PX5 stuck in bootloop

[marchnz]

This Post details how OTG/LOADER can be achieved to recover a corrupt PX5 or PX3 - any RockChip MTCx SOM

Update 2/2018 - anyone can make a gig without modifying their main board. BUT FIRST try a USB port, some have OTG out of the box.

From the work I've now done, Ive realised you don't need to hack your mainboard. You could temporarily solder connections to SOM or make JIG from header pins or make a simple cable from an old USB cable and two resistors: see: https://forum.xda-developers.com/an...px3-5-headunit-mod-recover-flash-som-t3766892

Making a jig is necessary , as MCU cuts power to SOM after a few minutes during OTG.( Theory is the MCU watchdog senses an unknown condition in SOM, so resets by power off.) This can be bypassed by another mod to board, but much easier to just make a simple jig.

Need:
Remove SOM from mainboard.
Header pins to fit SOM (or solder temporarily to SOM PCB header)
USB male cable, cut one end off for wires to attach to SOM.
5v and 3.3v power supply.
1. Using schematic as guide. Connect earth and +5v pins. Connect 3.3v to [rtc_3.3]
2. Connect USB wires to SOM [USB-OTG] connections
3. Connect wire to SOM [ctl_recovery] and wire that to 3.3v.

Colour codes & Notes:
GREEN = PSU Earth [0v/(-)]
RED = +5v to PSU
BLUE = USB
YELLOW = +3.3 Via 3.3v supply OR use two resistors as a voltage divider 10K to 5v (R1 in image), 15k to earth (R2 in image), center = 3.3v
NOTE: SINK [CTL_WAKEUP) to GROUND (0v) if unit does not power ON

+5v = 1,2,3
0v = 4,5,6,71,72,73 and SOM Mounts
+3.3 = 20,28 (and possibly 9 & 15 - see note re. 15.)

With USB connected, power on. Use tools/load driver as described in recovery thread.

Q/Why hardware mods?
A/Mods were necessary because out-of-the-box (default) both my GS and JY boards would not enter USB OTG/Loader mode to connect it to a PC for recovery. We now know that a "jig" can be made without sacrificing your mainboard and simply wiring to header pins or buying a SOM header from say Digi-Key.

- Your MTCD may be different -so test first!

Q/My SOM won't appear in device manager or Rockchip Android tool
A/ If all connections are correct and power is Ok, force MASKROM as detailed here:
https://forum.xda-developers.com/sho...&postcount=128
https://forum.xda-developers.com/sho...&postcount=129


***Updated 08/Dec/2017 - Attach image showing USB Male cable connections to mainboard***

** Note - SD recovery boot card did not work for me on PX5 - probably due to version of [rockchip sdcard image create tool] - if anyone has a version that works, please post here.


Updated Feb 2018 - Android 6 SDCARD create tool is now available. Recommend SDCARD method and then OTG if SDCard method fails

Synopsis
My corrupt NAND was successfully reflashed using OTG "LOADER" mode as follows. It became corrupt , stuck in a bootloop and unable to enter recover following flashing a custom (bad) firmware image file.

- Using an MTCD GS board - which I have as a spare part - no display, only board.
+ Windows 10 x86 notebook with rockchip batchtools v1.8, rkimagerepack tool, latest PX5 image.
+ USB stick formatted fat32 with latest PX5 image

Do not do the below mainboard mods, this is completely unnecessary and posted for background on the journey to find a simple way of getting OTG.
1. Modified the board as follows:
A. Desolder and lift pins 27 & 28 of USB hub GL850 (thanks to xda user @ikerg and 4pda - for the lead)
B. Solder Male USB a cable to GS board USB 1 connections (See below attached image showing where USB male plug wires were soldered)
C. Remove pin to RK PX5 SOM [ctl_recovery]
D. Solder a wire directly on PX5 SOM header on [ctl_recovery]. This wire is used to trigger recovery.

Process
Initial recovery image: Because the PX5 firmware (OTG) "LOADER" process timeout appears to be approx 1 minute, it was necessary to create a recovery only image, as the 1.2gb PX5 firmware file exceeds the timeout, aborting the flash.
Using RKIMAGEREPACK, I created a image with a dummy zero byte system.img. This resulted in a 200mb image file.

- Loaded [rockchip batch tools] on notebook.
- Installed rockchip drivers
- connected USB cable to notebook
- held [ctl_recovery] wire to [mcu 3.3v test pad]
- Powered headunit on
- Notebook 'USB chime' was heard and board became connected, as indicated as highlighted '1' in rkbatchtools
- wiped flash
- selected firmware image without system created above and flashed to PX5
- powered off, removed PX5 SOM and refitted to JY headunit board
- On JY headunit, powered on, entered recovery via [holding powerbutton for approx 15sec - wait for 3 flashes of led] press once more.

