802.11be QCA cards eg WLE7002E25, anyone?

Hello,

I've been waiting for an MTK option for a single miniPCIe slot to step up from my MT7915, there still are none. What about QCA-based cards, following reference design from my understanding?

I'd prefer a 5G6G or 6G, but can only grab a 2.4G+5G DBDC option on the cheap. Are there any obvious red flags like poor/no driver support in OpenWRT, or firmware schenanigans like with previous generation of QCA cards? Anyone using one of these in the wild?

It iis out for some while now.

The other thing you wish for is not supported in mainline.
https://wireless.docs.kernel.org/en/latest/en/users/drivers/ath12k.html

1 Like

@brada4 thank you!

The link you provided is for a 802.11ax card, whereas I need/want 802.11be.

AFAIR ath12k is available in snapshots, and other cards (client ones) using it are working fine. No word on AP ones like the one I am interested though...

Only industrial kind of that card is supported.
MBO is not really instrumented yet.

1 Like

@brada4 thank you!

I'll wait for MTK 802.11be AP and/or improvement of QCA 802.11be support, then!

Be aware that you're pretty much among the first to experiment with 802.11be cards, especially addon cards and ath12k. That means there is not much general wisdom present, yet - and ath10k/ ath11k/ ath12k are rather special in many ways (and, while needed to get reasonably recent drivers, the backports approach doesn't make that any easier (mailbox/ rpc API between kernel and backports)). Another hot topic for both mt76 and ath12k (but ath12k in particular) is finding a card from a reputable vendor, who programmed the EFUSEs correctly! You are in a world of pain if those aren't correct, which is often the case for these low-volume/ almost prototype cards.

Reconsider if you really want to go there, yet. And count on failure and potentially having to go through multiple cards/ sellers to succeed.

3 Likes

@slh thank you!

Possible vendor misconfiguration is the main reason I steered cleared of QCA AP cards so far. The only reason I consider them now is a very good price offered by a couple vendors. I'd rather wait for MTK with adequate driver support and no (major) headaches, then.

Btw., a 5+6 GHz DBDC card is not that likely to find. Mostly for two reasons:

  • the bands are too close to each other, so you have to fight more with interference from the other band of your same card.
  • in most cases, 5- and 6- GHz is more or less treated as different portions of (basically) the same band - and with that, the same/ similar phy and band filters. As a result it's easy to find 2.4+5 GHz or 2.4+6 GHz DBDC cards, but the combination of 5+6 GHz is technically much more difficult - I guess (in the future, quite a few months++[*]) it's more likely to find 2.4+5+6 GHz 'TBTC' cards, than 5+6 GHz DBDC ones.

--
[*] when considering the potential time scale, you have to remember that DBDC/ 'TBDC' is usually a way to cheap down boards, but 6 GHz is still considered a premium feature - which is not served well, if you cut your 4x4 card into 3 virtual 1x1+1x1+2x2 cards and the corresponding reductions in MIMO/ beam-forming and throughput reductions. 6 GHz effectively still means high-end, and high-end kind of implies 4x4 (or better) for each band, all bands.

1 Like

@slh thank you!

I mentioned 5G6G because Compex makes one.

My goal is to get as close as possible to saturating 2.5GBps link for LAN to WLAN traffic for a 2x2 802.11be client, Anything that can do 2x2 320Mhz 6G 802.11be is fine, be it 2x2 5G6G DBDC, 4x4 6G or something else - but I wouldn't mind getting a 2x2 160Mhz 802.11be at 5G on the cheap as an intermediary step, if that's my only (reasonably priced) option for now.

I considered getting BPI-R4 with native 802.11be TBTC card - but it's physically too big and somewhat expensive, packed with features I do not need. Would prefer a physically smaller and cheaper solution.

2882Mbps does not saturate 2500in 2500out both at once. Means cannot achieve target.