Need help to build my own 5G mobile broadband network

Hello to all,
I'm opening a new topic because this one appears to be dead.

I'm exploring the DIY option after I saw the limits of commercial 5G routers. Not only they are already expensive on their own, but they need to be purchased again each time the 5G network evolves.
Where I live we can achieve a maximum download speed of around 1-1.35Gbps, but ISP are constantly updating their infrastructures and since I'm building my network from scratch, I want it to be more future proof as possible with the ability to change the single piece when it's time to update.

Main problem is that all the embedded boards I saw with a 10Gbit ethernet port are extremely expensive and I don't understand for what reason given that a 10GBit PCIE network card can be purchased for around $50.

My network is not complicated. Other than the regular connectivity features I want to run a file-media server (like jellyfin, ProxMox and similar). I thought to a Mini PC or something like Dell Optiplex Micro, mini-ITX or similar (although they look like desktop boards and I'm not sure on power consumption).
Considering there is more than enough power, I thought that it could act as an OpenWrt router with additional stuff like pfsense and other desirable stuff to enhance the network. I was hoping it was possible to connect the router to a 5G modem-only powered by OpenWrt for a seamless integration. Or, much better, expand the router with the required pieces like 5G module, SIM card slot etc (in this case I would not even need 10Gbit ports anymore because all the connections would go through wifi6e/wifi7 which I suppose would be more enough to get all the bandwidth).

Regarding outdoor vs indoor, I will certainly keep the antennas outside on the roof. I also I have no problem to keep the whole thing outside, but I am not sure I can get any benefit. From what I have understood, if distance is more than 10 meters, it requires PoE to avoid losses and I think this is not my case (although I still have to do all the needed measures, I believe I should stay within this limit).
Unless there are other noticeable advantages, I don't see any reason for a full outdoor.

To conclude, although seemed something very easy to do, I realized that, even if the network I want to build is not complex at all, there are so many things that can go wrong.
It's better to ask experts for help.

Thanks again

1 Like

While cellular client devices are common within OpenWrt (i.e. a wireless cellular wan), running a 5G/cellular mobile broadband network requires specialized equipment that is not really in the domain of OpenWrt.

OpenWrt is a router OS which can certainly serve at the core of your IP-level network if you want, but it's not going to give you the operator level cellular radio systems you need.

Can you draw a diagram of the system topology you are considering and how and where OpenWrt fits in?

EDIT: It's also worth noting that in most, if not all jurisdictions, you'll likely need a license to operate a cellular network of any kind. And towers or buildings upon which to mount the antennas. And the cellular broadcast equipment necessary to setup a network. And a whole lot more, normally. So if you haven't gotten that far yet, you may want to start with that understanding the requriements for building out a network from regulatory considerations to logistics to equipment required. OpenWrt would probably only play a small part in your network.


Many thanks for your fast response.

Let me first clarify the misleading title (I realize this only after your edit). Please feel free to edit the title accordingly once I clarified.
I'm not a company who want to distribute mobile connectivity all around the city for which, of course, I would need government licenses and many other required authorization.
I'm a private with a regular 5G monthly subscription and all I want to do is to connect to their service through my home 5G router and share the connection between a couple of devices inside my home. Think it like 5G routers sold from famous brands to the final users in order to make them able to share connectivity with other devices of their home (ex: Zyxel routers).
Only difference is that I want to build my own one for the reasons above like does many people.
No license or any kind of authorization is required for this as long as it is for personal use and used for their home, same for antennas. It's a different story for a company that want to build routers for the market (in this case I think some certification about the hardware could be mandatory). Every one here is free to use an outdoor or indoor commercial router sold by the famous brands like the linked example or build a custom one by separately assembling the parts.

I never had chance to play with OpenWrt, so it will be more difficult to explain. I will do my best, but first let me see if I have understood well what you are trying to say.

Do you mean that OpenWrt is able to handle and do all the routing with a 5G connection (for example in the case of a 5G modem connected in bridge mode), but will not be able to handle all the specific features (example: bands selection and aggregation, enable/disable cell lock and all the other related configurations) ?
If so, does this apply to all or just specific brands?
For example some of wallystech router boards support OpenWrt through forks, some of compex through SDKs with QCA binary drivers available for software developers. I also saw some other brands advertising their boards as OpenWrt compatible, but I don't remember where. Some other through their own RouterOS.
I mentioned the above just for example because the ones with 10Gbit ports are too much expensive, but the mobile processor (and even the the whole 5G module) is the same even if I use my own board (mini-ITX etc etc).