Successful boot into recovery - started booting recovery and warned that system image was not valid. PX5 self-formatted system and awaited image.
- inserted USB stick with latest PX5 image and upgraded.

First boot waited for 5 minutes. Pressed reset and rebooted - successful recovery!

The warning here is that the PX3 SDCARD image create tool does not work for PX5 - even when using a PX5 image.


Why did I post this? Original Post:
I have a (JY) MTCD with HA 2..56 MCU PX5 stuck in a boot loop - "starting apps" appeared and then it reboots, looping. I was unable to access recovery via any of the standard recovery processes (e.g.holding the power button or boot into recovery via bootable SDCARD.
========================================
On SDCard boot - works OK on PX3 but on PX5, only a black screen is displayed without any text.
SDCard and button holding recovery methods worked OK on PX3 board, but not the PX5 Module when the SOM NAND was corrupt.

 

[marchnz]

Ok - need to figure out how to force this PX3 board to 'maskrom' - this is the procedure where the NAND is suppressed by grounding the I/O or Oscillator to effectively stop the NAND. This then places the RockChip PX5 into Maskrom mode to then connect/reflash via USB.

Anyone tried this?

Feb 2018 - MASKROM methods have been identified for both PX3 and PX5 Modules. See posts later in thread.

 

[nixfu]

You seem to think you have both a PX3 and a PX5 board....it is one or the other.

 

[marchnz]

You seem to think you have both a PX3 and a PX5 board....it is one or the other.

Thnaks for your reply - clarifying [I have a (JY) MTCD with HA 2..56 MCU PX5 stuck in a boot loop]

I have a PX3 Board that I have swapped back in to test - it is OK. The PX5 goes black screen if trying to boot from SDCARD recovery and I cannot seem to force it into [maskrom] to flash from PC either.

Would be interested to hear if other PX5 users who have swapped from PX3 to PX5 can access recovery OK.

Note that I foolishly screwed up and flashed a custom image thayt contained what I think may be the wrong achitecture of SUPERSU, hence the boot loop. BUT this doesnt explain why I can no-longer enter recovery, or does it?

Regards,

 

[ptclabs]

try holding the reset button for about a minute. it should auto restart into recovery. it will first go off, then it will either pop back on suddenly or start turning itself on and off. if it does the on and off thing i usually let it boot to recovery on the third one

 

[marchnz]

Thanks for the info.

Ive done everything but holding the power/vol knob or reset for 1 minute.

Tried forcing the unit into mask rom by very carefully exposing some of the obviously IO tracks between the cpu and flash and shorting a pair (this is the method described elsewhere, but usually there are copper test pads (you'll see these on the px3 module but not px5))

No matter what I try, I cant get my pc to see the head unit via either usb sockets. I made a 1 to 1 usb A male cable which works with a nexbox OK but not these headunits. The HU is a JY UL124 orginally px 3. It has 2 sdcard slots and two usb.

 

[marchnz]

After significant research and testing, I have a recovered PX5! See Post#1

 

[ikerg]

Good.....clap clap clap
Your tutorial is very clear and useful for other people....No more bricked devices....

 

[marchnz]

Good.....clap clap clap
Your tutorial is very clear and useful for other people....No more bricked devices....

Thanks - updated op with a couple of images of my test/recovery jig and software.

 

[audisean]

After significant research and testing, I have a recovered PX5! See Post#1

good work there - would like to check my px3 board so did you ground to GND or D0 for reset on px3 pads ? and if you could show which wires on usb lead you connected to pin 27 & 28 on GL850 chip that would be very useful thanks

 

[marchnz]

good work there - would like to check my px3 board so did you ground to GND or D0 for reset on px3 pads ? and if you could show which wires on usb lead you connected to pin 27 & 28 on GL850 chip that would be very useful thanks

Great question and good spotting - on px3 its possible to force "maskrom" by shorting [gnd] & [d0] pads. Unfortunately px5 do not have these pads and can only be achieved on px5 via carefully exposing track insulation to NAND - fortunately px5 will enter "loader" mode on recovery. So, no I did not do this, I used a GS board I have as a spare part and modified it to force recovery, by removing the SOM header "recovery" pin and grounding that pin via a switch - to force the SOM into recovery.


