HiSense 65H6510G no boot

[howardc64]

Re: HiSense 65H6510G no boot

One things important. on philips Android TV with same Soc ..we do have here many threat dealt with... Soc BGA caused dead /reboot. what we do know where are the BGA points.. what if this BGA was on of eMMC IO data rail !!!!?

just out the topic .. not all on the net are correct neither other issue to be general to other.
i still believe eMMC not dead. unless rework BGA of the Soc

Yes, certainly can be the SOC of course.

I think you are saying failure could be the common BGA lead-free solder ball cracking? Which is of course solved by reflow if BGA isn't damaged. I can try limited reflow effort (have pre-heater, hot air station but no BGA reflow station)

as the PWB always high to make warm start ( Android ) mean the Soc are always under currents stress even if its at STBY mode.

Have a question on PWB always high for warm start... I obviously can't not check STBY mode power rail status

Post #1 schematic power tree diagram shows SOC CPU and logic rail is enabled by POWER_ON/OFF. Can not find how this is turned on in schematic but I'll assume it means turning on the TV. If true, then SOC is turned off on STBY. This is surprising. I would have assumed at least 1 CPU need to be kept alive in low frequency low power state in STBY.

3.3V_Standby does keep DRAM powered up. Presumably for faster warm boot.

Can you explain what PWB always high means in more detail?

Worth a shot to replace it tho, if properly equipped to program it and the chip is 5 bucks.

yeah, not too many BGA connections but is probably at the limit of my hot air station skills Also need to research how to install boot + firmware. Looks like crypto signing is necessary.

[lotas]

Re: HiSense 65H6510G no boot

No, it’s definitely a dead eMMC, especially since it’s Hynix, which is afraid of heating, they also die in Sony TV, and in order to pull the data from the programmer via the ISP interface (without unsoldering it from the board and when freezing), because when soldering, when heated with hot air, it dies completely.

[Diah]

Re: HiSense 65H6510G no boot

Have a question on PWB always high for warm start... I obviously can't not check STBY mode power rail status

TV operation systems are different as you know.. Android system TVs or TV boxes they didn't have in circuit any SPI where developer can arrange boot script to let the TV parts start one after others. Android warm start as soon you press power on TV or on RC this command will read by eMMC / NAND where the boot / Uimage/ Kernel control the rest. so boot are always in tact / active. ( in this case the PWB are active high... and some line if need more high it will switch for example 9V to 13V ) ( while the 9V called STBY)

if you plug the tv out from wall .... then plug it in the boot will start from 0 sector .. ( this is the best way to trace boot log ) not just press power at time TV ready plugged in wall.

kernel init.rc include the function of i2c where they saved under /dev/i2c will play the function of other TV system where SPI in place /Mboot.

take for example Android TV box. we can see 5V adapter always on... or 12V always on in case there are DVB Tuner installed on the box.

so what are the different between Android TV box and TV. when both are just CPU RAM emmc/ nand.

[Diah]

Re: HiSense 65H6510G no boot

so how could emmc dead if there response to RC ? or power button !! i like to learn too...
RC Map saved on emmc...

[howardc64]

Re: HiSense 65H6510G no boot

No, it's definitely a dead eMMC, especially since it's Hynix, which is afraid of heating, they also die in Sony TV, and in order to pull the data from the programmer via the ISP interface (without unsoldering it from the board and when freezing), because when soldering, when heated with hot air, it dies completely.

I have been wondering about heat damage to eMMC. Inspected 4 different main boards. Its an interesting observation on these no forced air cooled main boards.

• ~2015 Vizio with frequent SOC BGA solder ball thermal cracks. Big heat sink (older gen logic chip) eMMC mounted below SOCs.
  
• ~2021 Samsung. eMMC mounted below SOC. Hole in heat sink over eMMC.
  
• ~2018 LG. eMMC mounted above SOC. Hole in heat sink over eMMC.
  
• 2020 HiSense (This TV) eMMC mounted above SOC. No hole in heat sink over eMMC.

Newer TVs have much smaller/lower mass heat sinks of course as they benefit from newer logic semiconductor process (lower switching voltage and leakage)

These boards of course have different eMMC technology/generation and SOC semiconductor fabrication generation. In general, cheap Chinese TVs use domestic/Taiwan cheap SOC manufactured on older semi-conductor fab with worse heat/logic density.

[howardc64]

Re: HiSense 65H6510G no boot

so how could emmc dead if there response to RC ? or power button !! i like to learn too...
RC Map saved on emmc...

Unfortunately, this TV doesn't respond to Remote Control or Power Button.

[lotas]

Re: HiSense 65H6510G no boot

But the Boot section is still alive, and the User section has collapsed.

