Proposal for next glinet travel router and MiFi router hardware

Proposal for next glinet travel router hardware:

  • CPU: no matter, e.g. like in glinet 750 or 750S, which is cheaper
  • LAN: 1 GBit or 2.5 GBit, which is cheaper
  • Wifi: Slot with Wifi 6 card, e.g. 10€ Intel 200AX card

Proposal for next glinet MiFi router hardware:

  • CPU: no matter, e.g. like in glinet 750 or 750S, which is cheaper
  • LAN: 1 GBit or 2.5 GBit, which is cheaper
  • Wifi: Slot with Wifi 6 card, e.g. 10€ Intel 200AX card
  • Slot for WWAN card, empty or with e.g. 10€ 4G card, exchangeable with 100€ 5G card

Seems this is not easy. I forwarded this to product managers.

1 Like

@alzhao, follow the short version for the glinet product manager:

  • a slot on which the user can replace outdated wifi cards by a actual one like today a 10€ WiFi 6 card like Intel AX 200

Proposal for next glinet MiFi router hardware:

    • a slot on which the user can replace outdated wifi cards by a actual one like today a 10€ WiFi 6 card like Intel AX 200
    • a slot on which the user can replace outdated WWAN cards by by a actual one like today a 10€ 4G card and in some month by a what ever € 5G WWAN card

The other hardware features are already ok as it is.

modular systems have never carried a small footprint, and you have to consider newer unknown hardware may require more power which can lead to excessive heat. sell used and upgrade or just slap a batterypack on an ar750s and plug your usb 4g dongle into it. why wait. if you need ethernet then your most likely not so mobile and footprint isn’t an issue. I’m all for an 802.11ax

I am not interested in footprints. Its a question of how to get in future actual hardware based tecnologys running in gli routers.

Actual are par example:

  • Wifi 6 and not WiFi 4

“running in” very confusing and many unknowns

I am not seeing how being able to swap a lot of modules will make a lot of sense for these products. Especially travel routers need to be reasonable sized and low power to allow them to work long from battery power.
WiFi standards are backwards compatible and let’s be honest: None of the devices in GL inet’s productline are not really going for the AX11000 class like the Asus ROG Rapture GT-AX11000.

Probably if you go looking, you will find routers with 2 or even 3 mini-pcie interfaces so you can indeed install your own wifi-chipset and wwan devices. But those are not aimed at what gl inet seems to focus on…

I completely agree. There are tons of brands focused on corporate and bleeding edge testing that have big bulky routers with swap-able modules. This is not really the idea behind GL-iNet. Routers should be as fast as possible, portable and using OpenWRT. They won’t run bleeding edge hardware that is not finalized, it will be mature hardware.

I’d love to see a unit that has 2.4 to pick up signal, wisp over 5ghz. Most of the time, I’m in the same room as my ar300m, but speed is much slower than direct connect to hotel WiFi. Similarly slow when I test at home with my Ar300m connected to my guest wifi.

@Henry_Bruns

For my usage cases with the Gl-iNet range, which is using a portable router while travelling, all the proposed flexibility and speed isn’t important to me. What’s important is a well supported, stable, small, portable unit. When travelling my normal usage is web browsing, email and a bit of video. The fact that my travel router doesn’t have the latest specs doesn’t worry my as even if I’m on holidays if my streaming video service (which is probably the most taxing usage) is HD rather than 4K isn’t an issue, heck if I’m on holidays SD will do :slight_smile: .

Actually case in point even not on holidays. I’m housesitting an old 1800s house , double brick construction, and my main B1300 router won’t even provide a good signal to a room about 15m away because of the 4 double brick walls it has to penetrate. I’ve just used my $28AUD Mango in WISP mode to boost the signal to my Android TV box to get a reliable signal. Android TV box is wired to the Mango and I should probably run in Extender mode rather than repeater but the technology in this cheap, stable box means running in repeater mode gives me 37Mb (on a base 50Mb connection) which is more than sufficient for my needs. To add all the technology you’ve suggested would give me marginal gain for the additional money I’d need to pay for a “specced up” device.

