Quantum Fiber W1700k support

For people having to 10G issue, that is addressed by two patches. You can first try the new 10G driver PR that fixed the issue the Realtek phy register got incorrectly toggled during boot up and caused phy crashed.

Another PR if you still having issue you can try is my 10G pcs patch, that fixes the issue when 10G switch speed negotiation fail to get IP.

So after testing 6.18 for up 6 days, 14:10 HW offload is still working. Sweet :+1:

@frollic Did you close this thread?

If it was me, it was unintensional, fat fingers, perhaps :slight_smile:
Gone now.

I've been using fanboy's test build which appears to include those. Didn't seem to do anything for the WAN port on the "problem" unit.

Works great on the other 2 units though.

By testing URLs, I found a new vendor firmware : http://gcs-firmware-details-prod-002.storage.googleapis.com/W1700K/WXK001-05.00.31.10.bin (2026-04-21).

It has a new u-boot : U-Boot 2014.04-rc1 (Apr 21 2026 - 07:22:08) AXON 2.5.
From the strings, it adds at least support of a new NAND (_SPI_NAND_DEVICE_ID_W25N04L) : this would be quite late, unless it's a rebuild of something already included in newer devices, but perhaps this is just from a newer Airoha SDK.

WXK001-05.00.30.62 has U-Boot 2014.04-rc1 (Apr 01 2024 - 06:06:47) AXON 1.8.
WXK001-05.00.30.64 / WXK001-05.00.30.76 / WXK001-05.00.30.82 / WXK001-05.00.31.06.bin have U-Boot 2014.04-rc1 (Jun 12 2024 - 08:14:34) AXON 2.0 (so it seems the two newer AXON 2.0 builds were never included in vendor firmwares, despite being flashed from the factory).

Also take note of WXK002-05.03.41.02.bin, uploaded on the same day but is about 10M larger than the other images.

Ships with U-Boot 2014.04-rc1 (Dec 03 2025 - 14:50:19) AXON 2.3and the squashfs date seems to suggest the same.

Looks like that 10mb is a DPI engine and (password protected) signatures, eww.

It might be a debug firmware : it has a few debug scripts, a diag.ko.

lionic-dpi seems to contain a "Cloud AV", a server able to decompress many archive files (on this slow CPU...), a program to identify devices on the LAN, ...
Several binaries were not stripped, and contain symbols and sometimes debuginfo : decomp_server, cwga, casa, dm_agent, lc-json, lcsh, wcfa.

It also has minissdpd and miniupnpd.

There is a /etc/init.d/uenv_update, which changes u-boot variables.
But this is weird, the changes are not made in the same versions which come with a new bootloader (and an older firmware would keep the newer bootloader but set its values :thinking:).

This also means the u-boot config could be different, depending on whether the vendor image has ever been booted (and which one).
But I'm not sure about the defaults, since there are both values in the compressed u-boot, and at 0x80000.

For example, AXON 1.8 / AXON 2.0 / AXON 2.3 / AXON 2.5 have serdes_ethernet=11 (LAN2 port in USXGMII mode) in the compressed u-boot, but serdes_ethernet=12 (LAN2 port in HSGMII mode) at 0x80000.
Since recent firmwares force 11, I suppose not all devices had this value.
This might matter for LAN2 port issues on boot.

@Notvis Before flashing any bootloader, you should perhaps compare printenv between your working and non-working devices.

WXK001-05.00.30.62 / WXK001-05.00.30.64 / WXK001-05.00.30.76 / WXK001-05.00.30.82 :

qdma_init=61bb
dram_limit=
serdes_wifi2=00

WXK001-05.00.31.06 / WXK001-05.00.31.10 :

qdma_init=61bb
dram_limit=
serdes_wifi2=00
serdes_wifi1=00
serdes_ethernet=11
serdes_pon=23
serdes_usb1=00
serdes_usb2=01

WXK002-05.03.41.02 :

qdma_init=61bb
dram_limit=
serdes_wifi2=000
serdes_wifi1=000
serdes_ethernet=411
serdes_pon=423
serdes_usb1=000
serdes_usb2=001

So values are changed for AXON 2.3, but not AXON 2.5.

Hi, I ordered W1700K from Amazon US (seller: Mile High Sales LLC) and the package has just arrived but something unexpected happened...

It says "LUMEN" instead of "Quantum Fiber" and the HW version is 2.1... Am I ruined? :frowning:

EDIT: If you look at this closely, it says “MXF-W1700K2”..

That's the W1700K2, which is a slight hardware revision.

I note that the FCC documents show a slightly different heatsink + fan vendors, and that it comes with RTL8261CE instead of RTL8261N. (I think that's just a different package style with pins as opposed to none, since you still have 10G labels for the ports.)

It's the same amount of capable. If anything, it might come with slightly newer AN7581GT stock with less cursed PCIe training.

That said, I have no idea if the layout/firmware is the same and whether it's compatible with current images. (Although I suspect it very much could be)

Gosh, there's more hardware revision!

https://ir.lumen.com/news/news-details/2024/Lumen-is-first-in-the-industry-to-achieve-a-Wi-Fi-CERTIFIED-7-device/default.aspx

Both devices also support cloud-based Digital Twin technology and use the carrier grade OpenWRT standard with [Wi-Fi EasyMesh™] and [prplMesh] enabling real time orchestration and advanced device management capabilities.

I like that they mentioned OpenWrt. :slight_smile:

But by reading this I assume Lumen is the maker and Quantum is just a whitelabel?

Power Logic and Fuzhineng fans to be specific. Also the power bricks are only specified to output 48 W (4 A) instead of 60 W (5 A). Different Nuvoton fan controller...

Lumen was the corporate parent, and Quantum was Lumen's more recent brand name for their residential fiber business. But Lumen recently sold off Quantum to AT&T.

The vendor names don't even match what's in the vendor (they're in the WXK002) image since they refer to them as asink and asrock for the fan curves? (Which are identical)

The fan controller may be problematic depending on how much it differs from NCT7511Y, given there's no datasheet available.

You don't need to test URLs.

In the xml from here:
http://gcs-firmware-details-prod-002.storage.googleapis.com/
you can find these firmware Links:
W1700K/WXK001-05.00.31.10.bin
W1700K/WXK002-05.03.41.02.bin
W1701K/W1K001-05.00.31.10.bin
W1701K/W1K002-05.03.41.02.bin

The WXK002 could be the Firmware for the W1700K2.
I don't have time to analyze it, but I'm sure some of you would like to take a look at it.

Indications appears that is the case, given I see checks all over the place for whether /etc/hwver == 20, and the fan curves being present.

I think their antenna gain might be ever so slightly different as well given they also use a different power table to abide by 6ghz Indoor/Standard Power limits as well.

# As original with W1700Kv1 HW version
#       SkuTableIdx is 3 --> Sku_04/Backoff_04   : Used for 6ID
#       Sku_21/Backoff_21 is used for 6SD
#
# In W1700K-v2 is new HW version, which uses different SKU and BF Backoff power tables
#       SkuTableIdx is changed to 4 -->  Sku_05/Backoff_05   : Used for 6ID
#       Sku_22/Backoff_22 is used for 6SD

Just the top photo of the internal.

I immediately noticed that there is no RF shielding sticker on the left bottom side. Also, the power choke is gone now.