Dell XPS-15 (9520) Mostly Bricked after common BIOS-Firmware upgrade - Full Repair Log

**Tesla** · 03-19-2026, 08:27 PM

So, taking into account:
- How it got Bricked in the first place (um, how it decided to Brick itself)
- Some "signs of life" but not really fully Posting any more
- No physical damage
- None of the published CMOS Resets help
- BIOS Recovery non-responsive

Do yall agree that I have exhausted all the "easy fixes" and is now a candidate for a BIOS chip code fix?

As I understand the procedure, I should now remove the motherboard, find the BIOS chip on the back, clip onto it, and try to get a good dump of the BIOS with my new XGecu-T76?

**m1ch43lzm** · 03-19-2026, 10:34 PM

Looks similar to this case: https://youtu.be/t2TBtKS8bqM?t=908
It's an XPS 15 9500, but it should be similar procedure
Make sure to get WSON8 probe, shown in that video, the SOIC8 clip won't work; you have to hold it for however long it takes to read the chip, also lower the SPI clock frequency to 8MHz on the XGecu software otherwise you get read errors/random changed bits due to extra wire length between programmer and chip
When using this type of probe with pogo pins, read the chip several times and make several backups, compare them with hex editor, should be identical, a slight movement can cause to lose connection and some portions won't be read correctly

S/T and Win key have to be transferred to new BIOS

**Tesla** · 03-20-2026, 09:59 PM

Originally posted by m1ch43lzm

1. Looks similar to this case: https://youtu.be/t2TBtKS8bqM?t=908
It's an XPS 15 9500, but it should be similar procedure

2. Make sure to get WSON8 probe, shown in that video, the SOIC8 clip won't work; you have to hold it for however long it takes to read the chip,

3. also lower the SPI clock frequency to 8MHz on the XGecu software otherwise you get read errors/random changed bits due to extra wire length between programmer and chip

4. When using this type of probe with pogo pins, read the chip several times and make several backups, compare them with hex editor, should be identical, a slight movement can cause to lose connection and some portions won't be read correctly

5. S/T and Win key have to be transferred to new BIOS

1. Nice find, thanks.

2. I ordered one, but it shipped separately. It will be here in a few days.

3. Acknowledged

4. Acknowledged

5. Acknowledged

It's a Winbond 25R512JVEN . The supplied clip does clip-on and stay. If it was made with a bit more precision, I think it would actually work. Also, it doesn't help that there is a very tiny 5-leg component soldered onto the board very close that the clip's black plastic might be touching. The pictures and actual pinouts in the XgPro are nice (like for this supplied clip and I assume other adapters)

With it set to W25R512JV , I notice it only wants to read Pins 1,2,5,6,8 (according to the failed pin-detection).

**Tesla** · 03-21-2026, 12:11 PM

Just killing some time waiting for my QFN8-6X8 Pogo-Clip .

I loaded Compal HDO51_HAD51 LA-L402P r1.0.cad into Open-Board-View.

Questions:
1. Why do I have to "mirror" view to make it look the same as the board on my bench? That insinuates a major physical change of the PCB (not just a different view).

2. This laptop is fully optioned, yet JDCI1 and JDUDC1 connectors (on top, easy to get to, near the SSD M.2 slots) are empty.
a. Do you think Compal added some Diag/Programming ports (could be used by only removing laptop's metal back-cover)
b. Or, if this same MB is really also used in Precision-5570 ... do those somehow have an option this one doesn't?

**m1ch43lzm** · 03-21-2026, 01:00 PM

SOIC8 clip won't work with WSON8 package, it's designed to touch the legs on a SOIC8/SOP8 chip, not with the pads on the motherboard (WSON8 chip doesn't have legs)

1. Who knows, it depends how it was exported from the original CAD program, if mirror works then it's what it is... Several boardviews are like that

2. a. JDUDC1 looks like a debug connector for the prototype version of the board that's removed from final production version, used for programming SPI flash and/or JTAG debug, there's a mux chip U740 on the back side of the board for that purpose (page 79 of schematics)
But, since U740 is marked with DUDC@, it may not be present on the board at all; anything marked @ on the schematics means it's an option which may or may not be present on your board (removed from final production version)
b. Different GPU (RTX A1000 or A2000) for the Precision variant, instead of GeForce RTX 3050/3050 Ti for the XPS