Very hard right now to describe a system topology, let me first see if my understanding about OpenWrt limitation are correct.

Ok... so if I understand your goal here, it is very simply to do the following:

  • use a 5G cellular modem as a wireless broadband uplink for internet access
  • use an OpenWrt router to distribute that to your devices.

This is absolutely possible and, your are correct, does not require any special licenses or anything else... basically just the normal consumer equipment and you're good to go. I'm sorry that I misunderstood your initial question.

There are many ways to achieve the goal -- the two most common are:

  1. get a router that can run OpenWrt and has a USB port into which you can insert a USB 5G modem dongle device


  1. get a router that has a built-in 5G cellular radio. These tend to be more expensive than most standard wifi routers, but maybe break-even when you combine the costs of the router and a USB stick from option 1. They are more limited in terms of brands/models, though, so your choices are fewer.

So with that in mind, you will probably want to describe your intended use so that you get the best suggestions... for example:

  • travel vs home (or both)
  • how many devices
  • expeced bandwidth
  • coverage areas
  • other needs

I'm glad to hear good news and sorry for the misunderstanding, this is totally my fault.
As an example let's take the Zyxel NR7501 5G NR Outdoor Router which actually could be considered the top performer.
This is a typical application diagram (let's forget full outdoor vs partial outdoor for now):

On the roof the Zyxel router running its own RouterOS, inside the home the residential gateway (which I suppose could be an OpenWrt router).

This follows the recommended principle to split the network into multiple devices, each one with its own dedicated function for the reason we all know and it's what I'm trying to follow, but the two devices are connected through a 10Gbit ethernet port and this is extremely expensive, unless they communicate wirelessly, but I don't know if this is possible at all, even with OpenWrt in both the devices.

So, let's start with just an "All in one" OpenWrt device and what I want this device able to do. Then, based on what can't be done we can split the network into multiple devices.

  • 5G connectivity for Internet (current expected bandwidth: maximum 1Gbps, future expected bandwidth: 2 to 10 Gbps, depending from the ISP updates and how 5G evolves)
  • Internet usage: only within the house (two apartments and a ceiling) and the garden, but this is mainly a job for the wifi for which I don't have any concern. Occasionally I need to be able to access the router while on travel through a mobile (mainly for streaming videos and/or manage files, more details on the next point)
  • I want it to be able to act as a file server (ProxMox or similar) and a media streaming server (Jellyfin, emby or plex). Storage devices are a couple of flash drives (max speed: 8GBit per device) plus a future HDD (even the faster one will not go over 2 or 2.5Gbit). The maximum number of users are 2 and each one with its own exclusive access to the storage. This means that there is only one connection per storage and the number of storages accessed simultaneously should be 2 (or 3). So, USB 3.0 gen1 ports should be more than enough to get the whole speed of the devices. I didn't thought to faster storages like Nvme because I don't need such speeds (and because most of the boards already comes with the needed slot)
  • I want it to be able to run pfSense or similar as a firewall along with a powerful adblock (and install OpenWrt packages to extend functionality, but this is an OpenWrt device and, of course, I can)

So, what could be a board most suitable for OpenWrt with the ability to fulfill the above requirements?

I saw such USB 5G pens, but I have not investigated because I can already imagine an infinite amount of issues and not even close to the performance of a router like above. It would be like trying to get 5G connectivity from a mobile. They are not intended to handle such kind of traffic and I can already see how they can fail in numerous way.
I suppose you mean a router board with a dedicated slot for a 5G module, right? In theory the performance should be the same since the Zyxel and a separate module share the same chipset, but how OpenWrt will handle it? What brand for OpenWrt support? Qualcomm. MediaTek or whatelse? For example the router above should use qualcomm snapdragon x65 (or x60).
I saw so many issues and different solutions in that thread and this is where I'm stuck.
Here an user is using Banana, but with troubles. I'm open even to such devices (included Raspberry or similar).
My budget is around a couple of dollars.

