820-01700-05 5V 0.00-0.01A on all USB-C Ports

It's not too often I see a device where all of the USB-C ports pull basically zero current (Starts at 0.01A and drops to 0.0A) with no looping or other activity, but that's the case on this 820-01700-05. No liquid damage or issues spotted with a visual inspection, pretty through check for shorts didn't turn up anything, customer reports sudden death. Anything in particular that's common on this model that I should check for before spending an hour going through every power rail in the CD3217 startup sequence? The best reference I've found so far is https://logi.wiki/index.php/CD3217_and_T2_Power_on_Sequenc, but that's a heck of a lot of rails to test and not specific to this board.
 
Actually PPBUS_G3H is 0.099V diode drop to ground, which is surprising given it didn't beep during an initial quick check. Every CPU related coil measures approx 0.1V drop to PPBUS_G3H, what all do I need to check before safely injecting power?
 
Okay, did a lot of measurements and determined the CPU fets should be good, so I injected 13.1V into PPBUS_G3H, consumed about 250mA and U9580 got warm, found PP2V5_NAND_SSD0 to have low resistance (About 6 Ohms, but randomly drops to nearly 0 on occasion), so I injected 2.5V into that rail which consumed 250mA, but on occasion randomly jumps to 4Amps for a few seconds (I've observed this repeatedly, I'm not accidently shorting anything). U9400 gets warm with 3 distinct hot spots that melt first when using freeze spray. Is this adequate to conclude that this NAND chip has died? What do I need to do to replace this NAND chip? Can I just swap this one and dfu restore afterwards or is it a more complex process?
 

2informaticos

Administrator
Staff member
Pufff, injecting 13V directly; too risky, don't do this again.

Did you check diode mode to ground at L9580 first?
What if there was 0.099 (meaning L9580 shorted to PPBUS_G3H)?

We have few threads about changing NAND chips, but nobody left a feedback about did it and succeed.
Of course, DFU is needed in such case.
Some people claim on Internet they got it swapping with NAND from the same position on the donor board.

BTW, do not write the end "V" after diode mode reading; avoid voltage confusion.
 
In the future, how would you recommend finding a partial secondary short with low power consumption if not by injecting on the main rail? I started at 1V and worked up very gradually while monitoring the outputs of every independent fet I could find to determine that there wasn't a significant risk before cranking up the power.

I didn't check at L9580 first, and certainly see your point, but also from experience I believe it's a lot less common for an integrated buck IC like this to short the input rail to the output than with independent fets. What do you suggest instead when most everything on the board shows a low diode drop?

I'm willing to go through with changing all the NAND chips + controller + rom if need be as this is a pretty expensive device, but I'm struggling to come up with any definitive answers on this. I suppose I'll just have to get a donor and start with one, and swap the rest if it doesn't work. I don't suppose it would be possible to use an iPhone NAND programmer to clone the low level data to another NAND chip I already have laying around, would it?

Will do, thanks for your help!
 

2informaticos

Administrator
Staff member
If a NAND chips is bad and need to be replaced, I bet you cannot even read it, no?

We talked a lot about how and NOT to inject voltage on PPBUS_G3H power rail on the forum.
ALWAYS check first for shorted high-side MOSFETs; including the ones integrated on the power chips.
If you search the forum, will find lot of threads with shorted U9080/580!

BTW, you've didn't mention to started voltage injection with 1V; just said injected 13V...
 
There's been some recent developments with reading sysconfig data from dead iPhone NANDs, generally it's more focused on when the damage prevents access to the main partitions but still allows the sysconfig regions to be read with some software tricks, but I've also had success with using this on partially shorted NANDs while injecting external power. Maybe I'll give it a go, probably can't hurt.

Interesting, I'm surprised the integrated fets short that often, I will check that in the future! Fair point, I was trying to summarize a lot of testing that I did.

Assuming the customer approves the quote, I'll let you know what I find wrt NAND swapability.
 
Finding donor NAND's for this board has been quite the saga, but now I'm afraid it might have been in vain.

I removed U9400, the NAND that was getting warm, after which PP2V5_NAND_SSD0 still has a diode reading of 0.018, which seems low to me. This is the exact same value as it had before, expect previously when I injected 1V into PP2V5_NAND_SSD0, the board consumed 5A, now it only consumes 80mA.

Upon powering the board, PP2V5_NAND_SSD0 alternates between 1.5V and 3V in 3-5 second intervals. Previously it was partially shorted and nearly 0V. This is not what I expected to find, and I'm pretty confused as to how I'm getting 3V on a 2.5V line that's switched off of a subsidiary of PPBUS. Does this point to swapping 6x NANDs being a waste of time? Or what could possibly be going on here?
 
