[sdiy] IC socket reliability/0.100 headers?

[rburnett at richieburnett.co.uk]

> I guess it's fair to say that MTA-type 0.100 style connectors can
> suffer from many of the disadvantages of IC sockets? They do make 
> life
> easy and neat, but I guess a soldered connection to a PCB pin with
> heastshrink for added mechanical strength is possibly the most
> reliable.

I tend to avoid connectors where possible, but it does depend what you 
are feeding through them.  I see less trouble with power or HV 
connectors where there is enough current flow on the supply rails to 
prevent contact degradation, and the power being passed is typically 
decoupled locally afterwards.  Small audio and weak RF signals seem to 
be most susceptible to contact degradation and non-ohmic contacts.

The worst problem I saw with a multi-way connector was whilst doing 
some consultancy for a company about 10 years back.  They made a 
portable measurement instrument that had a simple on board computer.  
Someone made the design decision to pass a portion of the data and 
address busses from one board mounted on the front panel to another PCB 
mounted on the bottom via fine-pitch multi-way connectors.  They used 
good quality gold-plated connectors from a well regarded connector 
manufacturer and it worked fine in the lab, but after 6 months of 
real-life service the units were being returned for spontaneous reboots 
and lost data when the instrument was picked up and plonked down on 
another table, moved from a cold van into a warm lab, etc, etc.  The 
moral of this story is don't pass signals as sensitive as a high-speed 
data/address bus through mechanical connectors in a portable appliance 
that experiences rough handling and harsh environmental changes unless 
you literally have no other choice!  And if you do have no other choice, 
then make sure any mechanical slop doesn't have to be taken up by the 
connector itself!

-Richie,