[Arm-netbook] EOMA68 interface update proposal: -GbE, +USB3.1, +GPIO

[Luke Kenneth Casson Leighton]

ok, soo...

i've been following the updates to USB3 (10 gigabit speeds now with
USB 3.1). in combination with:

* the development of both the tablet and the laptop not having room
for an ethernet port
* the development of lower-cost SoCs not really having ethernet
* the fact that the ethernet components cost around $2 when the SoCs
themselves are around $2 to $4

... i'm coming to the conclusion that Ethernet should be removed, and
the 9 lines used to

* increase the USB3.0 allocation (4 wires) to USB 3.1 (8 wires)
* this would leave 5 spare pins for GPIO.
* at least two of those should be EINT-capable GPIOs

this would make it possible to redesign the tablet PCB to a more
"standard" one - removing the need for the I2C GPIO chip for example.

if there are any products that need ethernet, it can be done as a USB2
or USB3 compliant external IC.

the IC3128 PCB can be reduced down to around a $12 BOM, the EOMA68-A20
PCB can be reduced by a further $2.50.

any thoughts or objections, much appreciated the input.

l.

_______________________________________________
arm-netbook mailing list arm-***@lists.phcomp.co.uk
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
Send large

[Hrvoje Lasic]

Depend really on your target market. For any application that your product
need to be any kind of server ethernet is kind of better, more relaible
compared to wifi.

If your target audience is not that kind of application, or at least not in
majority I guess you got it right. Just question any kind of paradigm.

Another angle is that at one moment you have to stop developing and ship
product, even if not perfect. Again, you decide.

Post by Luke Kenneth Casson Leighton
ok, soo...
i've been following the updates to USB3 (10 gigabit speeds now with
* the development of both the tablet and the laptop not having room
for an ethernet port
* the development of lower-cost SoCs not really having ethernet
* the fact that the ethernet components cost around $2 when the SoCs
themselves are around $2 to $4
... i'm coming to the conclusion that Ethernet should be removed, and
the 9 lines used to
* increase the USB3.0 allocation (4 wires) to USB 3.1 (8 wires)
* this would leave 5 spare pins for GPIO.
* at least two of those should be EINT-capable GPIOs
this would make it possible to redesign the tablet PCB to a more
"standard" one - removing the need for the I2C GPIO chip for example.
if there are any products that need ethernet, it can be done as a USB2
or USB3 compliant external IC.
the IC3128 PCB can be reduced down to around a $12 BOM, the EOMA68-A20
PCB can be reduced by a further $2.50.
any thoughts or objections, much appreciated the input.
l.
_______________________________________________
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook

--
Hrvoje Lasić
Vulpes d.o.o.
Gračanska 120a
10000 Zagreb
Croatia
tel [phone]
tel [phone]


****@gmail.com <***@gmail.com>***@vebbu.co <***@vebbu.co>*

*www.vebbu.co <http://www.vebbu.co>*

[Luke Kenneth Casson Leighton]

Post by Hrvoje Lasic
Depend really on your target market. For any application that your product
need to be any kind of server ethernet is kind of better, more relaible
compared to wifi.

well, USB3 ethernet is going to be reliable (and fast). even USB 3.0
is 5gbit/sec and GbE is 1gb/sec. and i had to take off SATA because
it was adding extra cost.

Post by Hrvoje Lasic
If your target audience is not that kind of application, or at least not in
majority I guess you got it right. Just question any kind of paradigm.

it's not the majority, for sure.

Post by Hrvoje Lasic
Another angle is that at one moment you have to stop developing and ship
product, even if not perfect. Again, you decide.

yehh very true :)

Post by Hrvoje Lasic

Post by Luke Kenneth Casson Leighton
ok, soo...
i've been following the updates to USB3 (10 gigabit speeds now with
* the development of both the tablet and the laptop not having room
for an ethernet port
* the development of lower-cost SoCs not really having ethernet
* the fact that the ethernet components cost around $2 when the SoCs
themselves are around $2 to $4
... i'm coming to the conclusion that Ethernet should be removed, and
the 9 lines used to
* increase the USB3.0 allocation (4 wires) to USB 3.1 (8 wires)
* this would leave 5 spare pins for GPIO.
* at least two of those should be EINT-capable GPIOs
this would make it possible to redesign the tablet PCB to a more
"standard" one - removing the need for the I2C GPIO chip for example.
if there are any products that need ethernet, it can be done as a USB2
or USB3 compliant external IC.
the IC3128 PCB can be reduced down to around a $12 BOM, the EOMA68-A20
PCB can be reduced by a further $2.50.
any thoughts or objections, much appreciated the input.
l.
_______________________________________________
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook

--
Hrvoje Lasić
Vulpes d.o.o.
Gračanska 120a
10000 Zagreb
Croatia
tel [phone]
tel [phone]
www.vebbu.co
_______________________________________________
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook

_______________________________________________
arm-netbook mailing list arm-***@lists.phcomp.co.uk
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
Send large attachments to arm-***@files.ph

[Lauri Kasanen]

On Thu, 3 Sep 2015 15:09:14 +0100

Post by Luke Kenneth Casson Leighton

Post by Hrvoje Lasic
Depend really on your target market. For any application that your product
need to be any kind of server ethernet is kind of better, more relaible
compared to wifi.

well, USB3 ethernet is going to be reliable (and fast). even USB 3.0
is 5gbit/sec and GbE is 1gb/sec. and i had to take off SATA because
it was adding extra cost.

I'll believe when I see it. So far seen zero reliable USB ethernet
adapters.

The lack of native ethernet and sata puts me out, but perhaps I
represent the small minority - get a product out, and then work on
revisions for the curmudgeons ;)

- Lauri

_______________________________________________
arm-netbook mailing list arm-***@lists.phcomp.co.uk
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
Send large attachments to arm-***@files.p

[Luke Kenneth Casson Leighton]

Post by Lauri Kasanen
On Thu, 3 Sep 2015 15:09:14 +0100

Post by Luke Kenneth Casson Leighton

Post by Hrvoje Lasic
Depend really on your target market. For any application that your product
need to be any kind of server ethernet is kind of better, more relaible
compared to wifi.

well, USB3 ethernet is going to be reliable (and fast). even USB 3.0
is 5gbit/sec and GbE is 1gb/sec. and i had to take off SATA because
it was adding extra cost.

I'll believe when I see it. So far seen zero reliable USB ethernet
adapters.

USB2 or USB3? USB2 i am keenly aware has been price-pushed so low
it's really quite ridiculous. the end-result is that, yes, USB2 SATA
and USB2 Eth dongles are... fun. there was word a few years ago that
the Genesys Logic USB2-SATA IC needed a "hard reset" on a regular
basis, it would crash so badly.

but, honestly, USB2 is only 480mbit/sec - trying to saturate that
with either GbE or even 30% of SATA-II bandwidth is... not going to
end well, given that USB2 is a "cooperative" bus.

USB3 i have much more hope for.

Post by Lauri Kasanen
The lack of native ethernet and sata puts me out, but perhaps I
represent the small minority - get a product out, and then work on
revisions for the curmudgeons ;)

well EOMA68 is designed to be a decade-long standard. so it's
all-or-nothing, here. i'd need to work on a totally different
standard, and a totally different package. well, that's the plan,
anyway :)

l.

_______________________________________________
arm-netbook mailing list arm-***@lists.phcomp.co.uk
http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
Send large attachments to