@Terran158 Why are you connecting them together? Are you extending wifi into a building or another 100m further on?

Have you made sure that the AC1350 unit is NOT operating on or near the 5.8G band that the 610 is using for its backhaul?  Keep in mind that the nominal channel being used, say 149, means potentially that 149-153 are actually being used because of channel bonding...so do the math on the TPLink mesh frequencies first and then configure the AC1350 appropriately....so they don't stomp all over each other.  Also, consider disabling user access to the 5.8G EAP610 to optimize the link for the 2.4G traffic AND the backhaul from the AC1350

  @Terran158 

What you are doing is logical and legitimate.  The way you are using the LAN ports is correct.  Changing out the POE injectors for a POE switch will change nothing.

I believe your issue is frequency related and you must be cautious with the RF planning of the mesh links.  Here is some advice from when I did something very similar.

From the main building, I had 2 wired APs that I would use to reach remote buildings.  The most important link (A) I fired up first and let it pick its frequency/channel in the 5.8G band and then I tweaked it and fixed it in the settings.  In my country and location Ch149 provided the maximum capacity AND maximum allowable power output on those channels.  For the second link (B), I fixed it at Ch157.  Now any meshed AP which homes back to (A) is forced to use channel 149 for its 5.8G radio and likewise any AP homing back via (B) is forced to use 157.  The other advantage to fixing the mesh channels is that upon power hiccup they come back up much more quickly.

Now, any client connecting to a 5.8G radio on an AP will take capacity away from the mesh and you can get to the point where local clients 'drown' out a far end mesh AP causing all sorts of packet loss and retransmits.  I would strong recommend disabling client access on the 5.8G band for the EAP610's (at least to get things stable).  You can use 5.8G on the AC1350's but NOT 149-151-153 or 157-159-161 or you will *really* clobber the mesh traffic (these are channels applicable to my setup).  You have to avoid the subsequent 2 channels if you are allowing Auto/80Mhz channel widths.  If you fix to 40Mhz, you only have to restrict the next channel up. 

For the 2.4Ghz radios, try to frequency plan around 1, 6 and 11.  Fix those channels at 20Mhz or you will be sorry.  40/80Mhz in the 2.4G band is only asking for trouble when multiple radios are at play in the area.  For instance a building might have the EAP610 on Ch1 and it's attached AC1350 on Ch6, and the building nearest that might have the EAP610 on Ch11 and it's attached AC1350 on Ch1 again...try to keep matching channels as far apart as possible.  There are handy apps for your smarthphone that will show signal strengths as you walk around...I'm particularly fond of the Android app by KaiBits.

It is unfortunate that you didn't just buy more EAP610's because they at least could have taken advantage of seamless roaming and AI wifi planning and all the rest, but presumably that ship has sailed financially.

  @Terran158 

My experience was with their slightly older brother the 225-outdoors, but I would do whatever you can to optimize the signal link for the wireless mesh backhauls.  Ideally you want a link that is -75dBm or better (ie not -80dBm) and ideally in the -60 to -70dBm range.  If you can get that inside the building envelope that's fantastic.  In my case I mounted high up on the facing walls about 0.5m below the soffit/eaves, positioned to minimize eliminate any foliage or other obstructions.

I would absolutely go 100% EAP610's just from a sparing/management point of view, but you could also use Omada-manageable indoor APs as well.  If the AP you choose also supports wireless meshing (some ceiling units do), that may make 'fixing' problematic dark spots even easier.  If the building construction is such that is presents significant degradation of signal from inside to outside that may also change your design.

If you don't have one already and you're planning on 6-10 APs in the final design, I would absolutely consider a controller, and probably the hardware controller like the OC200 since you'll have tenants dependent on it 24/7...you need this for the mesh to work anyways.

Sounds like you have enough 610's on hand to rework at least one building using them exclusively for the mesh and coverage.  Again, I recommend eliminating or at least minimizing the end users on the 5.8G radios of the devices that are used for mesh links (that includes at the main building on the wired end).  This will optimize the performance of the wireless backhaul and the experience of everyone downstream.

As for mixing and matching...you should ensure that if you need EAP610's for your solution that any other APs in the design are also EAPXXX's and are supported by your controller.  This will enable seamless roaming and the best possible user experience.

  @Terran158 

Which exact model is the 'AC1350' referring to?  If the EAP225 then no real need to swap them out, they are controllable, they can mesh, they are cheap.

The meshing is accomplished by a hidden SSID (network name).  So you can easily disable clients from using the 5G radios used for the backbone of your network if you have to.  Other units, such as the slave units connected to the LAN ports of the POE injectors of the EAP610's can have 5G radios intelligently enabled...ie make sure they are not running in the same frequencies as your mesh links are.  I create two different WiFi Profiles on my controller, one with 2G only client SSIDs and one with 2G+5G Client SSIDs, those key APs running the mesh links are assigned the 2G only access profile.  Once you get the network up and stable, you can experiment with enabling 5G client access...but I wouldn't if you don't have to in your environment.