iMac 2017 (18,3) fan on max loud, cpu on max

Georg3

New member
Hey y'all,

I work in the film industry, I have a good understanding of electronics and I design my own PCBs but I've never fixed an iMac logic board.
I am experimenting with battery powering my iMac.

I found that without the 7-pin flat data cable connecting the PSU and the logic board the iMac will work as expected but the CPU fan and the CPU will be pinned to max.
Then after some testing I figured out that pins 4 and 5 of this flat data cable are SNS_ACDC_N and SNS_ACDC_P, which if I'm not mistaken are two temperature sensors.

While measuring and testing things I have accidentally shorted these two pin (4 and 5), which resulted in the iMac shutting down and restarting and now even if this 7-pin data cable is connected the CPU fan and the CPU are pinned to max, like when it was missing those two temperature sensors.

So I suspect either one of those PSU temp sensors got damaged or something on the path to them.
I got my hands on an old iMac schematic which suggests these two temp sensors go to U5600, so maybe that could have gotten some damage?

Please help me troubleshoot and fix this problem.
PSU_BOARD.jpg
 

2informaticos

Administrator
Staff member
First of all, welcome to the forum!

Tell us the 820-xxxx code of your board.

iMac power supply should have a dedicated circuit which monitor the temperature and communicates with MLB through I2C bus; connected on the same bus as U5600 in your case.
Something got damaged in that bus; could be the chip from power supply, or other chip on the MLB.
 

Georg3

New member
Thank you for your prompt reply 2informaticos.

Please help me find the 820 code of my board.
Printed on the logic board I see two QR codes and some numbers below and another couple of numbers but unfortunately neither of those start with 820.
 

2informaticos

Administrator
Staff member
No schematic available.
820-00134 board is somehow similar.
Use it as guide.

This board has an dedicated ID for power supply.
Missing that connection can be the reason of your mod issue.
Now you have to solve the induced fault.
 

Georg3

New member
Thank you for the schematic 2informaticos.
When I was testing connecting those 7 pins one-by-one, I've found that only pins 4 and 5 (SNS_ACDC_N and SNS_ACDC_P) are needed for the machine to work perfectly, which makes me assume that one or both of these two might be missing now after I accidentally shorted these two to each other.

I have measured everything on the PSU side that is connected to these two pins and they seem to be OK.
I'll have to remove the CPU heatsink to get to measuring the components on the logic board side.

Do you have any suggestions about where and what I should check?
 

Georg3

New member
Exactly.
I haven't tried battery powering the machine yet, I always used the original power supply, this is what I've done:

- Because I knew when I will use big batteries I won't have the 7pin data cable my first test was to just unplug the 7-pin data cable and start the machine (with the original power supply), this resulted in: CPU fan is loud, cpu on 100% by kernel_task.
- To find out what it is missing I started connecting the individual pins of the 7-pin data cables one-by-one and I found that only pins 4 and 5 (SNS_ACDC_N and SNS_ACDC_P) are needed for the machine to work perfectly. Of course I've shut down the machine and unplugged power every time I connected a new pin.
- When I figured out that pins 4 and 5 are needed, I started the machine and I tried to measure what voltage is going through pins 4 and 5 so I can replicate it but while I was doing that I accidentally shorted these two pins and the machine resterated.
- Now even when the entire 7-pin data cable is connected (still using the original power supply) the CPU fan is loud and the CPU is on 100% by kernel_task, exactly like when pins 4 and 5 where missing.
 

2informaticos

Administrator
Staff member
After accidentally short, one of L5610/11 could be burnt; if you are lucky.
If not, change U5600.
Once you get it working correctly with all the normal connections, then disconnect J6901 and solder a diode, or NPN transistor between pins 4 & 5.
Check Q5605 in schematic, to make an idea.
 

Georg3

New member
Thank you so much 2informaticos, that is brilliant!!
I'll measure those as soon as I get home from work today.
If one of those have burnt, where can I buy a L5610/11 or U5600?
 

Georg3

New member
I've measured L5610/11 they are both 0.9 ohm. That means they are OK?

The actual logic board is a bit different than the boardview. The main difference is the IC that is where U5600 should be only has 8 pins instead of 10 as on the boardview.
I have attached a photo of it.
Am I looking at the wrong thing?
IMAG1724 copy.jpg
 

2informaticos

Administrator
Staff member
May be something else, not thermal sensor.
All Apple schematics I looked have only EMC1414/28 chips; none of them MSOP-8.
 

Georg3

New member
So it is somewhere else then, right?
I will have to probe through and find where the J6901pin 4 & 5 lead me... hopefully I'll find it quickly.
Do you have any advice how to trace them quicker?
 
Top