Adding the flexibility you’ve indicated adds complexity, heat, power consumption, size and cost. Of course depending on your usage YMMV :slight_smile:

@axebro
Mmm, aren’t you just describing the AR750 Creta, AR750s Slate or the new Mudi :slight_smile:

Customer voice about new revision of proposal for next glinet travel or Mifi router hardware:

  • no blue LEDs, because that is unhealthy for the eyes.
  • size of router, not bigger than AR300 or AR750
  • actual USB conector for power supply
  • DDR4 for better error correction
  • Flash which is big enough for actual firmware (it can be, from 3.1.003 to 3.1.004 was removed a part …)
  • CPU: no matter, e.g. like in glinet 750 or 750S, which is cheaper
  • LAN: 1 GBit or 2.5 GBit, which is cheaper
  • Wifi: Slot with Wifi 6 card, e.g. 10€ Intel 200AX or gl prefered vendor qualcomm
  • No external antenna or a standard connector for a true external antenna.
  • Slot for WWAN card, empty or with e.g. 10€ 4G card, exchangeable with LTE card

I think you’re going off the wrong angle here, @Henry_Bruns.

First of all, the CPU does matter. First of all, even the latest GL-AR750S uses a QCA9531, which has its own limitations (e.g. it doesn’t support DDR4 RAM, and I highly doubt the WiFi 6 throughput would be great). Qualcomm does have dedicated SoCs for WiFi6 applications, even ones that support DDR4: for example, the Networking Pro 1610 Platform, which, based on the specs, would be the perfect candidate for the next line of high-end routers, with the travel routers getting probably the lower end of this spectrum (say, the 610 or 810).

Also don’t forget that WiFi 6 will require at least 3 antennas (but rather, 6, to provide 2x2 MU-MIMO and be able to act as a repeater without major setbacks), which will increase power draw and heat as well.

For flash, 16MB SPI NOR flash is perfect. You don’t want to go bigger, as NOR flash is expensive (but secure, no sudden bad erase blocks). Then you can pair that with a larger NAND flash (like the MUDI does), and make the OverlayFS of OpenWRT out of that (read-only kernel and root file system reside in NOR flash, user configuration, including installed packages, reside in NAND).

For LAN, I’d go with 3x 1Gbps ports - 2.5Gbps is still quite expensive, and not a lot of uplink routers support it. Especially not while travelling, you’re lucky to get 50Mbps in a hotel, let alone 1Gbps; making 2.5Gbps a bit useless for a travel router.

For Wifi… you’re looking at a RECEIVER card. That’s why it’s so cheap. The 200AX (and its brothers) are fine for a device that is receiving a single connection from a router uplink, but is terrible for any sort of access point usage. You’ll get some massive issues the moment you have more than 3 devices connected to it. Consumer grade wireless cards meant for clients are NOT good substitutions for this type of hardware.

I agree on the WWAN card, however it adds further complexity and possible incompatibility issues. Because of that, I’d rather see more USB ports on the device (say, two USB-A 3.0/3.1/3.2, alongside with a USB-C power port similar to the MUDI, but also allowing proper USB connection over it). Maybe even a recessed USB port behind a door, so that a 3G/4G dongle can be plugged in and hidden without it sticking out?

Overall, designing a new router is not as easy as slapping together a list of wants. It will take time and effort, and if you really want all those features, well, it will cost.

For WiFi6 are 2 antennas are completely sufficient. As an example you can see the datasheet of an Inteal AX200 card. I have some of these in use. And yes, these are WiFi6 cards, cost around 12€ and are designed for two antennas. You could even use them with only one antenna and still have a much better channel handling with busy WiFi networks, protected management frames, WPA3 and also a higher speed because of the higher encoding. By how many peoples are the travelrouter used ? By one, two, 32 ? Does a two antennas WiFi6 card have a problem with 32 user at the same time ? No.

The developers at GL will certainly find a solution that will appeal to the buyers and thus satisfy the shareholders.

An alternative to an exclusively in-house development is to outsource parts of the required development or to order OEM routers and label them as gl.

Yeah and if you go to anything more powerful than the current processors that GL uses, you are locked into the Qualcomm SDK, and also insane heat and power issues. If you look at internal photos of for example the MV1000, you will see it has a big metal heatsink. Same goes for other more powerful GL routers. This means they are not good processors for portable routers. For fixed home routers sure, but they will be bigger. To get the full performance of Wifi6 you need multiple antennas. Just look at a Wifi6 router from Asus for example, they all look like spiders and are massive.