Now do the same search for 1000BASE-T and 2.5GBASE-T PCIe cards, at least around here those would be ~5 EUR and ~12 EUR each - and consider the cost that adds for the manufacturer. There are two more aspects to consider, keeping a 10 GBit/s network card busy (~driving it at full speed) also requires a considerable amount of CPU performance (so a faster SOC) and uses more power and generates more heat, so you have to spend more on the power delivery side of things and (active) cooling. All these add to the overall tally considerably - add the early adopter's tax and the economy of scale and you do see where the money is going.

That's all fine, also with OpenWrt as a bare-metal installation - but… would this be a good place to mount a mini-PCIe, M.2 or USB 5g card? Think about placement, as well as the effects of dangling from a 'big chunk of metal' (your PC case) blocking reception in at least one direction. Long feed lines are out of the question for WLAN, 4g/ 5g and other wireless communction signals as well, the attentuation would kill the signal for any reasonable cable distances completely. Considering that you want to reach really high speeds (10 GBit/s, to be frank, I'd be sceptical if your ISP can provide that long term - and even if they could, if they'd let you hog that much bandwidth for longer), you will need to place the antenna outside, probably in an elevated direction with good line of sight to your cell phone tower - not reaaallllyyy an environment to place a mini PC…

I personally would start by re-thinking the entire project. I agree with @slh, but will add some of my thoughts, too.

It generally is not recommended to run your router in a virtualized environment and beyond that it can be problematic if you're running all sorts of services such as file sharing on your main router. This can lead to issues ranging from performance to stability to security. It is simply best to run your router as bare-metal and to minimize the number of services that are running on the device. Then, setup a server behind your router that can do everything else you want.

Next, while it is possible to run multiple routing environments (such as pfsense and OpenWrt), why is this attractive to you? Choose one as your router.

Moving onto the physical infrastructure:

  • there are lots of CPE cellular modems that are designed to live outside and connect by ethernet (often PoE) to the customer equipment. Find out what bandwidth your ISP can provide now and in the near future (based on the prices you're willing to pay) and get a device that can handle the speeds. Even 1Gbps is somewhat rare for sustained transfer speeds over cellular (from a capacity and account/throttling standpoint for consumer grade connections at reasonable prices), so you may not need a 10Gbps device at this point in time. Buying a 10Gbps 5G cellular modem now will probably be more expensive than the combination of buying something like 1Gbps now and 10Gbps in 5 years.
  • Inside your home you can setup a 10Gbps network if you want, but those switches do tend to be expensive, so you may find that you'll get a better bang for your buck with either a small 10Gbps switch or a larger switch that has some 10Gbps ports for your highest bandwdith devices and 2.5Gbps or 1Gbps for the rest of the ports.
  • There are many router options which can accomodate wifi out there, but when it comes to wifi, you should get a device that is purpose built for the task. That is to say, either get a true all-in-one wifi router, or get a dedicated wifi AP. Don't try to stick wifi inside an x86 device or any other type of 'roll your own' router. The radio and antenna design can be better optimized when the manufacturer has control over the physical form factor of the entire device. That translates to better performance -- range, bandwidth, multiple client handling, thermal design -- and also lets you place it in the optimal location(s) for your home.
  • And as for the router itself, once you've selected pfsense vs OpenWrt (or any other contenders), then you can make hardware decisions that are meaningful (unless you've already chosen x86 as a platform, for example). Your router doesn't need to be a 10Gbps capable beast if your ISP speeds aren't going to get there for many years... save the energy and cost and get something more modest and upgrade when necessary. If you decide to invest in a 10Gbps switch now, you don't absolutely need a router at that speed unless you really will have that speed now (or very soon) and/or if you have a lot of inter-vlan routing. The switch will be ready for the 10Gbps internet when that is available to you and it makes sense to upgrade the router. But for the near term, that high performance router will just be an extra expensive energy hog (both purchase and operational costs) until you really have internet speeds approaching 10Gbps.
1 Like

One more thing, if you are in an environment where you could get 5g reception at 10 GBit/s speeds, it's very likely that fibre (ffth) would be available as well, which is likely to be cheaper - and will be faster and more reliable than 5g by a fair margin.


@slh Many thanks for your useful observations.

