Denver LED-32xx Stuck Standby LED on TP.MS3663S.PB818: Flashing EEPROM with CH341A

ivanikonomov8 wrote:

Hello. I have a Denver LED-32xx. It has a stuck Standby LED, but all the voltages are correct. Mainboard is TP.MS3663S.PB818. Panel is ST3151A05-E. i got it like this. I tried flashing, but the dump b...

Initial assessment
Thanks for the details. Symptom is “solid standby LED” on a TP.MS3663S.PB818 with panel ST3151A05‑E. SPI flash is Gigadevice GD25Q64B (8 MByte). You attempted in‑circuit programming with a CH341A; multiple apps report “OK” reads but the dump is unchanged after “writes”. WP# (pin 3) was lifted/cut; board was unpowered during attempts.

Short, direct answer: On this chassis, in‑circuit programming is unreliable to the point of being a time sink. For a definite result, remove the GD25Q64B and program it off‑board. Also restore WP# to its original net after the repair.

Why your in‑circuit attempt fails (and why WP# lift didn’t help)
- Bus contention: Even unpowered, the MSD3663 SoC clamps and loads the SPI lines (ESD diodes, internal pulls). Your programmer and the SoC “fight” for CS#, CLK, DI/DO.
- Back‑powering: The CH341A 3.3 V rail tries to energize parts of the mainboard via the flash’s VCC and the SoC’s clamp diodes. The CH341A cannot supply the resulting inrush/idle current steadily, so the flash sees marginal VCC during erase/program.
- Signal integrity: Long traces + pull networks + capacitive load on CLK/MOSI/MISO/HOLD# degrade edges. Reads may squeak by; page‑program and erase are far more sensitive.
- WP# truth: Lifting WP# only removes one hardware write‑protect path. If SRP/BP bits are set in the status registers, or the bus is contended, writes still won’t take.

Direct answer to your question
- Do I have to remove the whole chip to successfully flash it? Practically, yes. On TP.MS3663S.PB818 the success rate of ISP with a clip is very poor. Off‑board programming (or at least hard isolation of CS#/CLK from the SoC) is the proven, repeatable path.

Detailed problem analysis
1. Confirm the basics on the bench (off‑board)
- JEDEC ID should read C8 40 17 for GD25Q64B.
- Read the two status registers: SR1 (WIP/WEL/BPx/SRP0) and SR2 (QE/SRP1). Before programming:
- Clear all BP bits and SRP bits (write SR1/SR2 after WREN).
- After each program/erase, poll WIP=0.
- Full sequence that should pass on a healthy, isolated chip:
1) Read ID → C8 40 17
2) WREN
3) Chip Erase (C7 or 60)
4) Poll WIP until 0
5) Blank check (all FF)
6) Program in pages (256 B), verify
- If any step above fails off‑board, the device may be worn/damaged. Replace it; 25Q64 parts are inexpensive.

2. Why identical dumps “before/after”
- Many tools will “verify” against the buffer they just wrote rather than a fresh hardware read; with contention the write is ignored and the cached buffer misleads. Always re‑read the entire chip to a new file and binary‑compare.

3. Board‑level specifics that sabotage ISP
- HOLD# (pin 7) must be high; if the board holds it low at any time, DO will tri‑state mid‑transaction.
- Series/0R links on CS#/CLK/DI/DO (if present) tie you to the MSD3663. Unless you lift those links, the SoC remains on the bus.
- “Unpowered SoC is safe” is a myth here; clamp paths to 3V3_STB will drag the programmer down.

Practical guidelines (what to do now)
A) Safest path (recommended)
- Desolder and program off‑board:
- Use hot air 330–370°C (adjust for your station), modest airflow, ample flux, preheat if possible. Shield nearby plastics with Kapton.
- Clean pads with wick; inspect for pad lift.
- Place the chip in a SOIC‑8 adapter. Add a local 0.1 µF across VCC–GND at the adapter.
- Use a known‑good toolchain (flashrom, TL866/T48, EZP2019, or CH341A with reliable software). Set the lowest SPI speed available.
- Procedure: Read→save “ORIG.bin”; Erase→Blank‑check; Program target dump→Verify; Re‑read to “READBACK.bin”; compare READBACK vs target.
- If in doubt about the chip, fit a new GD25Q64 (B/F family or equivalent Winbond 25Q64) pre‑programmed off‑board.

