I think I just destroyed my Rock Pi 4B+ 1.6?

#1

I think I might have just destroyed my Rock Pi 4B+ v1.6. :frowning:

I have the M.2 extender, the heat sink, and a uSD to eMMC adapter.

This v1.6 Rock Pi 4B+ has a small eMMC module already bundled on it, with Twister OS installed there. It was booting just fine (off the eMMC). On a separate PC, I wrote the
Armbian_21.05.1_Rockpi-4b_buster_current_5.10.35_xfce_desktop.img
image to an NVME SSD and confirmed that a lsblk command on the RockPI showed the NVME. All was good.

The instructions at Rockpi4/install/NVME - Radxa Wiki clearly state that I need to remove the eMMC module to boot off the NVME. I popped the eMMC module off the board, and where I was expecting to see 2 eMMC connectors/socket/rails (like on the v 1.4 board), I instead see some glue and a scattering of little round contacts/solder balls. Where are the eMMC connectors/rails?

Now, even with the NVME in place, it won’t boot – and I can’t get the eMMC module back on tightly. It seems like that included eMMC module was really not meant to come off the board? Power gets to the board - the green light comes on - but then nothing. No blinking blue light. Nothing. Can I recover this? I have no issue with not using an eMMC.

  • Can I boot the 1.6 board without the eMMC module? I think I have lost the eMMC first stage loader, right?
  • What are the exact differences between the v1.4 and v1.6 board?
  • What is the difference between a B and a B+ board?

thanks in advance for any help

dbaman

#2

Can you post a picture? You mean you pull off the chip?

#3

Hello Jack.
Thanks for your reply. Here it is:

20210623_101103
20210623_101103.jpg1512×2016 1.12 MB

#4

It seems (from browsing the incoming problems) as if Radxa should perhaps reconsider their choice to make the eMMC permanent on the new B+, and revert back to a removable and replaceable (and interchangeable) eMMC.

1 Like
#5

Yes, I agree.

And I’ll admit that pulling that attached (soldered on?) eMMC chip from the board was pretty dumb.
However, before today I had never seen, much less removed, an eMMC chip. I am a long-time s/w guy.

I didn’t know these forums existed previously, and had I crawled around in them enough I would have eventually figured out that the chip was permanent. But you have to admit the official doc/wiki is seriously lacking (to say the least). There is no mention anywhere (that I can find) of the differences in the 1.6 4b+ board. Would you not agree (I’m honestly not griping at you, just asking your opinion) that making a chip that was previously removable suddenly non-removable is kind of a big change? Is it really that difficult to document that somewhere in the official wiki, particularly in the parts where it says to remove the eMMC module?

This is great hardware, and I really want to use it for my projects.

Can anyone please tell me if this 4B+ 1.6 board will boot in any shape or form w/o that permanent eMMC chip? Or should I just throw it in the trash?

I saw another post here about putting 2.02 Armbian on a microSD, pressing a MaskRom button, and booting w/o the eMMC. Will that work with this model 4B+? I can live w/o eMMC. The fact that the chip is permanently attached makes me think it has to be there for anything to work… Is the SPI on the eMMC?

Any help greatly appreciated.

#6

Try to boot the system from the micro SD card with pins 23 and 25 shorted.

#7

woow dude that was crazy lol. But… if I were you, I would go to a phone repair store since they have the tools to clean that spot (and see if is not damaged) where it was the eMMC, and if you are lucky, maybe you can buy a eMMC and ask some technician to solder it for you. Ask for radxa team if it is possible to do that with other eMMC.

#8

Show a picture of the bottom side of the eMMC to see if it is too damaged.

#9

I have to say, very impressive. I don’t know how you manage to do that. We have to heat it higher than 250 degree to remove it :slight_smile:

The hw difference can be found here:
https://wiki.radxa.com/Rockpi4/hardware/revision

With no eMMC board now you can still boot from SPI and put the OS in NVMe.

#10

That does hurt to see. But with some cleanup so no shorts are made it can still be used without eMMC.
I think you’re the one from reddit I answered.
If not, then multiple popl are pulling of their soldered eMMC modules.

I still prefer pluggable modules, except for lower tier boards.

#11

I’m not in reddit, it should be the same nickname

#12

:joy::joy::joy:

#13

It replied to the wrong person. @dbaman is probably the one I talked to on reddit.

#14

UPDATE:

I am happy (and relieved) to report that, even after stupidly ripping off the permanent eMMC module on this Rock Pi 4B+ v1.6, it still works. I have the heatsink, M.2 extender, and a booting 1TB nvme SSD (a 1TB XPG) all working happily after I followed the instructions at
Booting ROCK Pi 4(A/B/C) with mainline u-boot in SPI, NVMe and Armbian v20.11.x - Rockchip 3399 - Armbian forum

I think this demonstrates the quality of the hardware on this Rock Pi… Good stuff.

Now to have a look at Armbian…

cheers

1 Like
#15

Hello fellow Raxdaians. :slight_smile:

So I’m the guy who doesn’t know his own strength who popped off the soldered EMMC chip on my Rock Pi 4B+ v1.6. See above.

I reported a few months back that, even after doing this, it still seemed to work (which is impressive).

However, a while after that report, I would get occasional kernel panics (every 2-3 days). Since then, I have created a reboot command in crontab and I just reboot every 24 hours (which of course is not ideal). If I reboot roughly every 24-36 hours, it works fine. Yes, I do have the big black heat sink installed under it.

I’m not familiar at all with diagnosing kernel panics, but here’s what I’ve seen:
Message from syslogd@localhost at Sep 8 15:17:46 …
kernel:[131008.652050] Internal error: Oops: 96000004 [#1] PREEMPT SMP
Message from syslogd@localhost at Sep 8 15:17:46 …
kernel:[131008.853491] Code: 91002273 eb15027f 54000420 f9400274 (f9400681)

I’ve got a few spare cycles and am looking at this again.

Right now, the only way I can get it to boot is to place a jumper on pins 23/25 AND have a MicroSD card (I have a 128GB variety) with Focal OS inserted – and it will then boot the M.2 NVME (with Focal) OS (I have the M.2 adapter)

When it boots, lsblk reports:
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
mmcblk1 179:0 0 119.3G 0 disk
├─mmcblk1p1 179:1 0 3.9M 0 part
├─mmcblk1p2 179:2 0 4M 0 part
├─mmcblk1p3 179:3 0 4M 0 part
├─mmcblk1p4 179:4 0 512M 0 part
└─mmcblk1p5 179:5 0 118.8G 0 part
zram0 252:0 0 50M 0 disk /var/log
zram1 252:1 0 1.9G 0 disk [SWAP]
nvme0n1 259:0 0 465.8G 0 disk
└─nvme0n1p1 259:1 0 465.8G 0 part /

With the M.2 NVME attached:
-If the SD card isn’t there with an OS, it won’t boot.
-If 23 and 25 are not shorted, it won’t boot.

I’d like to do 2 things:
-remove the req for there to be SD card inserted, just boot from the NVME (and no EMMC of course)
-get rid of the kernel panics

I’m a bit confused with SP1 and boot process.
My theory is that eventually something tries to access that non-existent EMMC and it causes the kernel panic?

Can I disable the EMMC?
How can I otherwise cause it to boot from NVME w/o the SD?

TIA
dbaman