In the case of compex boards I don't see any active cooling (just an heatsink), so I'm still wondering where money goes to justify prices of 500-1000$ and even more. I thought to the processor, but I have been surprised when I discovered it's around $30.
And for such prices it is included an useless USB 2.0 port.
I fails to see where so much money goes, at least in such specific case.

Well, 10 GBit/s is not at all my target speed, it is more of a maximum limit to ensure my network will support any future ISP updates (the alternative is a 2.5Gbit port which is not enough for long term). I suppose the board slot for the 5G module will remain the same and able to connect all the future released modules (as long as the board itself has enough power to handle the increased performance, of course). If not, then this can complicate things.
Edit: No, as stated from the beginning, where I live we can achieve a maximum download speed of around 1-1.35Gbps. The only purpose of making right now a capable 10 GBit/s network is to make it ready in the long term for any enhancements ISP will do to their 5G (included the 5G evolution). As of now we are still in NSA. FTTH is available, but not at my location (and I doubt it will ever be, at least at reasonable times). And ironically, even with an FFTH, 5G is still better because a monthly subscription is much more cheap.

My only concern when I saw such boards has been the power consumption. They seemed to be something between a desktop and an embedded board (although I also saw embedded boards indistinguishable from desktop ones). Size itself was not even a concern, the needed enclosures for outdoor use make them no more bigger than the outdoor router I posted here as an example.
I didn't even checked for any ip68 certification since most people here says it is useless for our climate (we never reached extreme temperatures like -40C° or 60C°). However, this is not something that all agree and I just thought to stay on the safe side and take a board which can operate at such range.

The signal degradation and attenuation has been the main concern from the beginning and I am still confused about this because I continue to hear different opinions, but they all agree that, at least the antennas, should be placed outside (elevated direction with good line of sight to the target cell phone tower as you said).

Now, let's consider again the "All in One" device I have taken as an example. Everything outdoor and the only connection with the inside home is to AC socket to power it up. Let's forget about the accessibility troubles (and the USB flash drives kept outdoor!).
All devices connect to it wirelessly, no other connections of any sort.
In such case, the solution is extremely cheap and no loss can occur for obvious reasons. Am I correct?

This is an extreme solution, so let try to adapt to something more feasible. Antennas kept outside (in the same position as the previous option) and the same "All in One" device just below the roof porthole.
It's here where you see signal degradation (even if distance is reasonable) ?? And no workaround for this? Some special "no-loss" cable to connect the device to the antennas?

Yes, this was my initial idea, but once I saw how can become expensive a capable 10Gbit network I falled back again to the all in one solution (or at least reduce 10Gbit Lan connectivity to the minimum).

You mentioned performance issues, but as you saw there are only a couple of users in the network. A good processor with enough RAM are the first things I've looked for.
I didn't understood well what you mean for stability and security and if also applies to my specific case.

I was not aware that pfSense was an equivalent of OpenWrt. I've been told it was the best firewall and security system and I believed it was mainly intended to be integrated with OpenWrt. Based on what I learned now, of course I will forget about it and will go straight away with OpenWrt, but now how to get such features? Can you recommend something of the same quality? May be some package for OpenWrt?

Your thoughts are smarter than mines! Do you mean something like this or this (first option in both the links)? It has PoE, so I assume it is implicitly designed for outdoor. But can I expect the same performance through an adapter like this when compared with a board with integrated 5G module slot?
And with this I have also solved the problem with losses and signal degradation pointed out by @slh.
I never had chances to play with 5G, so even if I have done the appropriate researches, it's very difficult to estimate. What I can tell is that the most suitable tower is less than 2 Km from me with full visibility. Our 5G is only NSA and these are the details of the tower:

NR Info: N3 N78 (5G Status: TDD sub-6GHz AAU, FDD Passive, DSS)
LTE Info: B7 B1 B3 B20 B38

Regarding account/throttling standpoint, this does not occur, at least not with the specific carrier I will subscribe. All bandwitdh available from the tower goes straight to the receiver with no limit. The only limit is the data amount an user subscribed for, once reached the limit the connection breaks. I don't take in consideration when the service bandwidth is reduced due to congestion since this is something that involve all the connected users (and I never experienced this from a mobile). I looked at the tower history and bands has been updated constantly over the time. So, if not much more expensive, the same linked adapter, but with a 2.5Gb PoE port? If I can find one...
Such adapters run under x-wrt or similar forks and I don't know if I can be safe on the development continuation (after I have carefully checked they don't lacks features). Is there something officially supported from OpenWrt?