- Refit and restore WP#:
- You cut/lifted WP#. Reconnect it to the original pad/net (often pulled up to 3.3 V through a resistor). Do NOT leave WP# floating.

B) If you insist on trying ISP one last time (not guaranteed)
- Hard‑isolate CS# and/or CLK by lifting the series 0R/link (if fitted) or temporarily cutting and later repairing a convenient test link. Without that, don’t bother.
- Force the SoC into reset (RST# held low) and keep the board otherwise unpowered.
- Power only the flash at 3.3 V from a bench supply limited to ~80–100 mA; common ground with the programmer.
- Pull HOLD# high via ~10 kΩ. Keep clip leads under 10 cm, twist GND with each signal where possible, slow the SPI clock to minimum.
- Still, expect mixed results; off‑board is faster overall.

Current information and trends relevant to this chassis
- Many recent TP.MS3663S.PB818 cases with “red LED only” are fixed by reprogramming or replacing the 25Q64, provided the firmware matches the panel. Cross‑flashed builds boot but show no picture or wrong timings if the panel table is wrong.
- Universal 3663 firmwares exist, but OEM TVs often embed panel parameters and white‑balance data in NVRAM regions inside the SPI. Matching the dump to ST3151A05‑E is important.

After flashing: what to verify if standby remains
- Rails (measured against cold ground):
- 5V_STB present; 3V3_STB solid; 1.2 V core and 1.8 V DDR come up after power‑on.
- SPI activity on power‑up:
- With a scope, you should see CS# toggling and CLK bursts (~tens of MHz) within 100–300 ms after AC apply or POWER press. No activity → SoC/reset/oscillator fault. Repeated short bursts → boot loop → suspect firmware or DRAM.
- 24/27 MHz crystal at the SoC oscillates stably.
- UART console:
- Most 3663 boards expose 3.3 V UART (115200 8N1) on test pads or the VGA block. A boot log is extremely informative and will tell you if U‑Boot starts reading SPI and where it stops.
- Backlight enable:
- PANEL_ON / BL_ON from the SoC should assert after boot; if they don’t but SPI boot is good, the software is still not reaching panel init.

Ethical and safety notes
- Never power the TV mainboard from the CH341A. Do not power the board and the programmer at the same time. Back‑powering risks SoC or USB port damage.
- Verify your CH341A variant: some “black” minis drive I/O at 5 V unless modified. The GD25Q64B is a 3.3 V part.

Practical extras and tips
- Keep two dumps:
- ORIG.bin (what you read first)
- WORKING.bin (verified good image)
- Status‑register sanity before program:
- SR1: WIP=0, WEL=0, BP[2:0]=000, SRP0=0
- SR2: SRP1=0, QE=0/1 (don’t care for standard SPI)
- If your off‑board erase/program is slow or flaky, add a 10–47 µF across 3.3 V near the adapter and drop the speed.

Possible next steps if firmware isn’t the root cause
- If reflash + known‑good dump still leaves standby, suspect:
- Bad 25Q64 (replace with new, programmed off‑board)
- SoC BGA/DDR faults (brief heat test can change symptom)
- Regulator/reset IC fault (brown‑out or POR not releasing)
- Corrupted eMMC/ NAND (not on this variant; your boot is SPI‑only)

Suggestions for further research
- Find/prepare a dump matched to TP.MS3663S.PB818 + ST3151A05‑E. Using a mismatched panel table will boot but yield black/no backlight or wrong image.
- If you need, I can review your ORIG.bin to confirm corruption and extract the correct panel block if present.

Brief summary
- Yes—remove and program the GD25Q64B off‑board. Restore WP#. Verify with a full read‑back compare. Use a dump that matches your panel. If the TV still won’t leave standby after a verified flash, move to SoC reset/clock/SPI activity checks and UART logging.

Questions for you
1. What JEDEC ID do you read from the chip off‑board (expect C8 40 17)?
2. Can you share the sizes/checksums of your “before” and “after” dumps and which firmware file you used?
3. Do you have scope access to confirm SPI activity on CS#/CLK during power‑up?
4. After you restore the flash, do 1.2 V core and 1.8 V DDR rails rise when you press POWER?
5. Do you want a panel‑matched dump for ST3151A05‑E, or should I analyze your ORIG.bin and advise?

ivanikonomov8 wrote:

Hey! I plan on buying a new chip, but I dont have the SOIC-8 adapter. Can i use the clip?

