New router with miniPCI-e on new chip

GL-MiFi is currently available but based on the old chip and there is no possibility to power PoE.

Is it planned to implement the next version on a faster chip?

PoE is not related to the SoC. It is a separate module. It has lipol battery and not able to power PoE.

@alzhao i would love to see an upgrade to the domino board used inside the Mifi. That would make it relatively easy to upgrade old mifi’s to the new chips, desoldering the domino board. What do you think about this?

If you then use a quad core chipset or dual core, and with 5ghz too, only an extra antenna would be needed :smiley:

I think many people would be happy and willing to pay for such an upgrade, me included!

We cannot just upgrade the core. But we will surely upgrade the mifi soon.

Why can’t you just upgrade the core? Isn’t it a self contained unit, the rest around is just power regulation and the lan no?

If you can design a new core, even if its a little larger, as long as it has the same footprint as the original core, then why not?

The problem is that each SoC has different pins. If you need to change the Core board without changing the underlaying board, you have to make the pinout nearly the same. This is generally not possible because each chip is different.

We made the core according to the AR9331’s IO. Let’s say we need to make Core2 using AR9531, we will not use the same pin out. For example, AR9531 support 5G wifi chip using the PCIe interface, we will surely have the pin out of PCIe. So using the old pinout as Core is doesn’t make sense.

With a multi layered board it’s easy to move even pins from one side to the other without issues, if one really wanted :smiley:

I understand it would be simpler for you to make a brand new Mifi, but for us that already have one a domino upgrade would be just as great.

If you do make a mifi, i hope you understand that we live an a dangerous world now in regards to security, so please don’t use a slow processor, we need to have much much faster openvpn, so a processor with built in hardware decoding. Also not single core, it is not always enough.

As much as I hate wasting things, really you are asking:
a) Take extra time/effort and increased costs to produce a new core board instead of just a new product in general
b) Do this, because it will allow existing users to do what exactly? Save $20-$30 on the cost of replacing the entire base system and core together… and only if the user has enough technical abilities to de-solder the existing core, install the new one, and not screw anything else up in the process.

Even from an enthusiast perspective it makes almost no sense at all. What exactly are you trying to save by asking GL to do twice as much work? The two most expensive components of the MiFi right now are mPCI modem card, and base processor/PCB (and by my guess those two are about equal to each other in cost). So when new Mifi comes out, you already get to re-use your mPCI card (that’s why it’s a card… pull it out, re-use it. No special soldering/etc needed). Assuming cost of new core board is roughly 50-70% cost of the base PCB in general, you will have to buy that either way… so why not just buy a complete tested working unit and add your mPCI back in?

Now if you had 1000+ MiFi’s already in use/deployed there might be at least some justification, but even then I expect it would cost you more in time to get them, open them up, de-solder existing core, solder new core, run full battery of tests to make sure all is working, etc than just buying a new unit and doing a physical swap out.

As for new MiFi, keep in mind one of the main reasons MiFi does so well is low cost compared to other vendor solutions. You can already buy much more powerful devices with 3g/4G radio… you can even just buy a B1300 and attach 4G USB radio to it… but costs for that are more expensive than MiFi. I’m all for faster/better hardware, but only if enough time has passed to let those improvements come without significantly increasing the price of the solution.

You usually post nice things here but you missed the entire point here.

When you want to have a product be FCC compliant you have 2 options. One is to design the entire product and when it is finished, you send it in for evaluation. This is not free, it can cost a lot of money. If you have many issues and have to do many times, this adds up.

The second option is to design all the RF stuff and make that into a board, and then only things around it will change if a revision is needed. This way you potentially only need to do 1 revision of the processor board and only need to do FCC compliance testing one time. There can be infinite revisions of surrounding components and other products designed around it too. This is much cheaper since you don’t need to do any new testing since only the RF stuff needs to be tested to be compliant.

