Mac Air 2012 - A1465 820 3208A

[jpadie]

Hi

I've got another mac air in at the moment Coffee damage. Sugared too.

the whole backlight area is nasty. Fuse blown. I've replaced that and the green light goes nicely to orange. But fan stays at high speed throughout.

No power in the backlight circuit. Looks like there's no signal on LCD_BCKLT_EN but there is 8v on LCD_BCKLT_EN_L on Q9707. Is that normal?

some of the coils appear shorted to ground - L7100, L7720 and L7630 and every other coil towards the fan; and U1800 is getting hot to touch. all are at about 45C.

any thoughts on chasing down the short? alcohol test does not show anything particularly hotter than other things. likewise an infrared heat scan (with a gun).

thanks
Justin

 

[aprendiz]

You have a display connected??
What is the resistance to ground on L7720?
Fan at high speed is a sensor failure, also if trackpad is not connected....

 

[jpadie]

thanks aprendiz

resistance to ground over L7720 is approx. 13MOhm.

no display connected. Board is bare, so no keyboard, trackpad etc.

 

[jpadie]

Plugged a screen in .

The laptop boots to bios but u1800 gets very hot.

L7720 seem to vary in resistance to ground . Now around 13k.

L7630 is still close to zero ohms to ground . L7330 is a little better at about 1k .

I've prodded at most things that seem like a culprit and I'm not finding any bad caps or similar . None of the passives are getting super hot. And I get the boot chime (no SSD so no OS . I will create a usb boot drive tomorrow).

Is this likely to be a bad u1800? If so, I guess the board is a write off? Or is it worth trying some heat on it?

Separately the screen connector started smoking around the backlight pins. So I guess that wasn't as salvageable as i had originally hoped. It cleaned up nicely but probably was shorted inside. I've removed it now . The backlight pins of the cable have lost their copper but the underlying metal is still good. The resistance to ground didn't change after removing the connector.

The fuse went. So that's one positive!

 

[jpadie]

just following up on this. the laptop continues to power up but a couple of power rails seem odd. notably

PP18v5 is at 14.8v rather than the stated 18.5v (this feels like an error in the schematic...) ; and
PPVRTC_G3H is at 3.3v rather than 3v.
PPVCore_S0_CPU is at 0.8v rather than 1.25 and
PPVCORE_S0_AXG is also at 0.8v rather than 1.05.

I poured some mild heat on to U1800 and the coil resistances are now:

L7630 16Ohm

L7330 820 Ohm

L7110 20 Ohm

L7510 5Ohm
L7550 4Ohm

Any hints on where to search next?

many thanks
Justin

 

[dukefawks]

U1800 normally gets quite warm, but you should be able to keep your finger on it.

LCD connector needs to be replaced now and also the LCD cable. This is the reason why you use a cracked trash LCD for testing.

 

[jpadie]

LCD connector is removed. I'm waiting for new connectors so can't test with a different screen yet.

But the voltages on ppvcore still seem wrong . I've removed some of the coils and injected voltage into that rail directly but nothing gets hot.

​​​​​​

 

[dukefawks]

This thing was turning on with an image and you are injecting voltage on CPU Vcore?

Does it even still chime now?

 

[jpadie]

No idea about the image . The display connector was fubar.

Chiming normally. But low voltages on the 1.25 and 1.05 rails. High voltage on the 3v rail and top speed fan.

I've put it to the side until the connectors arrive . Loads of public holidays in France at the moment so deliveries are few are sparse .

 

[dukefawks]

It chimes so board is fine. Replace LCD connector and LCD cable. If you insert the fried LCD cable it will burn your connector again.
This is why you need a cracked LCD for testing this stuff.

 

[jpadie]

Thanks

I've got a few screens now .

​​​​ fingers crossed that connectors will arrive early in the week .

 

[jpadie]

replaced the lcd connector and plugged the fan etc back in. Spun up the first time then stopped and now will no longer spin.

U1900 still getting very hot to touch. not something I'd like to keep my finger on (and hotter than comparable machines).

Gut feel is that there is still a short somewhere in this board but nothing is getting hot other than U1900.

Any ideas?

thanks
Justin

 

[dukefawks]

You mean U1800? Is that getting hot without the fan spinning up?