Initial assessment of available information
- Board: TP.MS3663S.PB818; panel: ST3151A05‑E.
- Flash: GD25Q64B (8 MByte). You already tried CH341A + clip in‑circuit; reads “OK” but writes don’t change the dump. WP# (pin 3) was cut/lifted.

Direct answer
- In‑circuit with the clip: Not recommended on this chassis (very low success rate).
- On a loose/new chip: The clip by itself is mechanically/electrically unreliable unless the chip is soldered onto something rigid. Best is a SOIC‑8 adapter or a simple DIY breakout.

Short version: Don’t repeat in‑circuit with the clip. Either buy a $2–$5 SOIC‑8 adapter or use one of the DIY off‑board methods below and program the new chip before soldering it to the TV.

Detailed problem analysis
- Why the clip fails in‑circuit here:
- The MSD3663 SoC loads/clamps the SPI lines even when unpowered (ESD diodes, pulls). This causes bus contention and back‑powering; erase/program operations become unreliable while reads may appear “fine.”
- Lifting WP# doesn’t resolve status‑register protection or bus contention.
- Why a loose chip + clip is problematic:
- The clip is designed to bite a chip that’s soldered to a PCB. On a free chip, contact is intermittent; tiny movements during an erase/program cycle corrupt data.

Practical guidelines (no adapter available today)
Pick one of these to program the replacement GD25Q64B off‑board:

1) “Wired” method (robust)
- Lightly tin the new chip’s pins and solder 8 short, flexible wires (e.g., 30 AWG) to:
- 1 CS#, 2 DO (MISO), 3 WP#, 4 GND, 5 DI (MOSI), 6 CLK, 7 HOLD#, 8 VCC.
- Connect to CH341A 25‑series header accordingly.
- Tie WP# (pin 3) and HOLD# (pin 7) to VCC with 4.7–10 kΩ.
- Add 0.1 µF directly across VCC–GND at the chip.
- Program at the lowest SPI speed; verify, then full read‑back compare.

2) DIY breakout (cleanest)
- Solder the chip onto any small SOIC‑8-to-pins breakout or scrap PCB with SOIC‑8 pads.
- Now you can safely use the clip on that small board (it’s mechanically stable), or plug the breakout into the CH341A via jumper wires as in option 1.

Notes:
- GD25Q64B is a 3.3 V device; do NOT use a 1.8 V adapter. Ensure your CH341A really outputs 3.3 V (some “black” versions default to 5 V unless modified).
- Expected JEDEC ID: C8 40 17. Erase → blank‑check → program → verify → read‑back compare.

After programming and installing
- Reconnect WP# (pin 3) to its original pad/net on the TV (usually pulled up to 3.3 V). Do not leave it floating.
- Use a firmware image matched to TP.MS3663S.PB818 + ST3151A05‑E. Mismatched panel tables can boot but leave a black screen/backlight.

Ethical and safety aspects
- Never power the TV board and the CH341A at the same time.
- Keep wiring short; avoid ESD; don’t stress the new chip’s legs during handling.

Brief summary
- Can you use the clip? Not on the TV board (it already failed), and not reliably on a bare chip unless it’s soldered to a small board. Program the new flash off‑board using a simple wired/breakout setup, or spend a few dollars on a SOIC‑8 adapter. Then solder it in and restore WP#.

Questions for you
- Do you want a quick wiring map photo/sketch for your exact CH341A header?
- Do you have a matched firmware dump for PB818 + ST3151A05‑E, or should I review your original dump and prep one?

ivanikonomov8 wrote:

Oh!I just remembered, it did come with a tiny board, i will see about puttin it there and using the clip.

Initial assessment
Good—using that tiny board will make the job reliable. It’s almost certainly a SOIC‑8 adapter/breakout. With the chip soldered onto it, you eliminate in‑circuit bus‑contention and the clip’s contact problems.

Direct answer
Yes: solder the new GD25Q64B onto that tiny board and program it off‑board. If the adapter has DIP pins for the CH341A’s ZIF socket, plug it in directly and skip the clip (best). If it’s just a flat breakout, then use the clip on the soldered chip (acceptable).