[Diah]

Re: HiSense 65H6510G no boot

Unfortunately, this TV doesn't respond to Remote Control or Power Button.

oops. this first time you stated this.... then possibility of toast eMMC in place beside Soc BGA

[lotas]

Re: HiSense 65H6510G no boot

This is how the life of eMMC on the programmer shows, and, strangely enough, also Hynix.
https://www.badcaps.net/forum/showpo...2&postcount=19

[howardc64]

Re: HiSense 65H6510G no boot

But the Boot section is still alive, and the User section has collapsed.

I guess you are thinking UART output shows eMMC boot code is retrieved by SOC. Ran the boot code and failed on user partition.

I wonder if this is true. If CMD0002 (post #28)is indeed CMD2 of eMMC spec, then it is basic bus initialization process of eMMC (described in post #28's attached boot PDF file) so I guess before any NAND flash block transfer. But I don't know if UART output of CMD002 is this meaning or eMMC initialization details. I am new to this level of detail on eMMC.

I was guessing BIST maybe performed by first boot code loaded into SOC at manufacturing (Probably there is small embedded flash inside SOC)

This is how the life of eMMC on the programmer shows, and, strangely enough, also Hynix.
https://www.badcaps.net/forum/showpo...2&postcount=19

Thanks for the link. Some notes on the post's eMMC status

eMMC mode 1-bit makes sense. eMMC spec says all eMMC has to support lowest common denominator 1-bit data mode (instead of parallel 4bit or 8bit mode if have the wider data connection) for compatibility with any eMMC host controller. eMMC Programmer surely use 1-bit mode when starting.

Type A and B memory lifetime code shows it was worn out by wear leveling. So too much writes and used up majority of NAND flash write limits. For a 2015 SONY and 16GB (GBytes) of MLC (multi-level cell) flash. It shows poor software making too much writes to the eMMC. My Tesla entertainment console eMMC died the same way (post #30)



FYI, MLC cell write limit is < 10k while TLC is < 1k! I think all these lower eMMC capacity eMMCs for TVs etc (compared to a computer drive) use MLC (my eMMC says MLC) There is not enough NAND cell to spread out wear leveling to much larger capacity like computer drives (>= 128GB and typically >= 256GB) which can use TLC

[nomoresonys]

Re: HiSense 65H6510G no boot

Here's a question, does turning the tv completely off, as in turning off surge protection power strip over night, does that wear out the eMMC quicker than leaving the tv in standby with the strip just left on so tv is always in standby when not turned on?

[Diah]

Re: HiSense 65H6510G no boot

i found my self lost and in wrong again..
after i reread from start.. yours #16 post #5 #7

in #5 sound funny to me either #7 so let me talk on post #16
you have STB 3.4V then why you don't disconnect the MB and power the PWB as stand alone to see if the BL will power on and the 12V rail / 20 stable ??
i know you used tester for LED but tester i don't believe on it...

[howardc64]

Re: HiSense 65H6510G no boot

i found my self lost and in wrong again..
after i reread from start.. yours #16 post #5 #7

in #5 sound funny to me either #7 so let me talk on post #16
you have STB 3.4V then why you don't disconnect the MB and power the PWB as stand alone to see if the BL will power on and the 12V rail / 20 stable ??
i know you used tester for LED but tester i don't believe on it...

Have disconnected MB and PWB. 12v rail shows 9v (post #7) and no backlight (post #10)

The fact 12v rail shows 9v and LED doesn't light up with main disconnected is strange to me. But I have no idea how this PSU is suppose to behave.

I did use multi-meter DC and AC mode to check 12v with main connected. 12V DC Is steady and no AC value which is best I can do to check for noise without scope.

Certainly PSU seems strange but would be amazing to allow all proper voltage generation on main (I've checked AC component on all main voltage rails. All clean)

[Diah]

Re: HiSense 65H6510G no boot

okay. let us forget all the post you did before...

can you listed all V you have on the socket with out MB plugged in... and name the pin please

[howardc64]

Re: HiSense 65H6510G no boot

Here's a question, does turning the tv completely off, as in turning off surge protection power strip over night, does that wear out the eMMC quicker than leaving the tv in standby with the strip just left on so tv is always in standby when not turned on?

For NAND flash, I think its all dependent on write cycling. Read cycles are not a problem.

There are many possibilities to do software improperly (my background is system software, HW architecture. Developed system with flash memory before) There are lots of caches with different access speeds, retention and wear leveling.

SLC

I don't know if eMMC (or SOC) include SLC (single level cell) which has 100k+ write cycles. Systems use SLC to gather up random writes to a large block to reduce writes to MLC/TLC NAND cells (you can't write single bit NAND cell, its always a block write)

RAM Caching

One would hope video streaming temporal caching is done completely in RAM to avoid killing the flash memory.

Logging. Software are usually written with multiple logging levels including maximum when debugging. Need to put the log in non volatile memory (flash) just in case the system crash. So if software guys forget to drastically reduce logging in normal use, then eMMC gets worn out like my Tesla entertainment system haha.

Anyway, whenever downloading stuff or apps produce some write info. Software developers need to think about 1) how much data 2) how scattered 3) where to write 4) write frequency to NAND flash etc etc.