Not useing WiFi6 is a security risk.

You don’t understand that that card is a client, and not a host, do you?

Clients communicate with one AP, making the AX200/201 efficient for client communications. In fact the AX200 series does not even support WiFi 6E (the 6GHz bands now available for WiFi) - that’s why it only has two antennae only (one for 2.4 and one for 5GHz).

On the other hand, APs need to communicate with 2-128 clients simultaneously. Client interfaces, such as the AX200, are not designed for this, and would perform really poorly in this scenario. Yes, they can host (pretty much all desktop OS has a built in way of “sharing” WiFi, though the performance is not that good, and I believe the maximum client number is 8, at least in Windows), but it’s not a good experience, and certainly not something you’d want in a commercial product advertised as a router.

Depends. In my case, it would connect my laptop, two phones, tablet, camera, and a bunch of other portable IoT devices I have on me.

Again, depends on the card, if it’s meant for routing purposes (optimised for multiple client), or for receiving purposes (optimised for establishing a single, high bandwidth connection). It’s not just about the number of antennae, but how they’re used. For example, for a WiFi router to be able to act as a repeater (receive WiFi signal as well as broadcast it) with proper performance, you’d either need two separate radios (one for receiving, one for broadcasting), or a radio that can establish this receive-broadcast pattern over multiple antennas (so there’s no drop in performance). You can do a repeater with a single antenna and radio, but it will (at least) halve the bandwidth, as it will be receiving the signal 50% of the time, and broadcast in the other 50%.

That’s not true. The AX200 uses two antennae for the two separate bands - 2.4GHz and 5GHz. By removing one of them, you’re practically limiting yourself to a single band (not entirely true, since the antenna connector itself is able to act as a VERY LOW performance antenna, so you’d still get some signal, but it would be :poop:.

I doubt GL will ever go this way. By my understanding, the team wants to provide the best user experience (which is why the products are priced higher than other similar products, but also work considerably better), and by outsourcing hardware development, they lose the hands-on required for their goal.

I’m aware of this, sadly. So far, all the WiFi6 cards are either very low performance, or eat a lot of resources, and managing the bandwidth provided by WiFi6 also requires beefier hardware (which is why all Qualcomm SoC for WiFi6 is so heavy on power, and run on 4x Cortex-A53 cores).

Would Mediatek be an option? I’m aware that you’ve moved away from their platform in favour of Qualcomm Atheros, however I see no other low(ish) power options that would be good for a battery operated travel router.

This is bullshit. WiFi6 itself brings no security improvements at all, and the fact that most WiFi6 chips and cards have WPA3 support is not causation, just correlation (new product, new standard, of course manufacturers will put the next big thing in their new products…).

I’m stopping here for now. It’s a pity. Gl has a nice surface that some customers like.

Dude, you’re rambling all over the place. It’s clear you have very little technical knowledge, and a view on that that it’s complete. The GL interface has NOTHING to do with WiFi6. Security has NOTHING to do with WiFi6. WiFi6 is just a higher bandwidth solution over WiFi5 (802.11ac), with a bunch of technical improvements, and the requirement of WPA3. But just because WiFi6 requires WPA3, it does not mean it’s automagically more secure - OpenWRT has implemented WPA3 support on a lot of chipsets, going back all the way to the ancient MT7628.

Another analogy is how Wireguard vs OpenVPN is. Wireguard is a new program written from the ground up to use the latest encryption (ChaCha20), while OpenVPN has support for ChaCha20 in the latest version, but was not built for it.

In the same way companies can offer a firmware update to their wifi chips to enable WPA3, while it is required for new cards that have Wifi6 to have it.

Precisely. New hardware will have proper crypto acceleration for WPA3, older hardware will need to rely on CPU (which is slightly less performant, but works okay in most cases).

@Johnex what do you think about using Mediatek SoCs for WiFi6 support? They’ve released a few really good ones that would be a good match for a travel router (the MT7622 is a good candidate for a main SoC paired with an MT7688), and their WiFi6 Wave1 SoC, the MT7915 seems quite promising as well.