Note that OTG USB connections are made to the USB switch IC (if fitted) U27 [fsusb42]. If this IC/USB switch is not fitted, more drastic mods would be needed to disconnect the usb connections to the usb hub IC (by disconnecting the SOM header pins.) If you have a GS, you have U27. If not you will need to check.

Happy to answer questions here, we can all benefit through sharing.

 

[audisean]

Great question and good spotting - on px3 its possible to force "maskrom" by shorting [gnd] & [d0] pads. Unfortunately px5 do not have these pads and can only be achieved on px5 via carefully exposing track insulation to NAND - fortunately px5 will enter "loader" mode on recovery. So, no I did not do this, I used a GS board I have as a spare part and modified it to force recovery, by removing the SOM header "recovery" pin and grounding that pin via a switch - to force the SOM into recovery.


Note that OTG USB connections are made to the USB switch IC (if fitted) U27 [fsusb42]. If this IC/USB switch is not fitted, more drastic mods would be needed to disconnect the usb connections to the usb hub IC (by disconnecting the SOM header pins.) If you have a GS, you have U27. If not you will need to check.

Happy to answer questions here, we can all benefit through sharing.

Thanks for reply - my px3 board appears dead in unit thus far - think i've worked out how to solder usb A cable now just didn't want to cause any damage mixing tx/rx wires - i'll have to retest but i think I had 3.3 volts at GND & D0 pads - no time today but will have a go tomorrow (hoping this board is where Nand or fried chip fault lies) I have ordered a PX5 board so hopefully will get unit working

 

[audisean]

Had a try with px3 board - lifted pins 27 and 28 - put software and drivers on laptop - soldered usb lead to d+ and d- on px3 board and ground traced from u27 and ctl_recovery wire - nothing happens still - power at u27 is 3.3 volts

 

[marchnz]

Had a try with px3 board - lifted pins 27 and 28 - put software and drivers on laptop - soldered usb lead to d+ and d- on px3 board and ground traced from u27 and ctl_recovery wire - nothing happens still - power at u27 is 3.3 volts

Good work! How have you isolated the header pins - if they are still all connected (ctl_recovery, d+,d-) to respective ICs, it wont work because of contention with the connected ICs. E.g. ctl_recovery is connected to MCU and is pulled up to +3.3v by MCU - my solution was to remove respective header pins by careful desolder - they can then be refitted.

 

[audisean]

Knew i'd missed something - header pins still connected so will try to isolate guessing I need only connection at SOM board only..

 

[audisean]

pins isolated but no luck yet.... think maybe my chip is faulty (others report some px3 chips come from faulty batch) so will wait px5 board delivery - if that works I might order a replacement px3 chip and try replacement

 

[Cid6.7]

A bit over my head
Any chance I could get a hand ?
Maybe over Hangouts or something ?
Please Please Pretty Please lol

Here's what I'm workin with https://goo.gl/photos/uNTpFnUYJxZ3VwSJ7

 

[audisean]

Got the new PX5 board (12 days from China ) put into head unit and worked straight away - posting here as I had major trouble re-soldering pins 27 & 28 when I pulled a leg off GL850G usb chip - see attached pic of wire I had to solder in place... thinking if I did this again I would probably just make a break in circuit tracks and use conductive glue or nickel pen to remake circuit - PX3 board is still dead - never got usb to pc connection - as for Cid6.7 board - looks like you found the usb hub chip and switch chip (different boards to mine though) pin 27 & 28 are opposite dot on chip - usb switch chip should go +5v, sel, d+ , d-, and ground

 

[Cid6.7]

Is all this done just for the pc to detect the unit..?
If so what if the pc already detects it BUT unable to write to it with 0 modifications done to it..?
https://goo.gl/photos/m7NGyrkD4ef2YydD7

audisean what did you pay for the PX5 board if I may ask..? I'm wondering if I am able to swap chips from Sofia to PX5 with no mods

 

[audisean]

Is all this done just for the pc to detect the unit..?
If so what if the pc already detects it BUT unable to write to it with 0 modifications done to it..?
https://goo.gl/photos/m7NGyrkD4ef2YydD7

audisean what did you pay for the PX5 board if I may ask..? I'm wondering if I am able to swap chips from Sofia to PX5 with no mods

I paid about £95 for the board with airmail dispatch - not sure if it would work on yours - fitment looks similar with pin header - Witson 32814450733 is what I bought - worked straight away and recognised settings from previous android system in unit. The PX3 mods were necessary for otg cable to link pc and flash - if you have recovery by SD card or front usb you won't need it. Sofia looks pretty different but if you can link pc did you try rockchip batch tool ?