Need a working image for Quartz64B
#1
Just got a model B and tried to do some verification. I have tried the following 2 images, but none of them is working correctly:

Manjaro-ARM-minimal-quartz64-b-20220808.img.xz: Serial output is entirely corrupted. It seems the system is booted into the OEM setup mode, and whenever I press the enter key it will immediately return some junk back. I assume it was asking for some question and won't take an empty response as answer, so it was asking again.

Armbian_22.08.0-trunk.0114_Quartz64a_jammy_edge_5.19.0.img.xz: No image for Model B from Armbian, so this A image has to do. The serial output is less corrupted with some recognizable words but far from a clean output. It is in a boot loop so not usable.

The serial cable and settings works correctly on a Radxa ROCK 4 so this has to be the board's issue right? (Right?) I tried to replace the DC power supply with another one with no avail since I think there might be some noise from the ground plane. They are 2 prong design though.
  Reply
#2
I just want to share that I have the same issue, but forgot to open a question about it (got distracted with more fun stuff).

I guess it should be a software or configuration issue, because my board did boot successfully before with an image from last year.

This month I tested with:
1. Manjaro-ARM-minimal-quartz64-a-20220801.img
2. Manjaro-ARM-xfce-quartz64-a-20220808.img
3. Manjaro-ARM-minimal-quartz64-bsp-20220516.img ( a bit older and seems it's from an archived repository, so not worked on anymore)
4. Armbian_22.08.0-trunk.0109_Quartz64a_jammy_edge_5.19.0.img
5. slarm64-current-aarch64-server-quartz64-5.19.0-build-20220807.img

All input from the serial console is garbled with some bits and bytes looking like real words (for example a string looking like a filename with extension "conf").
  Reply
#3
make sure you connect the ground pin and set the baud rate to 1,500,000. The serial adapter you use also matters, as does keeping the serial wires short and away from any power lines, apparently the model B is less nice about logic signals or something. A CH340G based adapter like the PINE64 "Woodpecker" will work.

Occasional Linux Kernel Contributor, Avid Wiki Updater, Ask Me About Quartz64
Open Hardware Quartz64 Model A TOSLink Adapter
Pi-bus GPIO Extender For ROCKPro64 And Quartz64 Model A
Plebian GNU/Linux
  Reply
#4
1) For ”uart” troubles, it is recommended to check the circuit diagram.

2) PINE64 "Woodpecker".
The original 3.3V output signal was measured to be about 3.6V. This is beyond the acceptable range.

---
As for PINE64 "Woodpecker", I purchased two of them about a year ago and have them on hand.
This is obviously strange.
In my experience, when two pieces have exactly the same symptoms, the whole lot is affected.
I'm not sure if the ones being sold now also have the same problem.

As far as the "Quartz64B" is concerned, uart's trouble is probably caused by the former (see schematic: improper level shifter design).

The latter is not desirable, but in the case of "Quartz64B" there will be no significant impact
However, in the case of Rcokpro64, it is dangerous because it exceeds the input maximum range.

---
"Quartz64B", unfortunately due to delivery troubles, I still haven't received it, so I can't check it in person...

What the hell is going on ? at the recent "PINE64 Store".
  Reply
#5
If anyone of you has experience with electronic work, please try the following measures.

CON1 (40pin Header)
UART2_TX(8pin) <- "R-pu" -> 3.3V1(1pin)

"R-pu" is about 4KΩ as a guideline, however the smaller the value, the better the effect is expected.

--
As for this kind of issues, it is rule to make a proposal after confirming it myself.

However, due to repeated delivery troubles, the item never reaches me.
Unfortunately, I am unable to verify this myself at this time.
  Reply
#6
Although I'm a complete noob in electronics I did try to add a few resistors about a few hundred ohms (to both RX and TX), but nothing changed significantly.

But after I edited the configuration (the command line for the kernel) on the manjaro image and set the speed of the serial console to 9600bps I saw two lines[1] on the console: both mentioning something like "early hooks (udev)" and "early hooks (plymouth)".

I'll leave it for a night (again), but the board probably won't boot. I won't speculate what could possibly be wrong.

Well, that's as far as I can test. Smile

[1] Some characters before those lines couldn't be printed, but I'm not worried about those.

Note: i'm using the woodpecker adapter and model A of the quartz64 board.
  Reply
#7
Are you sure your serial interface is working well? Is RADXA using 1500 000 baud speed too? If the serial works at slower speed this is not indicative of it being capable of working at 1500 000 baud (unless it uses CH340G chip which is known to work).

Additionally, just in case I'll mention the CH3640g chip can run with 5V IO or 3.3V IO. Some adapters have a jumper to set this. Others can be easily modified if they are set for the other voltage.

