kicad-developers team mailing list archive

[Alex G <mr.nuke.me@xxxxxxxxx>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 09/01/2010 04:49 PM, Lorenzo Marcantonio wrote:
> On Wed, 1 Sep 2010, Alex G wrote:
> 
>> Thanks!!! This should prove very useful. I like their idea with the RMS
>> sum for tolerances.
> 
> It's actually a rule from statistic... summing the whole tolerances is
> the `worst case' method. If (and only then) the tolerance have gaussian
> distribution (and since we're talking about manufacturing thats probably
> true) you can take their RSS (Root Squared Sum) instead, and still have
> a representative value. Google for 'summing tolerances' to get copious
> explanations. In other word, they simply applyed the usual rules :D
> 
I'm (partially) familiar with statistics. I took a year-long
college-level course in high-school on the matter. And then there's the
common RMS measurements for AC.
I like the idea of applying this idea to PCB design. :)

>> If I had the standard, I would quote the tables (with proper references
>> of course). No, I'm not asking you to do it; I already have that info in
>> LP Calculator. In fact, I'm not even suggesting it. It's just my way of
>> saying I live on the dangerous edge of life. :-p
> 
> Also, they are biggish (look at the list of the pin type, each pin has
> a table for the three manufacturing densities with all the fillet
> targets and courtyard excess) and easy to misquote. Since they would be
> hand copied anyway I just offered a `reliable source'. And they copied
> the `round to 0.05, 0.10, 0.15 and so on' explanation in *each table* as
> `roundoff factor'.
>
And your source is very well appreciated. :)
Don't worry, if it comes to it, I will quote the tables from LP
calculator. Your help is already immense and of immeasurable value.

> I'm looking in the rest of the standard for juicy bits... it's mostly
> stuff like: QFN should survive reflowing at least 5 time, blah blah, or
> what differ in etching if you use phototools vs laser engraving and
> liquid resist vs dry resist, fixturing for panelization and so on (i.e.
> very low level stuff for the people actually fabricating boards).
> 
> Oh, and *vital* information like: Tape peel strength [they're talking
> about components reel] shall be 40 grams +- 30 grams (/me imagines
> a Q.A. technician measuring with a dynamometer the peeling strength of
> one reel :D:D)
> 
Ya' know what. For the stuff that I do, a 650W Bosch drill with scotch
tape around the smaller drill bits does the job. The only thing that I
would be crazy enough to measure its peeling strength with a dynamometer
is too inappropriate to discuss on the list.
Anyway, the bits relevant for kicad are the footprints, and a footprint
calculator. That will allow kicad to penetrate deeper (no pun intended)
into the professional market.

> And of course there are IPC-7251 (i.e. through hole components, we need
> it to replicate the corresponding function in the LP Calculator) and
> IPC-2222 (mixed wisdom about overall PCB design, layer stacking and so
> on -- including test coupons in panels and using some little resistor
> clips if you need a board for the space shuttle :P).
> 
Remember those headaches I mentioned earlier?
(This is by no means an "I don't care"; it's just that I have too much
stuff going on right now to dedicate CPU time to processing it).

>> I say you should include a schematic for a GPS guided missile so that
>> someone may send a free copy of your work to whatever entity wrote the
>> standard (non-destructively, of course). :-p
> 
> I suggest to send it to Mentor Graphics which acquired PCB Matrix (the
> developer of the calculator). Or, also, to IPC itself which in the back
> cover essentially says:
> 
By the way, you should probably license your quick intro under the GFDL
before we do that.

> 'You have to pay because we spent a lot of time researching it and
> partecipating meetings with a lot of red tape with ANSI and so on'
>
"And you can just imagine how dull a meeting looks without refreshments,
wine, beer, and bikini girls"

> They don't even offer an `upgrade license' for the owner of the old
> standard, you have to pay it fully... (you only have 50% off if you have
> *membership* with IPC, I don't dare to ask what they want each year:D)
> 
Or whether they charge their membership in money or one of the last
items I mentioned earlier.

Alex

P.S. Preemptive apologies to anyone who may find this email offensive.
The truth of the matter is, I can only only be furious or sarcastic at
those who make exaggerate profit from us common people. (And honestly, I
think I found the most hurtful sarcasm for those old guys trying to rip
us off).
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
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=DndZ
-----END PGP SIGNATURE-----