**Tesla** · 03-21-2026, 02:21 PM

Originally posted by m1ch43lzm

0. SOIC8 clip won't work with WSON8 package, it's designed to touch the legs on a SOIC8/SOP8 chip, not with the pads on the motherboard (WSON8 chip doesn't have legs)

1. Who knows, it depends how it was exported from the original CAD program, if mirror works then it's what it is... Several boardviews are like that

2. a. JDUDC1 looks like a debug connector for the prototype version of the board that's removed from final production version, used for programming SPI flash and/or JTAG debug, there's a mux chip U740 on the back side of the board for that purpose (page 79 of schematics)
But, since U740 is marked with DUDC@, it may not be present on the board at all; anything marked @ on the schematics means it's an option which may or may not be present on your board (removed from final production version)
b. Different GPU (RTX A1000 or A2000) for the Precision variant, instead of GeForce RTX 3050/3050 Ti for the XPS

0. Agreed. The metal contacts end-up around 0.5 mm away from making contact. It would definitely have to be modified (hand-filed-down at the very least).

1. Thanks, that's what I needed to know.

2a. Very interesting (so, kinda what I thought). Yes, and you can see by the full motherboard model-number, this is a Rev: 0.3 (X02) . As you might expect, it also initially came to me (for evaluation) with an early "X" experimental-bios. However, I've always just treated it like a "released machine" and it's done fine. BIOS-Firmwares always applied fine, Intel-ME and sub-controllers updated properly, all as it should be. I'm retired, so I don't work for them any more but that was the mission ... use normally for real, keep current (firmwares and OS), attach and test peripherals and networking ... evaluate and report.

2b. Acknowledged. That's what I've heard, but never seen or used-one in real-life (and certainly never been inside one).

**Tesla** · 03-21-2026, 03:04 PM

Originally posted by m1ch43lzm

Make sure to get WSON8 probe, shown in that video,

Ok, I got the pogo-WSON8 , and it got past the Pin-Detection on first try 😀

I have XgPro set to "W25R512JV @WSON8" and my chip on MB is a Winbond 25R512JVEN.

EDIT:

The Check-ID option is also working (never-mind what I said before)

The Verification after Read passed for both separate reading procedures
I got the same log-window CheckSum for both reads
I selected FLASH on the Select-Buffer option when I saved each file.

So, while I got the same Checksum for both tries, what is the accepted way of comparing the files (to be sure they exactly match) before uploading one?

Also, should I upload them here, or inside the "BIOS Request" sub-forum?

So, with a T76 ( and a pogo-clip with a 11-inch cable) are you sure we have to slow all the way down to 8Mhz?
The reason I ask is because it's recommending/defaulting to 40Mhz. But you are warning against excessive speed, so I did one of mine at 8Mhz and then the other at 16Mhz.
The reason I ask is because it (8mhz vs 16mhz) takes twice as long, and the longer you have to hold it risks an error.
I don't mean to question you (the expert) just asking.

Thanks again for all your help.

**m1ch43lzm** · 03-21-2026, 05:47 PM

ID check should return EF 40 20 for W25R512JV, it could be a bad connection, or the rest of the circuit is interferring with proper read

You got a prototype/debug unit, didn't zoom in your picture so I didn't notice JDUDC1 was fitted to the board, in theory it should be possible to read the BIOS chip using that connector
Last resort is desoldering the chip

**Tesla** · 03-21-2026, 06:18 PM

Originally posted by m1ch43lzm

1. ID check should return EF 40 20 for W25R512JV,

2. it could be a bad connection, or the rest of the circuit is interfering with proper read

3. You got a prototype/debug unit, didn't zoom in your picture so I didn't notice JDUDC1 was fitted to the board, in theory it should be possible to read the BIOS chip using that connector
Last resort is desoldering the chip

1. Yes, ID: 0x EF 40 20

2. Please see my edit post above ... It's working fine now. Sorry, but I think I had something set a bit wrong. I'm still trying to get familiar with the software and its use.