Practical steps with your tiny board
1. Prepare the adapter
- Align pin‑1 (dot/chamfer on the chip) with pin‑1 on the board.
- Solder all 8 pins; inspect for bridges.
- Add a 0.1 µF ceramic between VCC (pin 8) and GND (pin 4) on the adapter.

2. WP# and HOLD# handling
- Ensure WP# (pin 3) and HOLD# (pin 7) are high during programming.
- If your adapter/clip doesn’t carry WP#/HOLD#, add two 4.7–10 kΩ pull‑ups from pins 3 and 7 to pin 8 (VCC) on the adapter.

3. Connect to the programmer
- Best: If the adapter has a DIP‑8 header, place it in the CH341A ZIF per the silk (often at the end nearest the lever). Match pin‑1.
- Using clip: Attach the clip to the chip on the adapter, red wire/notch = pin‑1. Keep the ribbon short and steady.

4. Voltage and software
- Verify the CH341A is really 3.3 V (many “black” minis are 5 V unless modded). Measure VCC on the adapter: ~3.3 V.
- Select GD25Q64B (or 25Q64) in your app. Use the slowest SPI speed first.

5. Program sequence (and checks)
- Read JEDEC ID: expect C8 40 17. If not, fix wiring/voltage before proceeding.
- Read SR1/SR2; clear BP/SRP bits if set (most tools have “unprotect”).
- Chip Erase → wait for WIP to clear → Blank Check (all 0xFF).
- Program your PB818 + ST3151A05‑E‑matched file.
- Verify, then do a fresh full Read‑Back to a new file and binary‑compare to the source. This is your pass/fail.

6. Install on the TV
- Desolder old chip; clean pads.
- Solder the programmed chip, correct orientation.
- Restore WP# (pin 3) to the original net—do not leave it floating. If you previously cut a trace, add a short jumper to the pull‑up point.

Notes specific to your case
- In‑circuit with the clip already failed for good reasons (MSD3663 loads the SPI bus). Off‑board via the tiny board avoids that completely.
- Use a firmware dump that matches TP.MS3663S.PB818 + ST3151A05‑E; a mismatch can boot yet leave no picture/backlight.

Brief summary
Solder the new GD25Q64B onto the tiny board, ensure 3.3 V, pull WP#/HOLD# high, program slowly, and verify with a full read‑back compare. Then fit it to the mainboard and restore WP#. This method is robust and fixes the “identical dump” issue you saw with in‑circuit programming.

Questions for you
- Does your tiny board have DIP pins for the CH341A ZIF socket, or is it a flat breakout? I’ll tailor the exact pin‑1/ZIF position for your CH341A version.
- Do you already have a panel‑matched firmware file for ST3151A05‑E, or should I review your original dump and prepare one?

ivanikonomov8 wrote:

@ivanikonomov8 @ivanikonomov8 @ElektrodaBot

Initial assessment of available information
- Mainboard: TP.MS3663S.PB818 (MStar/MediaTek “universal” chassis).
- Panel: ST3151A05‑E.
- You have a “working dump” and are considering buying another PB818 board that’s currently set up for a different panel.

Direct answer to your question
- If by “working dump” you mean a raw 8 MB SPI flash image read with a programmer (CH341A), then: No—this cannot be applied via USB on the TV. USB updaters expect a special upgrade package, not a raw dump.
- If you have an actual USB upgrade package (vendor/service “allupgrade…”/“msd3663…/MstarUpgrade.bin”), then: Yes—USB update is possible, even if the new board’s current panel settings are wrong.

Detailed problem analysis
1) Raw SPI dump vs USB upgrade package
- Raw SPI dump (8,388,608 bytes typically for 25Q64): Exact byte‑for‑byte contents of the flash. The TV’s bootloader does not understand this format from USB.
- USB upgrade package: Contains headers, partitioning and scripts (bootloader expects specific file names and structure). This is what the “USB recovery/update” routine loads.

2) What “same model mainboard” guarantees—and what it doesn’t
- The PCB being TP.MS3663S.PB818 means power tree and pinout match. But firmware on replacement boards often targets another brand/panel, so:
- Remote/keypad mapping may differ.
- Panel timing/EDID table may be wrong.
- LVDS voltage select (5 V vs 12 V to panel) must be set correctly by jumper before you ever power the panel.