 

[jpadie]

yes U1800 gets hot without fan spin.

the fan 'jolts'. rocks on its spindle.

there is voltage on the fan pins though. is this a case of a broken fan ?!!

 

[jpadie]

nope. nothing on PPVCORE.

14.4v on PPDCIN_G3H.

the other rails are mostly there but:

I'm not getting 1.8v at all the points that it should be present. for example - there is nothing at R2719 but u2700 is fully provided.

PP1V5_S3: 1.35v
PP1V5_S3RS0; 1.35v (but PP1V5_S0: 1.5v)
PPVTTDDR_S3 : 0.675v
PP0V75_S0_DDRVTT: 0v
PPVCCSA_SO_CPU: 39mV
PP1v05_S0: 76mV
PP1V05_S0_LDO: 1.05v
PPVCORE_S0_CPU and AXG: 0v
PP1V5_S3_CPU_VCCDQ: 1.35v
PP1V05_S0_CPU_VCCPQE: 0v
PP1V8_S0_CPU_VCCPLL_R: 1.8v

If I've not included a rail then it is reading OK. I have not checked the TBT rails.

 

[jpadie]

the 1v8 issue was measurement error. It is present (1.78 - 1.80) in all the right places. ​

I had some spare time so probed some of the main caps around the coils. C1768, c1680, c1683 and c1682 (PPVCore caps) all have very low resistance (4-15Ohms). I'll have to chase this through.​

 

[jpadie]

Is there an easy way to isolate the core so that I can safely inject current into those rails?

 

[Inwerp]

the reason you're reading low resistance on CPU core voltage is that it has CPU installed. CPU is basically your short unless its dead.
CPU/GPU has very low resistance between power rail and ground, this is absolutely normal. just play a little bit with the numbers: CPU eats, say, 20W. now divide it by CPU voltage, this would be your current. now divide voltage by current and this is your resistance, 10-20 ohm, yup, multimeter CT will beep, "ahaa! we have a short here!"... nope.
Also, in "shorted circuit" conditions you can't really test capacitors without desoldering them.
wrong voltage does not necessarily means short circuit. maybe one phase does not work, maybe one of the CPU Power MosFETs failed, maybe one of the mosfet drivers failed, maybe CPU asks for the wrong voltage for some reason.
but it's all does not matter since your board chimes. brain-dead boards do not chime. board needs CPU to run the bios as well is to decode and play the audio file.

 

[jpadie]

thanks. I think that a resistance across a cap in that area that low seems wrong though. Are you saying that you get similar readings on a known good board?

this one has stopped chiming. corrosion in the intervening period perhaps. Or a shorted cap (which is my bet).

 

[Inwerp]

this is not "across cap" resistance, this is net resistance of all components between your probes. the only right way to test capacitor is to desolder it.
the only stuff besides voltages you can really test with MM in circuit are low resistors or fuses. caps could be tested with an oscilloscope on working board but you need this only for electrolitic ones since they are known to fade.
i am working primarly on PC Notebook boards, so sometimes i have an priveledge of removable CPU
so yup, 5-10 ohms on GPU or CPU power line is absolutely normal for working board, no mater which board it is. it is also normal to have "low resistance" on RAM. if something takes 1 Amp current, sucking 1.3 volts, it will and it should "beep" on continiuity test.

once again: if you had chime, CPU Voltage rail was almost surely ok. if it's dead now after playig with it, i guess you killed something like voltage sence or power driver on this line.
if you have a water damage with an ongoing corrosion, the first step was to stop it, then troubleshoot its consequences. i hope you did it, so it should be out of the table.
voltage injection is not a magic trick for everything, it should be applied only when the short circuit is obvious but hard to find.

bierf example:
you have a short on high side mosfet of the CPU Circuit, you inject the voltage right to the V+ rail somewhere. hello, your entire CPU power system is probably dead now since you gave higher voltage to the low voltage line. capacitors will be killed one by one: you inject voltage, you see some heat on one, you remove it, aha, now the next one is dead, then you end up with the dead voltage sence, dead feedback, dead driver (since mosfets usually short between all terminals, not just drain and source), dead caps, and dead CPU if your LPS is strong enough.