WiFi router using same antenna for both 5GHz & 2.4GHz. I want to separate bands.

[bob808]

Hello folks,

I got a TP-Link WDR-4300 router, dual band. It has three antennas. Internally there are 5 amps in total. 2 for 2.4GHz, and 3 for 5GHz. The two of the 2.4GHz band are joined with two of the 5GHz band via a mixer or whatever it's called.
I want to know if it's feasible to separate the bands. I would solder two extra pigtails, then connect proper antennas tuned for each band, for a total of 5 antennas.
5GHz works fine for what I want. Really I want to boost the 2.4GHz line a bit.
I also noticed that on the output of the joined lines there's a LC filter.
As you can see from the photos (left side is 5GHz area, right side is 2.4GHz), the 2.4GHz line goes into a via from the C207/C360 capacitor, then on the backside, then up again near the F1/F2 mixers (?). I could either unsolder the caps and offset them 45-90 degrees, and solder the tip of the coaxial cable directly on the side of the cap. The mesh I could solder on the golden trace.
Now, would I need a LC filter as well? Or is that in place because of the longer run of trace on the pcb? Is this whole thing feasible?

[bob808]

I found out that there's a similar version of this router sold in Asia, under the code name of TL-WDR4320. It has 5 antennas, and the internals are almost identical. Same chips, same power amplifiers. And the 2.4GHz lines have their own antennas.
I attached the pictures. So if I break the links to the 5GHz lines on mine, the only thing that I would be missing would be the lc filter on the output of the amps. Any idea on how important this filter is? I'm in a pretty clean radio area. Would it be a must? If so, what values should the cap/inductor have, and what frequency rating?

[Monkeh]

You really, really should not go messing with it.

If you want separate antennas, buy a model with them or use two APs.

[bob808]

Care to to explain a bit? It's not going to explode I presume...

[Monkeh]

You obviously have no way of verifying correct operation, which means you could be pissing all over the band (and/or others) and not know it.

Messing with (relatively, for the unlicensed user..) high power RF circuits is a bad move.

[bob808]

Well, let's take it some other way.
There is a trace, that joins the two bands together, it's noted in the pictures above. If I drill a bit in the via, the signal stops at the top of the cap. Then, I connect the coaxial tip directly to the cap. Apart from the LC filter, it's virtually identical to the other one.
Now, if you know, tell me please the function of the LC filter, and the values used, and rating of the parts. If I emulate that filter, the setup should be identical to the commercial one. I don't see what could go wrong here really, by any account. I won't mess with the amps, and that output/input filter should be a pretty simple thing for someone that knows RF and knows it's about 2.4GHz. I don't think that there's a real worry here. Just look at the pictures.

[Monkeh]

If I did know the values, I doubt I'd say. And yes, the presence and values of the matching network are critical.

Just get a more suitable device instead of hacking at one - better yet, use wires. I doubt your antenna choice would work any better, they're already using dual band antennas.

You could just get a dual band antenna with a bit more gain, if they're detachable.

[bob808]

If I did know the values, I doubt I'd say.

What do you mean by this?

better yet, use wires

I just made a set of omnidirectional biquad antennas for 2.4GHz.
The 5GHz band has three SE5005L amps : http://www.skyworksinc.com/uploads/documents/SE5005L_202446B.pdf
The 2.4GHz band has two SE2574BL-R amps: http://www.skyworksinc.com/uploads/documents/SE2574BL_R_202675A.pdf
Each of them has max rating of about 18dBm. That means a total of 21dBm gain for the 2.4GHz band and 24dBm for the 5GHz band (Assuming there's no attenuation via firmware). The 5GHz band doesn't matter that much as I have no reception outside my house. I use it only for the first floor, and don't need to increase the gain at all. If anything I might reduce the output as it's useless everywhere else apart from the first floor.
The 2.4GHz reception goes on for about 10 meters around the house.
I live in a pretty remote area, there's nothing around me in a 500 meter radius. Even if let's say I start removing all the caps in the way for increased power (which I don't plan on doing, I'm mostly interested in adding the correct filter), how much damage do you think I can cause? There's nothing around me.

I say that you are being ultra cautious, and if I wanted to learn something I would just be scared by you that I might cause a blackout for the nearest town. It would be more helpful to explain a bit the role of the missing filter, give some references, and explain how not adding that filter might cause leaks in other bands.

[Kalvin]

The antenna array is used for the MIMO functionality: https://en.wikipedia.org/wiki/MIMO The distance of the antennas needs to be very precise in order the MIMO to work correctly. Separating the antennas will definitely impact the MIMO and the performance of the WIFI connection will suffer. However, for a point to point connection with a directional antennas you should be able to separate the antennas.

[bob808]

Interesting.
This is the way they are on my router and on the one that has them separated. I attached a drawing.
I'm not interested of reaching outside my house. I want good penetration to the upper floor. If I use one directional antenna directed towards the upper floor, would it have enough juice as it is to penetrate some concrete?

[Monkeh]

If I did know the values, I doubt I'd say.

What do you mean by this?

I do not condone the modification of regulated RF devices to attempt to correct for installation failures or technology limitations.

I say that you are being ultra cautious, and if I wanted to learn something I would just be scared by you that I might cause a blackout for the nearest town. It would be more helpful to explain a bit the role of the missing filter, give some references, and explain how not adding that filter might cause leaks in other bands.

I'm not an RF expert, so I'm being quite cautious, yes. As far as the 'filter' goes, that is an antenna matching network and it is very much dependent on external factors (like the PCB). You can't just throw values at it.

If you want wifi to work reliably on multiple floors, use multiple APs or put the AP upstairs.

[bob808]

Here's a nice article on this subject:
https://www.lsr.com/white-papers/antenna-matching-within-an-enclosure