3. So I'm fine. Yeah, I never considered that they might only be on the early-revision boards. I just had a feeling that if they Brick this easy, Compal likely has a way of re-writing the roms without pulling the motherboards.

Please respond to my edited post above when you get a chance. I looked-up Peru ... I'm in Texas.

**m1ch43lzm** · 03-21-2026, 07:39 PM

I've had problems reading chips at the fastest speed (30MHz on the T48) when using clip/pogo adapter before, random bytes being different each time; there's also a note on Xgpro software that says "(Reduce clock frequency if using clip or with ISP port)", the faster speed works fine with a chip desoldered from the board and placing it in the appropriate adapter (in your case WSON8/DFN8 8x6, pictured below)

If 16MHz works fine and you get consistent reads (binary identical), then that's great, since you have a 64MB BIOS chip, otherwise it'll take ages to read and write to the chip
* Additional note. I've had issues with SOIC8 clip and using it with the blue (or green) PCB that has header pins and solder pads for SOIC8 and SOIC16 chips, that thing wouldn't work properly at anything higher than 4MHz; the DIP8 to header pins adapter works better

For comparing binary files i use WinMerge, it will highlight differences in yellow, otherwise it will tell you that both files are identical, you can also use UEFITool to open the BIOS file, it should show as Intel image, expand it and it should show Descriptor, ME region, BIOS region

Upload in BIOS request forum, in the Dell XPS 9520 bios thread, https://www.badcaps.net/forum/troubl...520-bios/page2
Also type the service tag from bottom cover ST: (it's the rules there https://www.badcaps.net/forum/troubl...before-posting )
Don't know if prototypes have a service tag like retail units, mention that if it doesn't have one

3. So I'm fine. Yeah, I never considered that they might only be on the early-revision boards. I just had a feeling that if they Brick this easy, Compal likely has a way of re-writing the roms without pulling the motherboards.

Retail units don't have those connectors installed, just the solder pads are there (engineers don't bother to remove those "leftovers" when designing the PCB, just remove them from the BOM, it saves few cents for each board..)

**Tesla** · 03-21-2026, 09:30 PM

Originally posted by m1ch43lzm

1. I've had problems reading chips at the fastest speed (30MHz on the T48) when using clip/pogo adapter before, random bytes being different each time;

2. there's also a note on Xgpro software that says "(Reduce clock frequency if using clip or with ISP port)",

3. the faster speed works fine with a chip de-soldered from the board and placing it in the appropriate adapter

4. If 16MHz works fine and you get consistent reads (binary identical), then that's great, since you have a 64MB BIOS chip, otherwise it'll take ages to read and write to the chip

5. Additional note. I've had issues with SOIC8 clip and using it with the blue (or green) PCB that has header pins and solder pads for SOIC8 and SOIC16 chips, that thing wouldn't work properly at anything higher than 4MHz; the DIP8 to header pins adapter works better

1. Kudos to you for identifying the problem and discovering a viable work-around.

My thoughts are we shouldn't take for granted that these programmers were really designed to have the chips legs inserted directly into the main-socket (no extra connectors and long-ish wires).

2. Yes, I saw that, but they don't really say by how much. Maybe just 50% is enough? Maybe it's 8-Mhz for T48, but upgraded T76 (with the extra power-amps of USB-3.0 ... and I think internal PCB is different) can do more like 16-Mhz ? (and still be toned down safer). To me, Speed means nothing if its not reliable and/or viable.

3. Sure, but the skill to do that (successfully and without further damage) is not trivial. Still getting successful results with the pogo-tool is fairly amazing. I don't mean to jinx-it, but with yall's help, I might actually be able to turn this Brick back into a working laptop.

4. Cool and yes, that's true (even 34 seconds seems like forever). And I forget if the Verify-After pass is included or extra.

