 5 flashing lights (according to the manual, this means communications error), and the outdoor unit also begins flashing 5 times after the indoor unit shuts off (though sometimes it never even does that). The outdoor unit never starts up (compressor or fan) before the error code appears.

Investigation
I took a look at the indoor and outdoor circuit boards to see if there was any visible damage. There was nothing obvious on the indoor board (although there were a number of bugs and gunk around the connecting wiring. This has since been cleared, but it did not resolve the issue). I also checked the communications wire for continuity.

Looking at the outdoor unit circuit board (see below), there is visible damage to the trace above IC02K, and to the resistor near D02K. There also seems to be some discoloration on one of the optocoupler pins on IC01K.




The red wire is the communications wire, and the black wire goes direct to earth.

After testing with a multimeter with the circuit switched off (at the breaker):

• The two parallel resistors near R05K measure 15.2Kohms across them, but based on their markings they are each 40Kohms, so I should be measuring 20Kohms?
• The resistor at R05K measures 1.5kohms, which matches its markings
• The two series resistors at R08K and R09K measure 9.2kohms and 9.3kohms respectively, which is in line with their markings
• The burned trace still has continuity, it looks like only the surface covering came off.
• The burned resistor now measures at 52kohms, but I have gotten different measurements previously. The burned resistor has colors [Brown, Red, Green, Black, Green] (the picture shows the black as blue). But this can be interpreted as either 125 ohms /.5%, or 50.5Kohms 1%. The latter looks correct based on the measurement, but it's ambiguous.

I have attempted to find a circuit diagram for this board to confirm the components and values, but I have not been able to find a service manual for this model.

It is possible that the issue actually lies elsewhere, but the apparent damage indicates that I should be looking here.

The most interesting finding
If a place my multimeter probes on either side of Z003K (any polarity) and set it to either diode test or resistance measurement mode, I can sometimes get the outdoor unit to power up for a short time. However it shuts down again once the compressor is up to speed, at which point the indoor unit immediately turns off with the same 5-flash error code. My theory is that the indoor unit has not heard from the outdoor unit for a period of time, and sends a shutdown signal which is received correctly.

It is my understanding that both of these modes apply a small voltage across the probes, which could potentially affect either the zener or optocoupler and nudge them into working for a short time. Of note is that the LED on the circuit board flashes twice every ~0.5 seconds when the multimeter is connected like this. Otherwise the LED would either not be lit, or only show solid red.

This behavior does not occur when the multimeter is turned off or in voltage measurement mode.

Questions

The main reason for putting this here is to try and get some pointers in the right direction to understand how this circuit works, what might have caused the issue, and what I should look to do next.

My current understanding is that there is one optocoupler for TX (IC01K) and RX (IC02K), with matching ones in the internal unit. The diodes appear to form some sort of DC rectifier so that current only flows towards the communications wire, but I'm not sure what the purpose of the Zener diodes are here.

Planned next steps

• Replace the parallel resistors with two 2W 10kohm resistors in series to get 20kohms (since I don't have two 40kohm resistors, only 47kohm)
• Replace the burned resistor with a new 2W 47kohm resistor (close to 50.5kohm)
• Replace the electrolytic capacitor with a new 2.2uf/400V capacitor (current rating is 2.2uf/100V). Voltage across it measures about ~80V DC when live.
• Replace the optocouplers (Markings show "3G P421 4BL B13") with a similar part off aliexpress (Though I ordered the GB version, which is a slightly different spec according to the datasheet: 100% minimum current transfer ratio vs original 200%)

Do these sound reasonable?

[fzabkar]

Service manuals used to be available here:

www.jordansmanuals.com/default.aspx?Brand=LG&Product=Aircon

However, several categories of product manuals have been removed from the site. :-(((

[james_s]

At a glance it looks like you have corrosion below that electrolytic capacitor. Th first place I always look with faults like this is any electrolytic capacitors. They commonly fail and if you've got sagging voltages or other issues from that you can get all sorts of weird symptoms.

[Nyaarla]

@fzabkar I even look a look on archive.org, but no luck

@james_s Thanks for that insight. I ended up replacing a number of capacitors on the outside circuit board which seems to have helped a little bit.

---------------

So a bit of a late update on this:

I ended up replacing a number of capacitors on the outdoor circuit board that seemed related to the communications circuitry and it now (usually) starts up without having to probe it. Unfortunately, it still shuts off after 2 minutes with a communication error and no clear indication as to why.

After looking at the resistors a bit closer, it turns out that I was just reading them wrong. They're all within expected tolerance. Even the visibly damaged one.

-----------------

I also came across this document:
https://www.orionair.co.uk/PDF/lg/LG_Split_Systems_Fault_Codes.pdf and performed the recommended checks across the communications line and neutral with the following results:

2-3, outdoor terminals, connected: fluctuates between 30-40vdc
2-3, outdoor terminals, disconnected: fluctuates a few volts around 68vdc
2-3, indoor terminals, disconnected, steady around -3vdc

The document states that it should be fluctuating between 0-70vdc when connected, but it only seems to be operating in a very narrow range.

-------------------

My current plan is to also replace some of the capacitors on the indoor circuit board, but if that doesn't work I don't have too many ideas left. Maybe it's about time to look for an oscilloscope...

[HackedFridgeMagnet]

Maybe it's about time to look for an oscilloscope...

That's what I would do.

What's the comms protocol? I guess it that awful power line communications. Be careful of the mains.  Also check the DC rail voltages are something sensible.