BTW, this is why Samsung SSDs have been the premium and most reliable historically. They have been using DRAM caching on nVME/SATA SSDs since early days. All the cheap SSDs are DRAMless and rely on wear leveling to survive longevity so they fail with heavy computer usage in just a few years. Some SSDs add SLC as write buffering to reduce MLC/TLC NAND flash wear.

But to answer your question, I don't know unfortunately because it depends so much on actual cache hardware components and software implementation which is invisible. This invisibility is the #1 challenge of embedded software for repair effort.

[howardc64]

Re: HiSense 65H6510G no boot

okay. let us forget all the post you did before...

can you listed all V you have on the socket with out MB plugged in... and name the pin please

All DC/AC voltage readings stable (Except NC when main is disconnected as noted)

XP810 (Main Disconnected)

NC (0V DC 0.2V AC and keeps dropping towards 0)
ANA (0V DC 0V AC)
18V/12V (20.4 DC 0V AC)
18V/12V (20.4 DC 0V AC)
GND (0V DC 0V AC)
STB (0V DC 0V AC)
12V (9.42V DC 0.07V AC)
12V (9.42V DC 0.07V AC)

BSW (0V DC 0V AC)
BRI (0V DC 0V AC)
GND (0V DC 0V AC)
GND (0V DC 0V AC)
NC (0V DC 0.2V AC and keeps dropping towards 0)
GND (0V DC 0V AC)
12V (9.42V DC 0.07V AC)
GND (0V DC 0V AC)

XP810 (Main Connected)

NC (0V DC 0V AC)
ANA (0V DC 0V AC)
18V/12V (20.55 DC 0V AC)
18V/12V (20.55 DC 0V AC)
GND (0V DC 0V AC)
STB (3.45V DC 0V AC)
12V (12.5V DC 0V AC)
12V (12.5V DC 0V AC)

BSW (0.085V DC 0V AC)
BRI (0.085V DC 0V AC)
GND (0V DC 0V AC)
GND (0V DC 0V AC)
NC (0V DC 0V AC)
GND (0V DC 0V AC)
12V (12.5V DC 0V AC)
GND (0V DC 0V AC)

DC/AC reads same with LED connected or disconnected.

[nomoresonys]

Re: HiSense 65H6510G no boot

It does seem strange how the stby is only there with mainboard connected, is that a new thing they are doing or a fault?

[Diah]

Re: HiSense 65H6510G no boot

so the things are normal.. just to clear what funny Chinese as they like put their own sign label
i will focused on STB which are 0 V when MB out from circuit.. and then read 3.45V at time MB plugged in.

STB Chinese its PWR_ON ( power on.) which there are step_down circuit on MB which take from 12V rail ( 9.42 V ls stand by ) generate 3.5V power on which it send to PSU to power 12V (9.4V stand by ) to high as it should

the STB ( power on which generated with MB step down should be send too to power on the BL circuit. ( this didnt happened because its firmware script ) also what you need:

step downer circuit take out the rail 9.4 ( 12V) to generate 3.5 V feed it to STB... this will make 9.4 high to 12... and feed the 3.5V from yours step down circiut via 1K R to other BSW BRI to turn on the BL.... BRI could it need more R value how much it less the brightness on LED BL will be high... and the reserve wise.

of course this need to do with out main boards in circuit.

[howardc64]

Re: HiSense 65H6510G no boot

so the things are normal.. just to clear what funny Chinese as they like put their own sign label
i will focused on STB which are 0 V when MB out from circuit.. and then read 3.45V at time MB plugged in.

STB Chinese its PWR_ON ( power on.) which there are step_down circuit on MB which take from 12V rail ( 9.42 V ls stand by ) generate 3.5V power on which it send to PSU to power 12V (9.4V stand by ) to high as it should

the STB ( power on which generated with MB step down should be send too to power on the BL circuit. ( this didnt happened because its firmware script ) also what you need:

step downer circuit take out the rail 9.4 ( 12V) to generate 3.5 V feed it to STB... this will make 9.4 high to 12... and feed the 3.5V from yours step down circiut via 1K R to other BSW BRI to turn on the BL.... BRI could it need more R value how much it less the brightness on LED BL will be high... and the reserve wise.

of course this need to do with out main boards in circuit.

Thank you for the explanation! Very helpful. Now I understand why 12v turns into 9.4 without main.

One more question, how does main tell PSU to reset? I can hear faint PSU cap/transformer area squeal when main boot loops without firmware.

[Diah]

Re: HiSense 65H6510G no boot

no reset... mean there are over load on PSU. it could be BL .. it could be the 12V rail when it go high... or the MB step down circuit over load... there are must be reason.... simple answer move load... let the panel always disconnected from T-CON. wifi BT out from circuit... speaker out... to narrow the cause

EDIT: when you run the PSU as stand alone as i described before and there quit and BL on then you know where yours eyes should focused.