5. Interesting ... another good observation and workaround.

In XgPro, for motherboard-chip in XPS-15 (9520): Winbond 25R512JVEN
select "W25R512JV @WSON8"
-----------------------------
W25R512JV
Passes ID-Check: 0x EF 40 20
64MB-BIOS
67,108,864 bytes
34-sec reads at SPI-Clock of 16-Mhz

**Tesla** · 03-22-2026, 03:18 PM

Originally posted by m1ch43lzm

Upload in BIOS request forum, in the Dell XPS 9520 bios thread, https://www.badcaps.net/forum/troubl...520-bios/page2
Also type the service tag from bottom cover ST:

OK, I uploaded it.

**Tesla** · 03-22-2026, 06:19 PM

m1ch43lzm ,

My fixed file is ready and I downloaded it.

I did a little research and this is what I came up with.

I will first use separate Erase and Blank-check. Seems required so I'll get that done and out of the way.

Using the same options as before:
- Selected IC: W25R512JV @WSON8
- Pin Detect, Verify After, Skip Blank, and Check-ID
- Address Range: ALL
- Unprotect Before Programming.
- 16Mhz SPI-Clock (or maybe 8-Mhz if trouble Verifying at end)

I'll LOAD the new .BIN-file (using that window's default-settings, as I was later instructed)
Finally, click Prog/Write button.
As usual, look for a good/perfect Verify.

Looks like the motherboard will need to be pretty-much fully-reinstalled to see if it all worked.

EDITED

**m1ch43lzm** · 03-22-2026, 09:00 PM

The way you do it looks OK to me, you can rest your hand between erasing and writing
Or, you can do it in one go if you tick "Erase before", you just have to hold the probe for longer time while it does Erase - Write - Verify

If all goes well, the first boot will take a very long time, it may be stuck at Dell logo or a black screen (some Dell laptops will display a Dell logo even without RAM installed, I think the logo is programmed into the screen somehow)

**Tesla** · 03-22-2026, 09:51 PM

Erasing and Blank (check) were fairly straight-forward (they should both just run with no errors). The Erase took like 4 minutes for this 64MB-chip. I'm still using the 16-Mhz Clock Frequency (speed).

Strangely, out of no-where ... the T76 freaked-out and started complaining about too-many Vcc amps or a short (on all functions, even READ-ing). First time that happened... I thought my chip/mb got fried somehow.
Simply power-cycling the T76 fixed it. I almost had a heart-attack.

I recall that during a file SAVE, I selected FLASH on the Select-Buffer option ...

Do I now select "FLASH" for Buffer Region on file-loading?

**Tesla** · 03-22-2026, 10:00 PM

Originally posted by m1ch43lzm

1. The way you do it looks OK to me, you can rest your hand between erasing and writing

2. If all goes well, the first boot will take a very long time, it may be stuck at Dell logo or a black screen

3. some Dell laptops will display a Dell logo even without RAM installed,

1. Thanks for the verification.

2. Acknowledged

3. Yeah, I think this one does that.
Also, even though it's soft-bricked, the Display-Test still runs like nothing is wrong. One of the only signs-of-life ... very strange.

**m1ch43lzm** · 03-23-2026, 09:58 AM

Originally posted by Tesla

I recall that during a file SAVE, I selected FLASH on the Select-Buffer option ...

Do I now select "FLASH" for Buffer Region on file-loading?

Just leave the default settings when loading the file, you don't need to worry about that, result is the same anyway as long as you don't change anything else
Only when reading/saving the file, as the latest versions of Xgpro software append a -Configuration- thing at the end of the file with a copy of the status registers, that can be safely trimmed with hex editor

**Tesla** · 03-23-2026, 10:32 AM

Originally posted by m1ch43lzm

Just leave the default settings when loading the file, you don't need to worry about that, result is the same anyway as long as you don't change anything else

I'm glad I asked. So, options should look like this when loading:

Thanks.

**m1ch43lzm** · 03-23-2026, 11:09 AM

That should work, if I remember correctly the options get reset every time you click on load
I'm not near my PC right now, I'm writing this from my phone

Dell XPS-15 (9520) Mostly Bricked after common BIOS-Firmware upgrade - Full Repair Log

Dell XPS-15 (9520) Mostly Bricked after common BIOS-Firmware upgrade - Full Repair Log

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Related Topics