Fwd: [onerng talk] BOM Information

[Jim Cheetham]

---------- Forwarded message ----------
From: Paul Campbell <taniwha at gmail.com>
Date: Sat, Jan 3, 2015 at 4:15 PM
Subject: Re: [onerng talk] BOM Information
To: jim at gonzul.net


(this one to the list also failed ... could you please repost it for me)

On Thu, 01 Jan 2015 08:34:50 CitizenX wrote:
> What's up with this spam detection?

I'm not sure - I was in Hanoi using a local SIM card, not sure why my mail
server is doing that on outgoing mail, maybe the local cell network is spam
listed - weird - I didn't think it did that on outgoing mail - hopefully this
one through a  hotel wifi is better

(for those who don't know I'm on a family vacation for the next week or so)

> OK, here's a good sampling of parts with insufficient information for exact
> (I'm talking suppliers and identical part feature replication) BOM
> replication.
>
> Ferrite bead (Lots of unknown's here)

actually  just about any generic ferrite will do

> The resistors with the 4k7 values (I assume thats 47K?)

no 4k7 - what's there is a tweaked analog circuit trying to get the noise duty
cycle to as close to 50/50 as possible with off the shelf resistor values
(hence the 1%) - the values are essentially designed to be sampled by that
particular CC2531 GPIO port - you'd have to tweak the values to match another
microcontroller or digital logic interface with different real-world sampling
thresholds

> The oscillator (Part #)

32MHz xtal - you don't need the fine tolerance that zigbee requires

> The transistors (Part # would be great)

2N3904

> The LED

again, not critical, any cheap 0603 LED will do

> The inductor (ESR, amperage level, etc)

not that important, the actual current is miniscule

> No data on caps, type, dielectric used, etc.

again not that important - standard ceramics work great

-----------------------------------------

Note in general there is too much decoupling here (which is why I'm being a
bit blase here about caps/etc) there's a lot of  'belt-and-braces' engineering
in here - more caps and decoupling than is probably really needed

What is important here is layout - you're making a noise source, something you
normally try and avoid doing in any electronic design, you want to protect the
circuits around it from the noise generator, and the noise generator from the
circuits around it  - you also want to reduce radiation (so you're not giving
information away, and to keep the FCC happy)  and stop any potential external
interference from potential attackers - so the big deal here (and why I didn't
try and make it any smaller) is that the entire noise generator portion of the
design, and its power supply is laid out on 1 layer with a solid ground plane
(stitched to the shield) underneath. You're also constrained in size by the RF
noise antenna which limits the length of the board (though arguably since it's
a noise antenna rather than a signal one could have a more 'unique' design)

Also as Jim points out we do want our design to be visually inspected.

You're right though I do need to publish a full BOM, I'll probably wait until
we have a formal one for long term manufacturing

        Paul