 Probe the comms link at both ends but only on the DC side.
If your anywhere near Hobart I might be able to help with the scope.

[pepona]

Hi, if you want I can upload the modem circuit of another brand that may serve as a reference.
I drew the circuit in a little hurry, maybe it contains errors but it works.

Sorry for my bad English

Regards
Marcos.

[Nyaarla]

@HackedFridgeMagnet I haven't been able to find too much information on the communications protocol in use. It seems to be based on a TX and RX optocoupler at each end to manipulate the DC voltage of the line. The voltage is rectified from AC via a simple rectifier circuit on the outside board, with multiple diodes ensuring that the current flows towards towards the indoor unit.

Speaking of mains, the entire low voltage portion of the circuit (5v, 15v, gnd) is 115vac to earth. not sure if that's a design decision or caused by something going wrong.

Thanks for the hint about checking the rail voltages. I ended up doing some further investigation and found that while they usually seem stable enough when the system is operating, it there are some circumstances where 5v and 15v sit at 0.7v.

Thanks for the offer with the scope, but, alas, Sydney is a little too far away 

----------------
@pepona A schematic for a similar circuit could be very helpful in determining how this one is supposed to operate. I'm also planning to look for some technical manuals for similar models of LG air conditioners - hopefully they decided to reuse the same design.

----------------

After replacing a few capacitors on the indoor unit circuit board that seemed to be related to the communications circuitry, there still doesn't seem to be any improvement.

That being said, I think I'm getting a bit of a better understanding of how how the different parts of the circuit work and what issues it's experience (or at least the symptoms).

I ended up getting pictures of both sides of the circuit board and tracing out the connections between various parts of the circuit. Especially the power rails, as I'm trying to work out what generates and filters the power supply.

Screenshots:
https://www.dropbox.com/s/lbz1yn2gxbilhnb/board_bottom.png?dl=0
https://www.dropbox.com/s/5f8m5o9906fvc9o/board_top.png?dl=0

Red is 5v, Orange is 5v rail on the other side of a resistor (incomplete). dark blue is 15v. Grey is gnd (incomplete). Magenta is AC live. Light blue is AC neutral.

I also have a Google photos album of all the pictures I have so far showing more details: https://photos.app.goo.gl/LUf87jHKtXdvBbU47

-----------

One thing I did find was that the TX trace (visible under the led) had been corroded and didn't have any continuity - that definitely would not have helped here  . After scraping away the corrosion, adding some solder, and verifying continuity, it didn't seem to help at all . Still a communication error.

 I noticed this when checking the low voltage side of the TX optocoupler and found that one leg (anode) was connected to +5vdc via a 600ohm resistor, the other was supposed to be directly connected to the microcontroller (and connected to ground via a capacitor).

The microcontroller side (cathode) was measured at 6.3-7vdc to gnd, whereas the anode was measured at 5v. A reverse voltage across the optocoupler diode didn't seem right. My guess is that this voltage was measured due to the other end being disconnected, but I'm not sure if it could be some voltage from the transistor side of the optocoupler leaking through.

-----------

There seem to be 3 different symptoms that may or may not be related.

1. Once the indoor unit powers on and provides power to the outdoor unit (verifed by checking voltage at the terminals), there is no sign of life or activity from the outdoor unit. There no led, fan or compressor, and both the 5v and 15v rails measure 0.7v at the test points. This seems to occur less often when I've been testing in the middle of the day so I'm thinking maybe it is related to temperature of the board or a component? . It may also randomly go back to this state for anywhere between a few seconds, to a few minutes, to giving up for the day.

2. (1) resolves itself after some period of time. There is then correct and stable voltage (relying on my multimeter here) on both the 5v and 15v power rails. The red led flickers rapidly, which I understand to mean that it is recieving communication from the indoor unit and is expected behaviour. However it will just stay like this for an arbitrary period of time. Sometimes only for only a few seconds before it starts up, but other times it could be tens of seconds, a minute, or never.

3. (1) and (2) both resolve themselves after some period of time. The fan starts up and then the compressor begins to come up to speed a few seconds later. Either the system will go back to state 1 after a few seconds (rarely), or the indoor unit will trigger a communications error (after exactly 2 minutes of turning on the indoor unit, regardless of when the outdoor unit starts) and the compressor will immediately shut off with the fan running for another ~30 seconds or so.

------------

My current next steps will be to replace some more capacitors on the outdoor unit which I identified as being part of the power circuitry (that circuit trace was very helpful).

One thing that I'm having trouble with though is working out how the microcontroller and low-voltage circuitry actually gets it's power. I see two possible candidates: the component marked POWER MODULE labelled "μPM1515-W" (have not been able to find a datasheet) below the 4 large caps, and the bridge rectifier and associated circuitry to the right of the 4 large caps.

[pepona]

Hi, no exact the same but it can serve you.


I think it´s very difficult to find the original schematic.

Regards
Marcos.

[HackedFridgeMagnet]

1. Once the indoor unit powers on and provides power to the outdoor unit (verifed by checking voltage at the terminals), there is no sign of life or activity from the outdoor unit. There no led, fan or compressor, and both the 5v and 15v rails measure 0.7v at the test points.

this must be the problem.
It wont receive comms unless these rails are up.

another thing that might be useful is a thermal camera. might help find if something is shorted.
also when power is off you can check the semi conductors to see if they are shorted. The fact that it is intermittent kind of says that they aren't shorted though.
also while live you can push components around a bit with and insulated tool(back of a screwdriver). See if you can somehow get the DC power to come on. Or push the wires around too. Do it safely of course.