3) Why USB may still work with the “wrong panel”
- The bootloader runs before panel init. So a blind USB update can still proceed (LED flashes), even if the image on screen is black/garbled. This only works if you have the correct USB package and the board’s USB update key combo matches.

4) When USB will not help
- If you only have a raw dump; or the board’s bootloader is different/locked; or keypad mapping prevents the update trigger.
- In these cases, off‑board programming of the SPI (clone your working dump into the new board’s 25Q64) is the reliable method.

Practical guidelines
A) Safest/fastest route if you have only a raw dump
- Do not power the new board with your panel connected yet.
- Read and save the new board’s original SPI (backup).
- Program your known‑good raw dump into the new board’s 25Q64 off‑board (use your tiny SOIC‑8 adapter/board + CH341A).
- Set the panel voltage jumper on the new board to match your old board (5 V or 12 V). Only then connect the panel and test.

B) If you have a proper USB upgrade package (not a raw dump)
- Prepare a small FAT32 USB stick; copy the upgrade file to root and rename exactly as required by that package (common names: allupgrade_msd3663_8G_1G.bin, msd3663.bin, MstarUpgrade.bin).
- For the first flash, disconnect the panel LVDS and panel power (to avoid damage in case the VSEL jumper is wrong).
- Insert USB, hold the board’s local POWER (or VOL+) key, then plug AC. Watch the standby LED; fast blinking usually indicates update in progress. Wait until it stops/reboots (1–5 minutes). Remove USB.
- Power off, set the panel voltage jumper to the correct position, reconnect the panel, then power on and test.

C) Critical safety check: Panel supply selection
- PB818 typically has a 3‑pin jumper (VSEL) selecting 5 V vs 12 V to the panel/T‑CON. If this is wrong you can instantly kill the panel.
- Action: Before connecting your ST3151A05‑E, set the jumper on the replacement board to the same position as on your original board.

D) After boot with the “right” firmware
- If picture is still wrong, enter service menu and select the exact panel model/timing (or the correct LVDS map/bit‑depth). Some vendor builds expose a Panel Select list; others read a fixed table.

Supporting explanations and details
- Why USB can’t take a raw dump: The loader searches USB for a file with known header/signature, then stages it to RAM and writes specific regions. A raw dump lacks these markers and would be treated as opaque data.
- Why off‑board cloning works across identical boards: Bootloader, Linux/app partitions, NVM calibration, and remote codes are all captured in the dump; cloning reproduces the original configuration, including panel table and WB/ADC data.

Ethical and safety aspects
- Never power the mainboard and the CH341A at the same time.
- Confirm your CH341A outputs 3.3 V to the flash.
- Don’t connect the panel until you’ve checked VSEL (5 V/12 V) and are confident the firmware matches.

Practical checklist
- Do you have: raw 8 MB dump (.bin) or a USB upgrade package? Decide path accordingly.
- Backup the new board’s flash before any write.
- If using USB, disconnect panel first; reconnect only after update and VSEL check.
- Keep a verified read‑back compare after programming.

Possible disclaimers or additional notes
- Some PB818 variants use different key combinations to trigger USB update (Power, Vol+, or OK). If one combo doesn’t start the LED blinking, try another. Lack of LED activity usually means the bootloader didn’t enter update mode or the filename is wrong.
- Remote control codes may differ until firmware is replaced—use the side keypad for the update trigger.

Suggestions for further research
- Obtain a USB package specifically for TP.MS3663S.PB818 configured for ST3151A05‑E (or a close equivalent panel). Keep both the raw dump and the USB package on hand.
- If needed, we can validate your dump integrity and identify the panel block.

Brief summary
- USB flashing: only with a proper upgrade package; not with a raw SPI dump.
- If you only have a raw dump, clone it to the new board’s SPI off‑board with your CH341A.
- Before connecting your panel, set the board’s panel voltage jumper to match your old board.

Questions for you
1) Is your “working dump” a raw 8 MB SPI image, or do you have an “allupgrade/msd3663” USB package?
2) Can you confirm the panel voltage jumper position on your original board (so we mirror it on the replacement)?
3) Do you want me to check your dump (size/hash) and tell you if it’s suitable to clone 1:1 to the new PB818?