To socket or not to socket?

[Grant Richter]

Hi All,

Generally accepted engineering practice for sockets
goes something like this:

1. Sockets cost money. The more money you spend
on the socket the longer it will last. 

2. Always fully socket prototypes and the first few boards
built after a revision. This makes troubleshooting your
screwups 1 Googleplex easier.

3. For routine production, socket chips containing
LSI and MSI circuits that are less reliable than
SSI like CD4000, 74 series, and op-amps which
are ridiculously reliable. Socket OTAs because they
are delicate and break easy.

4. Socket any programmable parts and parts containing
software as it will surely change :^)

----------
> From: Magnus Danielson <cfmd at swipnet.se>
> To: batzman at all-electric.com
> Cc: mgm at starlingtech.com; synth-diy at mailhost.bpa.nl
> Subject: Re: To socket or not to socket?
> Date: Saturday, February 05, 2000 12:00 PM
> 
> From: "Batz Goodfortune" <batzman at all-electric.com>
> Subject: Re: To socket or not to socket?
> Date: Sat, 05 Feb 2000 12:19:54 +1000
> 
> > Y-ellow Michael 'n' y'all.
> > 	Sockets do have problems (sometimes) in high shock, high vibration
> > environments. Where the IC can literally be shaken out of it's socket.
Some
> > sockets get dirty due to poor materials and or vibration. Sockets
suffer
> > with expansion and contraction due to temperature as well. This in
> > conjunction with dirt can cause poor connections to the IC after a
time. We
> > use to have this problem with early PCs. The solution was to go round
and
> > press all the ICs firmly back in their sockets.
> 
> You should notice that my experiences are coloured by may days as
repairman of
> PA-system gear where not only transport but also normal operation would
hold
> pretty impressive shock and vibration experiences for electronics. When
your
> electronics is standing on the same wooden floor as a bunch of bass
speakers
> pumping the base-line for larger event, then things will shake seriously,
I
> tell you. I have also seen gear go bananas from that treatment while this
was
> their intended operating environment. What do you guys feel about
> self-oscillating FIR filters? ;)
> 
> But, I also seen what happends when gear becomes 20+ years of age and
then the
> sockets really show their downside. I have this effect box where the
socket
> based interconnects seems to be THE only real sign of age, it makes any
other
> age problem be hidden in the troubles. I have been moving an overhaul of
it
> up for years since I just don't like debugging this type of trouble.
> 
> > However. Most of this can be avoided if you (a) never move your gear.
Or
> > (b) place a dob of silicon adhesive or hot-melt under the chip before
> > pressing it into the socket. This prevents the IC from moving and
because
> > of it's rubbery nature, also provides a kind of suspension support. You
can
> > replace the IC by simply prizing it out of it's socket as usual but it
> > won't possibly move of it's own accord.
> 
> Right, not moving your gear is one reason where sockets should be less of
a
> threat. But there are places where your living conditions still may have
> constant vibrations. For instance, New York area is the 4:th most
earthquake
> active area in the USA. There are a constant set of small earthquakes, so
they
> are barely noticed. These sources are however weaker than a shakey
transport.
> 
> > I have heard RF people complain about sockets causing problems but
that's
> > hardly an issue with synth gear I would think.
> 
> Yes, the reason is simple, you add lead length and thus add inductance.
Also,
> you increase the cross-coupling in both inductive and capacitive sence.
> This is also a reason why sockets are basically band of high speed
digital
> designs, it is much more analog up there.
> 
> > On the whole I use sockets extensively. Especially when running up a
> > prototype. You never know when you might want to exchange that pin
> > compatible OP-amp or something. And as John says. Replacing an IC in a
> > socket is 500% easier if it's in a socket.
> 
> It is sure easier, but I would at least like to point out that one migth
want
> to be a bit picky about which sockets to pick. I rarely use sockets and
> besides, at the few times I need to replace a chip I get to perfect my
> experience of unsolder chips. It's a nuance but it doesn't happend that
often.
> For experimental curcuits sockets play a diffrent role, but once I have
> settled for an IC I would solder it hard.
> 
> Cheers,
> Magnus