Why do you think that the ESP32 and other modules sell like hot cakes? Those are designed once, FCC complaint and anyone can just take that and use it in their product without having to do any FCC stuff. It saves time, it saves money, and it’s so much simpler.

GL-iNet did the second way, they designed first the board that has the processor and everything else with RF on it, then just have an antenna connector.

I think this way is brilliant. Usual issues come from power regulation and external components, you won’t have any issues with the main processor, since everyone is just following the design guidelines for it, and the memory etc can only be connected in 1 way.

You can also have the board on its own, selling it to other companies to embed into their products.
You can also spin out different products based on the same processor/rf board.

GL-iNet chose to design the Mifi in that way, and i think it was great.
If they do decide to make a Domino v2 and then a Mifi v2 around it great.

Making a backwards compatible board with just a tiny tiny bit effort is not bad either, then you can have a Mifi v2, people (AND companies that already designed things around the Domino, GL-iNet has no sales numbers for the Domino, it could be thousands or more) can just swap in the new Domino into their device as well.

There was a reason why they designed the domino boards and why they based the Mifi around it. I think it’s a great idea, and i just hope they continue with it.


I think you mis-understood my response also. Completely agree having the core and host board as two different parts makes total sense. All I was trying to say is from GL’s perspective there is probably very little business case to make a new “Domino v2” pin/backward compatible with the “Domino v1” unless there are many thousands of customers that have used Domino v1 in something other than MiFi.

Like you said there are no sales # one way or the other so only GL knows. I would suspect it is the opposite. GL when they launched Domino had huge development effort, lots of resources and several other products/host boards designed around Domino. All have since seemingly been either discontinued or abandoned/not available for purchase anymore. Perhaps they sold millions and want to keep it under wraps, but I suspect the truth is they produced many Dominos, found the uptake was not what they had hoped for, and decided to use them to produce MiFi instead.

Think about it, if Domino was such huge success, it would be inside all of the other mini routers (or at least more than 1…). Unfortunately that fact also contradicts the modularity argument. Clearly from GL’s perspective it is far more efficient/cost effective/easier/whatever to make inidividual boards based around chips… otherwise every other mini router they sell would have the core/host board combination that is seen in the current MiFi… right?

Don’t get me wrong I am all for modular design and re-usable components/etc I’m just saying from GL’s perspective there is probably very little business justification to make a “domino v2” that would be directly pin compatible with the existing one. Of course that’s just my opinion as an outside observer ;o)

For me its either way really.

I just wanted them to have a second option to consider, its up to them to decide what they do with it :smiley:

And yeah you are right, we do have the module we can reuse so no big deal.

If they don’t have a module bay like the current one, i would not buy it though. :stuck_out_tongue:

I am very happy that there will be a new mifi. For me it’s even better that it will be a completely new album. If you can give a hint, please add options for PoE. Such a device will be able to act as a complete CPE for LTE. From the user it will only require inserting into a waterproof casing.

One thing, even the core board has FCC the final product need to get FCC approval as well.

Making a core board, we need to sell a lot. It also takes a lot of effort as well. Making another board without core board may save some effort. That is two products vs one products. Not too much will be saved in certificates.

I’d rather to redesign the mifi totally:
Add a small screen
Easy exchange battery
Monitor battery from openwrt
Easier expend the interface

Is this still happening?
I’m waiting to a new router with m.2/mPCI connector and USB-powered.

1 Like

Yes. Redesigning the mifi


Are there plans to keep a large number of GPIOs broken out/available for use on new MiFi? For us this is one of the main benefits of the product as it allows us to wire into relays or inputs without the need for additional MCU/components.

1 Like

I understood. Yes we will try.

But can you just use current mifi for that? Current mifi is best for GPIO applications.

We are using current MiFi for this already. My concern is that new MiFi v3 when released will replace current MiFi, and if similar GPIO access is not available on the new unit then we will be stuck adding more components and completely redesigning things on our side.