–WORK IN PROGRESS —
My late 2009 i5 iMac had a fault on the motherboard which shorted the PP12v_G3H to ground, not even the first of the diagnostics LED lighted up. No diagnostics lights normally indicates a dead power supply, so I replaced the power supply with a used one from Ebay, which didn’t change anything, well it actually did, as a matter of fact I now had 2 broken power supplies ;-(.
It turned out the short on the mainboard (820-2733), was due to a dodgy LCD connector and a blown MosFET Q8040. Now that this has been repaired, it is time to focus on the power supplies (while waiting for yet another power supply from ebay).
Unfortunately I have not been able to find a schematic for the power supply, but there is a good thread discussing this unit on MacRumors Forum Looking for schematic diagram of 27″ 2009-2011 power supply, PA-2311-02A (liteOn)
There are 4 versions of the a1312 27″ iMac Main Logic Boards
- Late 2009, LGA775, Core 2 Duo, 820-2507
- Late 2009, LGA1156 1. gen Core i, 820-2733
- Mid 2010, LGA1156, 1. gen Core i, 920-2901
- Mid 2011, LGA1155, 2. gen Core i, 820-2828
The schematic and .brd file for all the above main logic boards is available on the internet – Google will gladly help you find them.
If you have a LGA775 based a1312, it is worth considering upgrading the Main Logic Board from a 820-2507 to 820-2733, There is a major performance difference here, since Intel integrated what was the Northbridge in the Core2 design into the CPU-chip itself so the RAM is now controlled directly by the CPU, much like the current (year 2022) design, and the performace of the 1 st and 2nd generation Core-i processors is not that far off the current Intel standard. You also get Quad Processor with Core i5/i7 (8 threads with i7). I do not own a Core2 iMac so the notes here only apply to Core-i based boards.
Unfortunately you cannot upgrade to the 2nd generation Main Logic Boards, the internal LCD connector is different and the connections on the back are different too. But the Power Supply and Inverterboard are compatible over the 4 iMac iterations (except LCD-sync and other cables)
All of the a1312 iMac share the same power-supply but as is obvious from the above 3 excempts from their schematic, the cable from the PSU to the MLB is not the same.
- Late 2009 and Mid 2010 uses the 593-1034 cable, with power to SATA hard drive
- Mid 2011 uses 593-1383 cable (SATA power is handled by 593-1317 connected to the MLB)
I didn’t realize this differences before, so I might have killed the PSU by connecting PSU to the wrong MLB with the wrong cable. It’s not LEGO, just because it fits dowsn’t mean it is OK.
+-------+ GND | 16 1 | GND backlight 12v | 15 2 | GND backlight 12v | 14 3 | PS_ON PP12V_S0_PS | 13 4 | GND PP12V_S0_PS | 12 5 | PP12V_G3H_ACDC PP12V_S0_PS | 11 6 | GND SMB_ACDC_SDA_RC | 10 7 | SMB_ACDC_SCL_RC GND | 9 8 | GND +-------+
ON Semiconductor has a well documented Evaluationboard for the NCP1605/1396 which is what is used in iMac PSU. the powersupply in the iMac gets its bias power from the main transformer as illustrated on the diagram below, this is different in the evaluation board.
The way I think of the PSU, it consist of 4 parts
- PFC controller, is just an advanged Boost Power supply resulting in a 380 Vdc
- NCP1396 drives the Transformer
- The bias circuitry powers the primary side chips.
- The secondary side rectifies and smothes the 12Vdc and sends feedback to the NCP1396 via optocoupler
On the iMac powersupply there is also some MosFETs to control 12v to MLB and the Backlight pcb, plus some i2c monitoring chip.
This is already big progress, before we had 1 problem – a dead PSU. Now we have 4 potential problems, and most often only one of them is real a problem to be solved.
So where is the problem
- I see a 380v after the PFC controller, hence the PFC controller seems to work (400v to 0v on annotated PCB)
- I see a fluctuating bias voltage, indicating that PSU tries to power on, but then powers down again, the bias voltage must come from the transformer, so an inital guess is that the primary side works. (+17v to 0v on annotated PCB)
- I see no 12v on the capacitors on the secondary side, so it must the rectifiers that is dead. (12v to GND on annotated PCB)
- I tried to measure the diodes on the secondary side, but I could not determine if they were bad, so I desoldred them and tested them with my multimeter – they seems fine.
- I tried to measure the push-pull MosFETs, but I could not determine if they were bad, so I desoldered them and tested them in my component tester – they are fine.
- My conclusion at this point was that something is wrong between the rectifier and the feedback circuit on the secondary side.
You may wonder why I didn’t test the diode in the component tester – well now I do wonder too, a few days later, before I was putting the thing back together again, I did test them, and guess what the component tester think it is two resistors.
The secondary side
The main components on the secondary side are:
- 4 dual diodes SBR20A40
- 2 MosFETs P0603BDG
- quad op-amp and voltage reference TSM104W
- quad op-amp AS324A
- I2C Current/Power Monitor ina219
- I2C Temperature sensor emc1404
My next move is to connect a 12v Lab power supply to where the diodes were removed, and see if I can figure out how the secondary circuit works, this is also much safer than working on a mains connected power supply.
Supplying 12v from a Lab PSU to where the rectifier diodes were, I see this power drawn:
- 90 mA @ 11.8v PSU not connected to main logic board
- 120mA @12.1v PSU not connected to main logic board
- 130mA @ 12v when connected to a 2011 logic board, diagnostics LED1 is on
- 600mA @ 12v when connected to a 2011 logic board, diagnostics LED1 is on
Power button pressed gives no LED2, So the iMac decides something is wrong, and does not go to LED2 phase, There was a 2n3904 connected to one of the diode heatsinks, but reinstalling this into the PCB did not change anything. Next step would be to check if the two MosFET that turn on PP12V_S0_PS is ok.
- The 2009 logic board draws a lot of power so there are still unfixed faults on this logic board, maybe I should just let it be, and see if I can install the 2011 logic board instead, this is not just plugg and play.
At this point I decided to reinstall the 2 MosFETs and the 4 dual-diodes, so I double checked the component with my component tester and found the diodes to be faulty, I swear the multimeter showed there was no connection in one of the directions, maybe they just developed a low break down voltage like a zenerdiode, and the multimeter didn’t reach that voltage – another lesson learned.
So I desoldered the 4 dual diodes on PSU no 2 and, they were dead too, in hindsight I should probably have come to this conclusion from the very beginning since both PSU had been the subject to a dead-short between the 12v output and ground
So now I am waiting for components, and wondering what the fault on 2009 mainboard might be?