[Starlink] teardown of the starlink provided "router"

[Dave Taht]

On Tue, May 18, 2021 at 8:28 AM Nathan Owens <nathan at nathan.io> wrote:
>
> Here are the chips in the dish:
> https://twitter.com/VirtuallyNathan/status/1331709138276024320
> There's a bunch of the black ones, some sort of RF chip, and one of the heat-spreader capped ones. Parts seem to be custom, nothing I could find online.
>
> The Dish acts as the modem, so any router can be used. How were you testing? With their router? Without?

behind their router.

this explains my issues with ipv6, and at least some, with
bufferbloat. We will toss the starlink router over the cliff
soon and start over.
>
>
> On Tue, May 18, 2021 at 8:23 AM Nathan Owens <nathan at nathan.io> wrote:
>>
>> Here ya go: https://fcc.report/FCC-ID/2AWHPR201/4805890
>> There's at least one teardown video, but the FCC pics are better.
>>
>> Qualcomm IPQ4018, can run OpenWrt 19.07.7, or the latest 21.02.1 RC branch.
>>
>> --Nathan
>>
>> On Tue, May 18, 2021 at 8:20 AM Dave Taht <dave.taht at gmail.com> wrote:
>>>
>>> I would really like someone to disassemble, and take photographs of
>>> the chipset and design of the starlink provided "router",
>>> which is proving to be such a disaster, and put those up somewhere.
>>> Youtube videos seem to be popular... we could do a
>>>  kvetchy one of those?
>>>
>>> I have really goofed in that my visit to the starlink site I'd assumed
>>> the dishy was all there was! At one level I'm deliriously happy
>>> that that router can be junked, and with a decent OS, ipv6 etc can be
>>> enabled, as well as many other interesting local
>>> networking services. At another level I'm grumpy as to have to throw
>>> out all the bufferbloat related testing to date.
>>>
>>> In my dream world, the dishy would do ethernet flow control at a
>>> "single transmit upstream buffer" granularity, below 4ms, and
>>> emit pause frames that an fq_codel upstream could deal with, but that
>>> still is not enough, accurate parsing of the dishy's actual link rates
>>> up and down would allow for controlling the sch_cake queues tighter.
>>>
>>> --
>>> Latest Podcast:
>>> https://www.linkedin.com/feed/update/urn:li:activity:6791014284936785920/
>>>
>>> Dave Täht CTO, TekLibre, LLC
>>> _______________________________________________
>>> Starlink mailing list
>>> Starlink at lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/starlink



-- 
Latest Podcast:
https://www.linkedin.com/feed/update/urn:li:activity:6791014284936785920/

Dave Täht CTO, TekLibre, LLC