Do you perhaps have an oscilloscope you can connect it to and look at the signal? I don't have Quartz64-B hardware. Just a soquartz and Quartz64-A, but even when I tried images that did't work because of a kernel panic etc I still get good output on the serial port. I could see u-boot, the image trying to boot etc.

I doubt very much that there is some hardware problem on Quartz64-B board that requires additional resistors etc to make it work. If you measure voltage that is too high on your serial adapter my first guess would be that it is configured for 5V.

Here you have the link to the datasheet for then chip in case you need to modify the adapter. https://www.insidegadgets.com/wp-content...asheet.pdf Look at page 5 (application notes).
  Reply
#8
I got some time to play around. Here's a few things I tried:

1. For the woodpecker: I managed to get somewhat stable measure with a basic multimeter and frequently I got 3.7V (and almost as frequently my hands managed to short the pins and had to reinsert the device). I guess this is not so interesting anymore.

2. I tried another no-name usb-to-serial adapter, but didn't get good reading. Also the multimeter measure around 3.5V.

3. Lastly, I tried with RPi4 model B.
Confident that the guys writing the specs didn't get it wrong (i.e. pins voltage is 3.3V) I connected the uart pins from the RPi4 to the Quartz64-A at baud rate 1500000, but still got mashed up characters with Manjaro-ARM-minimal-quartz64-a-20220801.img and only 3 (three) readable lines with baud rate of 9600 and then the booting was stuck .

Then I tried with Armbian_22.08.0-trunk.0109_Quartz64a_jammy_edge_5.19.0.img, but nothing good was read from the serial port with speeds 1500000, 19200 and 9600.

Well, given that I had a RPi, today I bought a usb memory stick to install the Raspberry Pi OS on it and use the micro sdcard slot to look around. I managed to boot into both images (using the command systemd-nspawn -D <root-directory> --boot) and tried to finish the installation process and setup static IPs (i'm only confident about the armbian configuration). After booting both images on the Quartz64-A, nothing seemed to work and there was no ping.

I even tried on the armbian image to execute custom commands in /etc/rc.local and although the commands did write something on the micro sdcard during booting with systemd-nspawn, nothing was written when I tried booting from Quartz64-A.

Given that the Armbian gave me a "scary" message that the installation process is only meant for developers, I guess I should wait a bit more until I can try again. Smile
  Reply
#9
(08-11-2022, 09:48 PM)Excalibur Wrote: Just got a model B and tried to do some verification. I have tried the following 2 images, but none of them is working correctly:

Manjaro-ARM-minimal-quartz64-b-20220808.img.xz: Serial output is entirely corrupted. It seems the system is booted into the OEM setup mode, and whenever I press the enter key it will immediately return some junk back. I assume it was asking for some question and won't take an empty response as answer, so it was asking again.

Armbian_22.08.0-trunk.0114_Quartz64a_jammy_edge_5.19.0.img.xz: No image for Model B from Armbian, so this A image has to do. The serial output is less corrupted with some recognizable words but far from a clean output. It is in a boot loop so not usable.

The serial cable and settings works correctly on a Radxa ROCK 4 so this has to be the board's issue right? (Right?) I tried to replace the DC power supply with another one with no avail since I think there might be some noise from the ground plane. They are 2 prong design though.

Latest manjaro XFCE image works (with several monitors 1080P).
  Reply
#10
(08-11-2022, 09:48 PM)Excalibur Wrote: Just got a model B and tried to do some verification. I have tried the following 2 images, but none of them is working correctly:

...

The serial cable and settings works correctly on a Radxa ROCK 4 so this has to be the board's issue right? (Right?) I tried to replace the DC power supply with another one with no avail since I think there might be some noise from the ground plane. They are 2 prong design though.

I tried (an older version of) https://forum.pine64.org/showthread.php?tid=17779 and that worked fine, both on Model A as Model B (they're separate images!).

I did have problem with the serial console (woodpecker) on the Model B, but it worked fine on Model A.
Need to look into it further at some point.
But despite there being no serial console output, I did notice the NIC light going on and also on my switch and after some time I was able to SSH into it.
  Reply


Possibly Related Threads…
Thread Author Replies Views Last Post
  Quartz64B eMMC boot failure bsmorgan 2 1,164 11-15-2023, 12:00 AM
Last Post: balbes150
  can't get quartz64-A working jyoung 14 7,260 04-29-2023, 04:41 PM
Last Post: diederik
Tongue Trying to get Quartz model A working fasani 2 2,879 09-17-2022, 06:43 AM
Last Post: balbes150

Forum Jump:


Users browsing this thread: 7 Guest(s)