Additional note: Before removing U9400, the board consumed 0.01A at 5V when powered, now it consumes 0.110A at 20V, so I feel confident the NAND was the issue, but I don't understand the present situation.
 

Maltinho

New member
Finding donor NAND's for this board has been quite the saga, but now I'm afraid it might have been in vain.

I removed U9400, the NAND that was getting warm, after which PP2V5_NAND_SSD0 still has a diode reading of 0.018, which seems low to me. This is the exact same value as it had before, expect previously when I injected 1V into PP2V5_NAND_SSD0, the board consumed 5A, now it only consumes 80mA.

Upon powering the board, PP2V5_NAND_SSD0 alternates between 1.5V and 3V in 3-5 second intervals. Previously it was partially shorted and nearly 0V. This is not what I expected to find, and I'm pretty confused as to how I'm getting 3V on a 2.5V line that's switched off of a subsidiary of PPBUS. Does this point to swapping 6x NANDs being a waste of time? Or what could possibly be going on here?

Where did you take the donors from? Exact same board or maybe iphone?
 
Where did you take the donors from? Exact same board or maybe iphone?
I have not installed any donor NANDs yet. Just removed the shorted NAND and observed that somehow I'm getting 1.5 to 3v on a 2.5v line. I'm familiar with the need to transfer a complete set of NANDs from a select list of compatible boards and match up the clock request line IDs and not doing anything silly. I have an exact match donor board in this case.

If you're asking for yourself, unless you've developed a new programmer tool, you can't use an iPhone NAND, there's system configuration data on each NAND that's used to configure the raid array under the hood. There's not a programmer available for transferring or generating this, so you must swap a complete set of NANDs from another board.
 

2informaticos

Administrator
Staff member
"I'm pretty confused as to how I'm getting 3V on a 2.5V line"
Did you check the voltage at correct output pin of the coils?
 
Yes, both sides of L9581, and every other spot I've probed PP2V5_NAND_SSD0 show the exact same behavior. To get more specific on the timing, PP2V5_NAND_SSD0 is 3.07V for about 15 seconds, then drops to 1.51V for about 2 seconds, and then repeats in a loop. This isn't the output of the buck collapsing and residual voltage on the line leaking slowly. The transitions between 3V and 1.5V are sharp, and those levels are stable for their respective periods.
 
C9582 and C9585 are not shorted, capacity test in circuit didn't work but I can pull either if that's helpful.
R9583 is 10 ohms as expected.
R9580 and R9581 both measure 150K but show as 475K and 221K on my 820-01700-A schematic.

Could this be due to differences between the 820-01700-A and 820-01700-05 board designs? Looking at the circuit topography makes me feel like this isn't the result of them just being in circuit, as they have almost the exact same value. I can pull them and measure if pertinent.

Should I take voltage measurements throughout the buck circuit?

Just to be extra clear, I have a NAND removed, this isn't the cause of the elevated voltage, right?
 

2informaticos

Administrator
Staff member
The removed NAND cannot change the voltage.

Change C9582, just in case.
Is not about short there; its capacity sets the soft start time.

R9580/81 value must be checked out of board.
You can play with the mentioned resistors to adjust output voltage.
Increasing R9581 will decrease the output voltage.
Get the same result decreasing R9580.
 
No change with C9582, after removing the original I verified the correct capacitance of 2.2nF

R9580 and 81 have the correct values when removed from the board and tested externally

Is there any way that tinkering with the feedback voltage divider could solve my cycling problem?
 
Well, I think I figured it out... This board is a trip.

Board was sitting across the room with USB-C power connected with the same behavior as before I removed and replaced the passives, when U9580 suddenly exploded, and the right 12 pads were completely vaporized. After removing the remains of U9580, I found PP2V5_NAND_SSD0 to be completely shorted and PPBUS to be unaffected. The short is now in NAND chip U9100.

My analysis is that U9580 was intermittently failing and originally killed U9400, causing the system to cut power. After I removed the short, the elevated voltage on U9580's output was a result of it being on it's last legs, and when I saw it explode PPBUS was connected to PP2V5_NAND_SSD0, killing a second NAND chip and causing the short to return. Sound right to you?

So, from here... I was planning to replace all the NANDs anyway, is that still realistic in addition to U9580? Or now that we know that PP2V5_NAND_SSD0 probably got hit by PPBUS, is there a high likelihood of severe damage to other circuits such that I should call it quits? Looking at the schematic, U9500 is the only other IC that strikes me as likely to have been impacted.
 
Top