Lenovo NM-D562 v2.0 main EC programming help needed

[davidebeatrici]

Both 0V and OL. How's the continuity/resistance to ground in your case?

I'll check the following signals:

[davidebeatrici]

So... I checked the oscillator pins on the original CPU and in diode mode to ground there is indeed a voltage drop from ~0.4V to ~0.6V depending on the polarity.

I checked the signals anyway and they're as follows.

EC_RSMRST_R_N: 3.3V
CPU_RSMRST_N: 3.3V
CPU_PLT_RST_N: 1.8V
CPU_PWR_GOOD: 3.3V

I'm confident the processor is soldered correctly and I also confirmed continuity to one of the 4 oscillator pins...

[questore]

So... I checked the oscillator pins on the original CPU and in diode mode to ground there is indeed a voltage drop from ~0.4V to ~0.6V depending on the polarity.

I checked the signals anyway and they're as follows.

EC_RSMRST_R_N: 3.3V
CPU_RSMRST_N: 3.3V
CPU_PLT_RST_N: 1.8V
CPU_PWR_GOOD: 3.3V

I'm confident the processor is soldered correctly and I also confirmed continuity to one of the 4 oscillator pins...

I will check them when I'll be at my desk.

[davidebeatrici]

I got a working Rev 2.0 from another 16ACH6H, with a broken screen that stays black. I can measure a voltage drop in diode mode to ground on all oscillator pins and the CPU warms up way before the GPU does.

This confirms the CPU I was sent is indeed dead. I just bought a stencil for the CPU, I'll let you know how it goes when I solder the original back on.

[questore]

I got another 16ACH6H, that POSTs but with a black screen. I can measure a voltage drop in diode mode to ground on all oscillator pins and the CPU warms up way before the GPU does.

This confirms the CPU I was sent is indeed dead. I just bought a stencil for the CPU, I'll let you know how it goes when I solder the original back on.

Meter in diode mode YC1 has 0.748 Voltage drop on both sides to ground , YC2 has 0.746 on one an 0,830 on the other side

[questore]

My measurements:
EC_RSMRST_R_N: 1.7V
CPU_RSMRST_N: 1.7V
CPU_PLT_RST_N: 0V
CPU_PWR_GOOD: 3.3V
I still think its and EC firmware

[davidebeatrici]

Meter in diode mode YC1 has 0.748 Voltage drop on both sides to ground , YC2 has 0.746 on one an 0,830 on the other side

I get ~0.75 on all four but I think your measurements are good.

My measurements:
EC_RSMRST_R_N: 1.7V
CPU_RSMRST_N: 1.7V
CPU_PLT_RST_N: 0V
CPU_PWR_GOOD: 3.3V
I still think its and EC firmware

Interesting how the RSMRST signals seem to be floating in a half state. Not pulled up nor down, a quantum paradox basically!

Let's see if they change once you flash the correct EC firmware.

[davidebeatrici]

The seller sent me a video showing how they test CPUs. That test bench looks amazing.

No idea why the sample I received appears to be dead then. I will desolder it and check the voltage drop directly on its oscillator pins.

[questore]

The seller sent me a video showing how they test CPUs. That test bench looks amazing.

No idea why the sample I received appears to be dead then. I will desolder it and check the voltage drop directly on its oscillator pins.

Maybe too big or small solderballs, or bent motherboard. Check the board for warp when CPU is desoldered.

[davidebeatrici]

I looked at the CPU sideways and compared it to the one on the working motherboard: turns out it was sitting too high, the balls were not melted completely during the soldering process.

I put the motherboard on the IR plate once again, waited for it to reach the target temperature and then kept the hot air nozzle very close to the chip until I saw a reading in diode mode on the multimeter.
The gap now looks identical to the one on the working motherboard and... success! Activity on the BIOS SPI and a perfect sine wave on both oscillators.

Unfortunately it looks like the platform doesn't POST without a GPU installed. I'll have to fix the two broken vital pads in order to confirm a 100% working status.

However, since one of the 4 pads under the CPU is near the border (AY1) I managed to confirm continuity between it and one of the two pins on the 32K crystal.

Well, looks like the blade I inserted under the processor was closing the gap between the pad and the solder ball...

[davidebeatrici]

This makes sense if GPU shorted. I have seen many of them going to afterlife like this.

Actually, as I suspected, 0.00Ω are perfectly normal on the GPU coils. I double checked on the working motherboard.

[questore]

I looked at the CPU sideways and compared it to the one on the working motherboard: turns out it was sitting too high, the balls were not melted completely during the soldering process.

