[NLRS] LFCSP (was re:power measurement)

[Ed Marciniak]

Maybe someone was trying to do you a favor. Use a photocopier to reduce the 
eval board artwork and print onto PnP Blue film.

With cold ammonium persulfate, I've managed 5 mil airgaps and near vertical 
slopes on sides.

Perhaps I'm a bit biased after looking at 1 mil bond wires and bare chips.


----- Original Message ----- 
From: "Gerald" <geraldj at ispwest.com>
To: <nlrs at mailman.qth.net>
Sent: Friday, February 10, 2006 7:05 PM
Subject: Re: [NLRS] LFCSP (was re:power measurement)


>
>
> On Fri, 2006-02-10 at 16:59 -0600, Ford Peterson wrote:
>>
>> Chris moaned...
>>
>> ...snip...
>>
>> > And these are indeed very cool...  but how on earth do you work with
>> > them at home given the packaging?  The AD8317 (1 MHz to 10 GHz) Log
>> > Detector is in a 2mm x 3mm 8-Lead LFCSP package.  Geez!  First you
>> > gotta make a board and then you have to figure out how to solder
>> > it down.
>> >
>> > I just ordered samples of a cool ADI synthesizer (ADF4360-6) that will
>> > let me generate 1152 MHz from a 10 KHz GPS reference directly--  and
>> > I failed to pay close enough attention to the packaging... and that
>> > one too is in a 4mm x 4mm LFCSP package.  I know I am getting old
>> > (43) but I never had trouble working with stuff at home until I
>> > encountered these beasts.  I had to order the eval board to play.
>> > What looked like a $6 project is now $120.  Geesh.
>> >
>> > So, how do people make use of these things in a hobby context?
>> >
>> > Chris NØJCF
>>
>> I feel your pain Chris.
>>
>> I found one of the kewelest (turned into a cruelest ...snicker...) RF 
>> Gain Block I've ever run into.  The AD8354 is nothing short of awesome. 
>> 1 MHz to 2.7GHz fixed 20dB gain block.  Linear output power up to +4dBm. 
>> In/out matched to 50 ohms.  NF 4.2dB.  Vcc 3v or 5v.  Total supply 
>> current 23mA.  Compare that power to a MAR!  Input power maximum 10dBm. 
>> This thing looks to be awesome.  Digi-key p/n for either the 8354 (20dB) 
>> or the 8353 (25dB) is $1.45.  LFCSP package is new to me.  But the eval 
>> board on the data sheet looks like a 1/2 pitch SOIC8 part.  How tough can 
>> this be?  So I order up a few just to play...
>>
>> What a disappointment!  These are 2mm by 3mm.  8 leads!  4 on each side! 
>> Traces need to be like 0.005"!  The center of each pad is 0.0127" apart! 
>> The eval board graphic in the data sheet must have been blown up by about 
>> 4X!  No way does the graphic characterize just how small this stuff 
>> really is.  You don't have a good feel for it until you pick one up with 
>> the tweezers and see that the needle tweezers shorts out about 3 pins on 
>> each side!
>>
>> I tried super gluing them upside down to get at the pads on the bottom. 
>> (there are no leads)  Completely hopeless!  My smallest iron touches 3 
>> pads at once!  Short of having a gold wire bonding apparatus, I think 
>> this stuff is establishing the limits of homebrewed capability.  I too 
>> would be interested to know how these must be placed in production.
>>
>> I feel your pain OM.
>>
>> Ford-N0FP
>> ford at cmgate.com
>
> They probably are placed by machine over a tiny drop of epoxy to hold
> them until they get to the wave soldering machine. Or the traces may
> have had a bit of fluxed solder powder (called paste) silk screened
> (takes good screens or a machined mask, maybe etched from solid metal)
> and then after the epoxy sets the solder was melted with a jet of hot
> air. Some have reported for ham projects that applying the solder past
> with a syringe and insulin needle then holding the chip down with some
> sort of soft adhesive works. One suggestion was from GB of a product not
> available here. The other suggestion was to use Staples reusable
> adhesive. What I found has part number 10448-US, bar code 7 18108 00907
> 2. Says it conforms to ASTM D-4236.
>
> A week or so ago I started to attack old surface mount boards in a
> Motorola spectra and my tiny iron with a 1/16" diameter tip is still
> big. I may have to take a tip and hit it with the lathe to cut it down
> to .032. Maybe drill a hole in the end and silver solder in a bit of 20
> gauge bus bar and file a flat tip on the bus bar (how ironic that name
> which could apply to 1/4 x 4 copper bar or stiff bare copper wire even
> the size of hair.
>
> One other possibility might be that an adapter board could be made (or
> may exist) to hold the tiny 2 x 4 mm chip and bring out its leads to a
> workable perimeter, say 6 x 8 mm. Probably made of FR-4 but .016 thick
> so it can be laid on a hot plate, the solder paste melted on the chip
> pads and the chip lowered with tweezers.
>
> All these things probably have to be done under a 10 to 20 x stereo
> microscope. I have one, but its old and purely vertical so I'll have to
> create a special low work bench and there's no room for that around here
> right now.
>
> -- 
> 73, Jerry, K0CQ
> Dr. Gerald N. Johnson, electrical engineer
> All content copyright, Dr. Gerald N. Johnson
>
> _______________________________________________
> NLRS mailing list
> NLRS at mailman.qth.net
> http://mailman.qth.net/mailman/listinfo/nlrs
>