This is another smart idea, but... is there anything wrong to optimize the whole thing by also using USB ports? USB 3.0 ports are already bundled in nearly all boards and speed of the storages are much less than 3.0 capability. In any case, the router needs a 10Gbit port if I want to have a minimum setup for future 5G updates plus one for the switch, but, as you said, I'm not convinced at all is a good investment. What prevent me from totally abort the idea is the fact that 10GBit PCI Network cards are cheap and may be there are still chances to use them in some way. I don't know, but I should evaluate that option only when I can see the exact amount of money needed.

This is what I have been told. Wifi AP are pretty cheap. Do you recommend wifi6e or wifi7? I saw boths at nearly the same price, but I have read that the current wifi7 are some sort of fake (didn't investigated further).

OpenWrt !!!! of course!! I totally agree with what you say. Given that now the network is divided into multiple devices, does still make sense an x86 as a platform?
The services to run are much less and the high performance seems an overhead. What your recommend?

So, let me see if I have understood correctly...

The purchase list:

  • 1 CPE cellular modem or adapter (regarding the 5G module I will follow the same rule, the cheapest one that fullfill my current estimated speed, may a bit more to cover the immediate future updates if price difference is acceptable).
  • A switch (number of ports and speed to be evaluated based on what I find in the continuation of this list)
  • A server for network file sharing, Jellyfin, proxmox and everything I want to run. No OpenWrt here, right? Maybe a very lightweight Linux build with no desktop manager? May be a fanless Mini PC?
  • wifi access point
  • An OpenWrt powered router
  • Antennas (here I must choose very carefully and needs assistance)

The connection diagram (there are for sure mistakes!):

  • Antennas > 5G modem
  • 5G modem > switch > OpenWrt router
  • Media server > switch > OpenWrt router
  • Wifi AP > switch > OpenWrt router

Do you see any loss with this?

As of now, with the current 5G achievable speed, depending from the available ports, is there the risk that the whole stuff could be connected even without the need for a switch if I end up with no 10Gbit network? In such case, wouldn't be better to delay the purchase when it will be really needed?

I really appreciate your help. Believe me when I say that I spent weeks trying to understand how to build my network while, with just a couple of your messages, I have a much more clean idea.
And I found you at the right time because Aliexpress Summer sales has just begun! It's the right to make the purchases!

I'm stuck in the adapter and module selection!

And I still have to understand if the modem should go completely outdoor in the roof or just the antennas. @slh talked about signal degradation and attenuation but in this source I read that although the length of the antenna cable affects the attenuation of the signal, in most situations it will remain irrelevant whether the connecting cable is 5 meters or 15 meters. In my situation I should be able to remain within that range. I have no problem to move everything outside as long as there are valid reasons and noticeable advantages.

I found the adapter in USB or ethernet flavour. I don't know what I should choose, there are so many different models that make me very confused. I want to connect the adapter directly to the above router (I wrote above the connection diagram, but I don't know if it can work as is). Does the router will recognize both the adapters? To me seems preferable the USB version since every router has an USB 3.0 port and I will get full speed. With ethernet port I'm afraid I'm stuck to 1Gbit (or 2.5GBit if I'm lucky to have one port). I'm also not sure what of the twos works better in case that I have to move the modem outdoor. I supect these adapters runs through x-wrt and similar OpenWrt forks. If you can recommend something different fully supported from OpenWrt I think it should be safer. I'm not sure about the performance, but I suppose this only depends from the 5G module.
I prefer an OpenWrt fully supported device because if something does not work it will be harder to get help (same for the module and router):

This is where I'm more scared. I have heard that if it is not good enough, not all bands will be aggregated. I found Fibocom FM350-GL and RM520N-GL, a more expensive version and I see that prices for this one goes from $80 to $175, all marked as RM520N-GL, it only changes the colour of the plate (link1, link2). I can't understand what are the real differences in concrete terms. I hope you can suggest a good module. I know perfectly well that I must be very careful with modems because OpenWrt does not support all the processors and some of them only have very limited support.