I put the motherboard on the IR plate once again, waited for it to reach the target temperature and then kept the hot air nozzle very close to the chip until I saw a reading in diode mode on the multimeter.
The gap now looks identical to the one on the working motherboard and... success! Activity on the BIOS SPI and a perfect sine wave on both oscillators.

Nice to see some progress, good job.
I have rechecked my GPU (3050) resistance after the partial fulty Mosfet removed and it is 0.200 ohms. Do you have a higher end GPU?

[davidebeatrici]

Oh, you only have a 3050? I have a 3060 on both the motherboard I'm repairing and the working one.

[questore]

The seller sent me a video showing how they test CPUs. That test bench looks amazing.

No idea why the sample I received appears to be dead then. I will desolder it and check the voltage drop directly on its oscillator pins.

Can you point me to the seller? I want to buy trusted CPU-s.

Oh, you only have a 3050? I have a 3060 on both the motherboard I'm repairing and the working one.

That explains the resistance differences.

[davidebeatrici]

Can you point me to the seller? I want to buy trusted CPU-s.

Sure!

IR plate: https://www.alibaba.com/product-deta...426303276.html
CPU: https://www.alibaba.com/product-deta...524376441.html

They also sell the retail version: https://www.alibaba.com/product-deta...244544734.html

That explains the resistance differences.

I can ship you my RTX 3060 (GN20-E3-A1) if you wish, as I decided to install an RTX 3080 (GN20-E7-A1)!

[questore]

Sure!

IR plate: https://www.alibaba.com/product-deta...426303276.html
CPU: https://www.alibaba.com/product-deta...524376441.html

They also sell the retail version: https://www.alibaba.com/product-deta...244544734.html





I can ship you my RTX 3060 (GN20-E3-A1) if you wish, as I decided to install an RTX 3080 (GN20-E7-A1)!

Thanks for the info.
I dont think i need a new GPU since I dont have the hardware to change BGA, its further on the road

[questore]

Programmer arrived and works perfectly. EC firmware was very similar to what I have So I am assuming its another version but OK.
Tryed reprogramming what we have both version acts the same.
Also After reprogrammed the EC I could read the BIOS with a clip for some reason so i decided that i reprogramm the BIOS with the only version i had 1.4 volt on the BIOS pin.
Result is the same but since then I again cannot rewrite BIOS with clip (not even IC detect)

This lead me to a suspected conclusion that the problem could be that something is interferencing with BIOS communication and it makes sense that if I cannot read/detect/write EC when its on the board, then maybe the EC also cannot read it when it communicates.
I am opened to any idea what to check along this logic?

[davidebeatrici]

Perfect timing, yesterday I received GPU and GDDR6 chips. Soldered everything, along with the extra capacitors and resistors, but the board is not powering the GPU nor the VRAM.


According to the power sequence FBVDDQ should come up after NVVDD. The RTX 3080 is absorbing from ~4.5A at 0.8V, a known working RTX 3060 ~1.5A at 0.8V.

After a few power cycles the board didn't want to turn on anymore, I quickly found the cause to be a short to ground on 20V: turns out the first of the five VRMs/MOSFETs is gone...
I removed it and now the board powers up again, but still no voltage to the GPU.

I suspected the VRM controller expects all 5 phases to be alive, but that doesn't seem to be the case: https://www.youtube.com/watch?v=OwnmAAS3gnY
I remember you removed a VRM as well, did you ever replace it? Do you have NVVDD right now?

[questore]

I suspected the VRM controller expects all 5 phases to be alive, but that doesn't seem to be the case: https://www.youtube.com/watch?v=OwnmAAS3gnY
I remember you removed a VRM as well, did you ever replace it? Do you have NVVDD right now?

Multi phase controller should work fine without one mosfet.
Yes I have NVVDD 0.81 and NO I have not replaced the shorted mosfet only removed,

[davidebeatrici]

Programmer arrived and works perfectly. EC firmware was very similar to what I have So I am assuming its another version but OK.
Tryed reprogramming what we have both version acts the same.
Also After reprogrammed the EC I could read the BIOS with a clip for some reason so i decided that i reprogramm the BIOS with the only version i had 1.4 volt on the BIOS pin.
Result is the same but since then I again cannot rewrite BIOS with clip (not even IC detect)

This lead me to a suspected conclusion that the problem could be that something is interferencing with BIOS communication and it makes sense that if I cannot read/detect/write EC when its on the board, then maybe the EC also cannot read it when it communicates.
I am opened to any idea what to check along this logic?

At this point I suspect either the SPI chip is bad or there is something pulling down the communication lines. Could you check both resistance and voltage drop to ground on all of them (pay attention to the polarity/direction)?