[Cryptech Tech] PCB prototyping machine [Was: BCP prototyping machine]

[Warren Kumari]

On Wed, Feb 25, 2015 at 2:59 PM, Павел Шатов <meisterpaul1 at yandex.ru> wrote:
> On 25.02.2015 19:05, Warren Kumari wrote:
>>
>> On Wed, Feb 25, 2015 at 2:35 AM, Павел Шатов <meisterpaul1 at yandex.ru>
>> wrote:
>>>
>>>
>>>
>>> Yes, that is what I suggest. Have you seen my suggestion to use CPU+RAM
>>> module from http://www.imx6rex.com/ on the list? This module can be the
>>> core
>>> module of our Alpha Board, that end users are not supposed to modify.
>>
>>
>>
>> I'm a little uncomfortable with the "that end users are not suppose to
>> modify" wording there.
>> I'm assuming that you are just meaning just for the Alpha Board, and
>> that this is simply a ready made, convenience option? And that it
>> would be replaced with a non-prebuilt solution later in the project?
>>
>> I realize I may be being overly picky, but use of the words "are not
>> supposed to modify" and suggestions to buy a ready made widget with
>> all sorts of built in magic sets of alarm bells.
>
>
> Warren, let me explain this. "end users are not supposed to modify" means "I
> doubt that anyone will ever want to re-design a 12-layer PCB with DDR-3
> memory interface".
>
> The core of our Alpha Board is CPU + FPGA bundle. Having a CPU implies
> having some RAM for it too. In case of i.MX CPU (which is used in Novena and
> which we are going to use in Alpha Board) we need DDR3-1066 memory, so we in
> fact need to develop a board with CPU + RAM + FPGA. Such a board requires
> very much effort to layout. That i.MX module that I suggested to use is open
> source. It contains the CPU and RAM on one small board, so we will only need
> to develop baseboard for this module with FPGA of our choice. They have
> schematic, layout and all the necessary information on their website. We can
> download it and inspect for anything suspicious. If we don't trust them, we
> can use their layout to order boards from company X, their bill of materials
> to buy components from company Y and then ask company Z to solder them for
> us.

Cool.

I've already told that our baseboard (carrier) for this module can have
> expansion connectors, so end users may actually develop their own extension
> modules and connect them to our system. But I don't actually think that
> anyone will want to re-design the CPU+RAM core module though.

Quite possibly people won't redesign the CPU+RAM core module, but I
want to make it clear that, in the final design implementers have the
*freedom to do so if they choose to*.

Mainly I'm trying to avoid us putting in this time and effort, and
then at the end have people who haven't been following the project
(and don't know our history/ culture) turn around and point at the
imx6rex board and say "Ah, so *that's* where the backdoor gets added.
No wonder end users are not supposed to modify this bit...".
Nothing that (IMO) we need to change on a technical level, rather just
pointing out that people who were not involved in the project (and
some who may even be hostile to the project) may at some point look
through the list archives and, unless we are very careful,
misinterpret what is being said.

> You can get a
> basic idea of required amount of effort to do this here:
> http://www.imx6rex.com/2013/11/how-long-it-took-to-design-the-imx6-rex-module-prototype/

I personally wouldn't want to redo all that (nor, frankly do I have
the skill-set to!), but my tinfoil hat may be thinner than others.

Don't get me wrong, I'm not saying that we shouldn't (or should) use
this module, just keeping the messaging in mind...

W

>
>
> --
> With best regards,
> Pavel Shatov
>
>
>
>
>



-- 
I don't think the execution is relevant when it was obviously a bad
idea in the first place.
This is like putting rabid weasels in your pants, and later expressing
regret at having chosen those particular rabid weasels and that pair
of pants.
   ---maf