I thik the whole thing would be much simpler for them to get a cellular amplifyer and just relay the wholle cellular network to the inside of the house.

I set one up for my ocusin nd it wi=orked great. andnot only did it repeat her carrier but eveyone in the house could now use their phones.

Something like this.

Maybe the first question you need to ask is, do you need an outside antenna?
Have you checked what kind of signal quality you are getting inside?

Personally, i am so close to the antenna near me that i am doubtful i would see any advantage of an outdoor antenna, and i have just set up my system inside.

I have a USB sleigh containing a Quectel rm520n-gl modem, plugged into a cheap fanless x86 box i bought used, this is connected to a zyxel GS1900-8 switch that has one wifi AP plugged in, a Netgear wax220. My flat isn't huge, so this setup is basically enough.

My considerations was initially that the place with the best antenna reception wasn't the ideal place to place an all in one wifi router, so the 5G-modem>router>switch>AP became the best solution.

But, back to the antenna part, i simply downloaded the app cellmapper on an android phone, checked my flat and ran some speedtests with different SIM cards from different ISP's in order to find the place with best reception and speeds. And then basically expanded from there.

I initially ran a LTE cat16 modem to an rpi4, and then directly to a GL-Inet B1300 AP, i then later added the switch and started wiring up and adding a NAS. The local telecompanies are right now upgrading to 5G, so recently i switched the modem to 5G and then finally i exchanged the rpi4 to an x86 box, this way i have a little more horsepower for different experiments i do with SQM.

One thing to note, currently with a solution like this is that 5G speeds are slower on an USB sleigh than a PCIE, so to get full speeds in your case i would choose a PCIE sleigh to start if you wind up choosing a similar approach.

The zyxel is probably the most easy and headache free, all things considered

EDIT: Only now i see you already looked at board, i would probably look at the Mcuzone 2.5g boards, they can do both pcie and USB, and can accomodate 1gb+ 5g speeds as far as i know.
The rm520n-gl is great, the difference between the 2 is that the cheaper ones usually is pcie only (RM520NGLAP), while the other is USB (RM520NGLAA), with option to switch to pcie.

If they want mmWave (as diagramed) they will need an outside antenna.

I had not yet chance to study the antennas side of the project, so I coul have misunderstood what you are suggesting. Is such $369 signal booster just to avoid placing the whole modem outdoor? How much speed you gained (with and without) ?

Thanks @Nomid ! This is a very smart observation. I saw people gained much higher performance with external antennas. So, the choice for me was very obvious. I think you're right, I should (and I'll do) evaluate such option only once I have the modem in my hands after experimenting indoor and outdoor.
However, I believe I must ensure in advance that the hardware will be compatible with outdoor usage just in case I will end up outdoor. One thing I can think of is the PoE feature. I don't think I can later add it and even if I can it could be problematic.

I could only test 4G+ through mobile and I got 21Mbps in download and 23Mbps in upload. With the tower at nearly 2 Km in full visibility, I suspect I will need antennas.

I have heard about Quectel rm520n-gl in many places. Is this one the best? It is exactly this one ? $80 seems a reasonable price, but what about bands? I need to be sure I will be able to aggregate all bands from the tower. Some people suggested the local version rather than the global, but I'm not sure between the twos. I also saw that it is only cat 16 and I'm not sure if this translates in less bandwidth. Any other recommendation?
With one additional kilometer of distance from my main target tower, there is another one with mmWave, but without visibility. I have read it does not work without visibility, but in USA people still get fast speeds in bad conditions. I suppose because they use more powerful frequencies. Here in Italy I think the maximum is 100MHz, at least now. In addition to that, lot of towers have the mmWave inactive and I'm not sure about mine.
If I can find a module with mmWave support with just a bit more of $ it will be fine, otherwise I prefer to renounce to such feature because there are great chances of failure.

What fanless x86 box you used for router? Is good the ones I see in AE? What's wrong with an embedded system instead of an x86 system.
So, two identical fanless x86 boxes, one for router and one for the media server: is this a good design? it looks very strange to me.
You're using OPNsense for such box, right? I was initially sure about OpenWrt, but I have read that for x86 routers OPNsense is more suitable.

