William Chops Westfield wrote:
> Programming parts IS pretty expensive.  With most modern parts,
> the programming time is too short to go off and do something else
> while it's happening, so you're looking at someone dedicating time
> to programming parts.  At the US minimum wage, you'd have to program
> 73 chips an hour to equal the 0.07 charge you mentioned from microchip.
> (and you don't want a minimum wage person programming 10f chips!)
> And then there's the cost of the programming HW... I think you MIGHT
> be able to get that rate, but the Microchip charge seems pretty
> reasonable...

Programming a PIC in circuit isn't done by itself.  The board has to be
tested, possibly calibrated, etc.  If you assume that the product is going
to go thru some test procedure with a custom test jig (as any volume product
would), you should compare the $.07 charge against the incremental cost of
programming the PIC in circuit during production test.

We've been envolved with a number of 10F designs that are programmed as part
of the production test procedure.  We found that 6 seconds is a reasonable
figure for the incremental test time to program a 10F.  The actual
programming takes less, but there is time to mess with the power supplies,
possibly click some relays to mux this or that, etc.  If this 6 seconds cost
$.07, the your production test cost is $.70/minute or $42/hour.  This is
very high, even considering the amortized cost of the equipment over time,
the space, the total burdened employee cost, etc.  It may be a reasonable
tradeoff for a low to midrange volue product where the cost of the test
station dominates the production testing cost.

Let's say you are building a production test fixture anyway.  The
incremental cost to add ICSP capability is $300 for a ProProg, and let's say
another $1000 in engineer time for the *incremental* design work to include
it in the system, the added software, etc.  Let's say the incremental 6
seconds actually costs you $.04 (reasonable for domestic production).  That
means you save $.03 on each unit over having Microchip program the 10Fs.
This means your purely $$ break even is 43K units.  Then you have to decide
what it's worth to you to be able to quickly put revised firmware into new
units and to possibly retrofit existing finished goods with updated firmware
in case a bug is found.

So it seems Microchip's offer is worth it for low to medium quantity
lifetime volumes of a few 10K units.  Of course you wouldn't likely be doing
anywhere near those quantities here if $.05/unit matters one way or the
other.  Rounded to the nearest $.01, the extra 6 seconds in China cost $.00.
Now the payback is $.07/unit, so it takes 19K units for break even on pure
$$ terms.  Again the flexibility of changing firmware quickly may well be
worth far more than that, but that is highly product dependent.

All in all, I think Microchip's price is reasonable for what it is, although
its not the right answer for most products.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist