I have an Intel BE200 Wi-Fi card in my laptop in the room next to the room where the router is.
I just found that opening the laptop lid, I get a real improvement in the speed (tested with iperf3). This makes sense since the antennas for a laptop usually are in the lid behind the display.
Strangely, the speed for my smartphone, which is 6E, not 7, I get a real poor upload speed when using the 6GHz band (also tested with iperf3).
To be honest - those devices don’t help with their proprietary physical interfaces: the physical distance between two (in principle standardized) M.2 connectors is itself proprietary, so it won’t fit into any board unless they built it themselves (or, in other words: you buy a board which is made for exactly one card and thus it is no better than the boards with on-board WiFi cards).
It is just the same with the compex WIFI 7-cards: They do have MLO-connectors (which they officially promote: https://compex.com.sg/compex-launches-wi-fi-7-modules-with-smart-connectors-supporting-seamless-multi-link-operation-mlo-integration-on-standard-cpu-platforms/), but those are only marketing and not available for the customers (I talked to a sales-engineer to understand how this could work with vanilla OpenWrt). The real MLO happens via some proprietary mPCIe/M.2-signal, and for this you need the Compex boards. And those boards run some proprietary OpenWrt-version with some special drivers, which still seem to be under Qualcomm-NDA, which is why a couple of configuration options need to be done by the Compex team. And they would be eager to do this for me, as the engineer told me... 
Sorry vendors, either you offer something that can be installed in an already OpenWrt supported device with some official driver software (in my eyes the firmware blob itself can stay proprietary, but the driver must be FOSS), or we won’t buy your crap. I am fed up with this situation…. 
what about COmpex cards which are MLO capable and are dual-band. there is no issue with creating MLD na use MLO ?
i assume that you are referring how to establish MLO with two physical wifi 7 cards?
Yes, my idea (after reading this announcement) was to combine 3 cards with each 4×4 to cover all 3 bands.
How to setup the SSIDs for a DBDC-card is something I can ask the engineer for sure. But this does only partially help you, as you will need a special firmware file to make it work with ath12k.
do you know or explain a little bit more, why is that?
i use latest linaro 1.6.1 firmware… and board.bin from compex support it self- latest ath12k driver with kernel 6.19.rc-6
↓
Which card exactly are you using?
this one exactly
wifi mPCI-e card ath12k_pci 0000:07:00.0: Hardware name: qcn9274 hw2.0
router: https://www.newegg.com/global/mx-en/p/22Z-009R-00052?srsltid=AfmBOoo0BvxTleyrni5oyy03sihDfub5iuQsGlkMe9lCPbxmnHXMH624
OS - linux
wifi range- great.no issues.
Which card do you use for 2,4GHz band?
in another very simmilar miniPC i use this card:
WLE7002E25
for now what i wanna to achieve is MLO between 5GHz and 6GHz because i do not have too much 2.4GHz Clients. 2.4 Ghz Band and it is too slow anyway….
I sugestie to move discussion about ath2k to another already open topić Here
Can you share the ath12k driver?
I've been trying to figure out how to backport it from kernel 6.19.rc-6 for my WCN7850.
i do not use openwrt, just building each kernel version (6.19.rc-6) for amd64 and run it on ubuntu…
i use openwrt amd64 image, only to take configuration fir desired wifi/networking setup. and then apply it on ubuntu,
will see soon if the latest openwrt supports MLO for my DBDC Compex wifi 7 (5+6GHz)card.
I compiled a firmware with this patch and didn’t work. The signal strength, measured using my smartphone, was bad and also on LuCi just appeared 255 dBm. I applied this patch to `mt76`, not the kernel as the original patch proposed.
EDIT: Sorry, I got confused with the Banana Pi forums and my mental struggle to get the most out of this paperweight. The photos are with the ImmortalWrt patch:
... With that ImmortalWrt patch, my faulty BE14 seems to report TX correctly.
However, its performance is significantly lower in both range and speed compared to a GL-MT6000.
I'm going to buy some new pigtails and a plastic case from AliExpress to see how it goes... I'm already broke...
UCI has MLD config but hostapd doesn't have mld_ap, mld_id, mld_addr parameters.. how so?
OpenWrt doesn't translate wifi-mld UCI to hostapd MLD parameters yet.
The scripts (/lib/netifd/wireless/mac80211.sh) don't have code to handle mld_ap, mld_id, mld_addr.
But hostapd configs are missing: mld_ap=1, mld_id=X, mld_addr=XX:XX:XX:XX:XX:XX
Result: Two separate APs with same SSID, NOT true MLO