Is this the adapter you use for the module?
I've been told it doesn't work with the one I previously linked because it is PCIe interface. It must be installed in a device like a raspberry which is fine for me because I think it should be easier to adapt it for outdoor usage.
The reason I want the PCIe module version is because it is 3 times cheaper. If there is no difference in terms of performance, than I'm perfectly fine with this version.

Quectel rm520n-gl is cat 16 if I'm not wrong. I don't like the idea of buying a module limited to cat 16. I prefer a one upgraded to the current standards (and still able to get all bands from my tower).
rpi4 is cheaper than rpi5 and I saw it working fine with 5G modules. Can you confim that rpi5 is a waste of money? May be even a banana or orange? They seems to be more powerful.
What I have excluded are mikrotik boards or similar. They are not OpenWrt and I saw the cheap ones (ex: RBM33) getting lower speeds than rpi (reviewer said the reason was probably not enough power).

  • 5G-module>rpi4(or similar)>router(x86)>switch>AP(wifi)
  • media-server(x86)>router(x86) or (if there are no ports available) media-server(x86)>switch>router(x86), my choice for PCIe 5G module is even more justified. Not only much cheaper, but also faster!

$60 for MR5210 only board sounds good, but what about OpenWrt? Seems to be much less common than raspberry. It has the right power or there is the risk of ending up with lower speeds like the mikrotic boards? Better to stick to a rasp?

Situation with mmWave is very obscure here in my country. I can afford no more than 20-30 bucks for such feature because there are good chances that will not work.

Unfortunately the mobile I can use for testing only support 4G+ and the speed test I posted above does not seem good. I'm not sure how much could be reliable testing 4G+.
However, once I get the modem, it should be easier to perform tests and find the best setup.

It amplifies the cellular signal. All cellular devices will get a stronger signal inside the house, including your 5G router. And it comes with an antenna desinged to be outdoors.
As with all radio signals, the stronger the signal the better the throughput. It might even relay mmwave into the house.

Without she had one bar if she placed her phone in her bedroom window and used bluetooth just to make calls. It was so specific placing not even stepping outdoors got a signal strong enough not to constantly drop calls.

We put the flat antenna on the ceiling of the first floor facing down (had to drill a hole but antenna covered it) and had all but 1 in the basement. We did place it on a pole as high as the cable allowed (32' routed through the vent to attic floor so ~20' up above roof) and used a cell tower map to point it directly in that direction and then tweaked it using its app.

I was able to stream video with it @ 1080p. I never did a speedtest.

In most jurisdictions mobile repeaters (boosters) are illegal unless permission granted for use as they can cause interference. Something to consider, not saying it is not a good solution.

What FCC has to say

Gotta leave that part of the quote in :stuck_out_tongue:
If user is in US, carry on! Just pointing it out :slight_smile:

1 Like

This is very interesting! You linked an expensive consumer device and I saw there are much cheaper versions, but apart from this, it could become another fun DIY project. I have a motorized satellite dish on the roof that I don't use from very long time. Do you think I can reuse it and build something similar?

I will have to check my local jurisdictions, but apart from this, why you say is not a good solution ?

Anyway, now it's better I focus on the 5G module because this is where most of the speed comes from.
It seems that most people here use Quectel RM520N-GL and I will probably end up with it but I still have some cash for something better and hope someone can recommend some other better models. I am undecided between global and local version and this is where I'm stuck because if I choose the wrong one I could end up with some bands not aggregated.
@Nomid suggested to look into an MR5210 board, but I have some concerns:

  • Is this board good for outdoor? It would be very convenient to buy the bundle with the metal case and although it can be insulated, all the connectors would remain exposed. So, I'm afraid I will need a different case.
  • I don't see any PoE connection. It seems to exist a MR5210P model with support for PoE, but price reaches nearly 100$.
  • RM520NGLAP is the one I want to buy (much cheaper and faster):

But this is very scarying and I don't know how it should be interpreted:

  • I don't see anyone here using this board.

@Nomid Is the one that really know well such stuff. So, it's still an hazard to make the order now, better to wait for him. If you guys want to share your ideas, it would be very appreciated. For now I switch investigation into a rasp or similar. There must be some version dedicated for 5G with at least one 2.5Gbit port.