ZyXEL GS1900-10HP revision B1 support OpenWrt firmware?

I would like to buy ZyXEL GS1900-10HP revision B1, but I didn't know that this firmware is okay
https://downloads.openwrt.org/releases/21.02.0/targets/realtek/generic/openwrt-21.02.0-realtek-generic-zyxel_gs1900-10hp-initramfs-kernel.bin
Check also https://openwrt.org/toh/zyxel/gs1900-10hp

In this datasheet documentation I see two revisions
GS1900-10HP A1 and GS1900-10 B1

and different hardware comparison

So the question is GS1900-10HP A1 and GS1900-10 B1 can possible the install the same OpenWRT firmware or do I need to wait for a new release?

Thanks in advance
Regards, Georgi

1 Like

You should probably check in Support for RTL838x based managed switches

1 Like

Any new information on this? Did you try it?

No, I bought Zyxel GS1900-8HP v2

I bought a GS1900-10HP unit, and it differs from the "A1" as described in your datasheet (barrel plug, and no ON/OFF switch), and labeled on the device as "Revision: B1".

Over serial it calls itself "v2":

U-Boot Version: 2.0.2.1 (May 21 2021 - 12:21:20)

CPU:   500MHz
DRAM:  128 MB
FLASH: 16 MB
Model: ZyXEL_GS1900_10HPv2

I can confirm, that the install instructions on the OpenWrt wiki you posted above work with self-compiled images from current master (r0-a181b9f0). This is after connecting via SSH:

root@gs190010hp:~# uname -a
Linux gs190010hp 5.15.132 #0 Mon Oct 2 20:13:10 2023 mips GNU/Linux

I haven't tested all functions yet, just flashed it minutes ago, but I think your question can be now answered in the positive "Yes". The Wiki-page is a bit outdated though, and the filename in the tftp command needs to be adapted.

Edit: After testing, it turns out, that the POE-part is different in rev B1: Power delivery works, but no control, or measuring the used power. ubus call poe info returns 'unknown' everywhere.

1 Like

Run dmesg after bootup and check kernel log messages. See if there any any major errors or warnings about hardware missing or changed. The best way to be sure is to open the chassis and see what the chip numbers are (compared to A1).

Overview shot of the Rev B1 board:

The POE part (Nuvoton NUC029ZAN, like in Zyxel GS1900-24EP)

The PCB area surrounding the POE part

For the SFP cages, I think

Yes, I know, I should not quit my day job to become a photographer...

I have not compared them yet to the A1 revision board.

Get close-ups of the 4 chips closest to the SFP cages. Can't read shit from the photos you posted except for the RTL chip. Take photos at an incident angle where light bounces off the chip at an obtuse angle, such that chip numbers are... readable :slight_smile: Rotate the board so that text reads across the photo and is in the depth of focus plane that is in focus.

Take your time and review the photos before you close the chassis. Get as close as you can without using digital zoom (only optical) which quickly produces artifacts; some cameras use AI/weird algos to compress better which can mess numbers/lettering up in weird ways. The 'PoE' chip photo was OK, but it's Nuvoton (an ARM micro-controller, for SPI GPIO I2C UART etc, probably for LED control, and Serial, judging by its connection to J1 and J2). All of those caps in the upper left which feed into the GS6014 Ethernet drivers are likely the PoE part.

It's not a race - clear information helps everyone in the long run.

Uploads photos, or crops of photos, but not screenshots of photos :slight_smile:

Of interest are the WinBond memory, and the 16pin square chip (prob NOR) between Winbond and RTL8231 by the cages.

The PoE chip is likely the one under the aluminium 3-fin heat-sink in the upper-left of the overview photo. You may have to prize the heat-sink off to get a clear picture of it. Only do so if you have some heat-sink paste to put it back on. I would advise the same to see if they refreshed the SoC, but that seems unlikely, since this is not trivial in a refresh. And if it boots, detects, and works, it's probably the same.

The Winbond, if you squint enough and let google autocomplete help a bit, seems to be:
Winbond W631GU6NB which is: 1G bits DDR3L SDRAM 1.35/1.5v.

2 Likes

Also, what does dmesg show (or just copy the text between uboot start and the openwrt text logo) after boot?

Yes, I can/will make better pictures, but the sunlight angle right now make my phone nearly blind, lol.
I posted the rough ones above, so that people could tell me, which parts I should post close-ups from. I'll edit the requested pics into this post sometime later today.

1 Like

I've compared the dmesg outputs:

Differences are few:

A1: Probing RTL838X eth device pdev: 82087600, dev: 82087610
B1: Probing RTL838X eth device pdev: 82087c00, dev: 82087c10

Both have these errors further down:
rtl83xx_fib_event: FIB_RULE ADD/DEL for IPv6 not supported
rtl83xx_fib_event_work_do: FIB4 failed

See here

The same SoC. Just the chip parent device and device address in memory.

1 Like

Has anyone got the GS1900-HP10 revision B1 (V2) to work properly so that PoE can be controlled (on and off via ubus -v call poe port '{"enable":false,"port":4}')?

The "new" Realtek POE chip is in use here apparently. I ran into the same issue.

There are 2 hack patches here(github.com/Hurricos) that may work, but I have not tried them yet.

Also: https://github.com/Hurricos/realtek-poe/issues/25

1 Like