Jack; I read your questions and some of the responses and have a few "experience"-based rather than data-based ideas about your question.

The answer depends somewhat on two things:  1) how many layers are you talking about; 2) how thick of a panel for this number of layers.

In the past we frequently found that higher layer-count boards, particularly thinner ones, with pads on every signal layer (this assumes one each power and ground planes, rather than reference planes) would create pad "float" because the stacked pads would be high pressure points where the full hydraulic pressure would be applied during the early part of the multilayer lamination cycle before the resin would squeeze out enough to equalize the pressure across the panel.  A good example of this was a 12-layer .053" thick 15" x 15" board I used to build moderate quantities of.  Layer-to-layer pad registration would sometimes be poor as a result.  This would also be particularly true if the flow pattern (that panel area outside of the board) was designed to "nest" rather than stack; dots on one layer that would fit into a space on an adjacent layer worked better for overall stability but would allow pads to move around more.

This has been reduced somewhat by the advent of vacuum-assisted lamination, but still if you would "guesstimate" the pressure on a few hundred to few thousand small pads when measured across the surface area of an 18" x 24" panel you can imagine what kind of pressure would be applied even if the lamination pressure was only 150-200 psi - there just aren't that many square inches of pad.  Therefore people began to remove some of the unused pads...

However, there are some drawbacks to this as well.  The less copper on a layer, the less likely that layer is to remain dimensionally stable, especially when compared to those power and grounds.  This means that once again your layer-to-layer registration can suffer (the misregistration will be more uniform, however I doubt that that's a lot of help).  In addition, if the layers are thin, small amounts of copper surface area mean more difficulty in handling the layers without some type of damage (frequently this means small cracks in the material, which sometimes aren't readily visible).

Another EXTREMELY good reason for leaving pads in place is hole reliability.  These interconnects act as "anchors" which can reduce the Z-axis expansion of the barrel of the hole during severe thermal cycling, which can reduce the incidence of barrel cracking or other catastrophic defects.  I know that some fabricators like the idea of reducing the number of interconnects (by eliminating those "extra" pads), with the reasoning of reducing the chances of post-separation interconnect defects, but most shops build most boards without losing too much sleep over ICDs.  

Therefore (finally) in most cases I would vote to leave the pads in place.  As always, there are bound to be exceptions to the rule, but in most cases I think it's to my advantage to have the pads on most layers.   But as the man says, these opinions are mine and mine only, and don't necessarily respresent the views of this station, etc. etc. etc.

Regards,
FJ

>>> Jack Olson <[log in to unmask]> 05/07/99 02:29PM >>>
It seems that most fabrication vendors prefer to remove the unconnected pads
from inner-layers, in fact we do it for them before we send out the gerbers.

But I thought I overheard Gary Ferrari say there was some test results that
showed the board to be more reliable with all pads included, but I don't
know where to find it or who did the testing.

Does anyone know if it is significant enough to re-evaluate our process?

Jack

################################################################
TechNet E-Mail Forum provided as a free service by IPC using LISTSERV 1.8c
################################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in the body:
To subscribe:   SUBSCRIBE TechNet <your full name>
To unsubscribe:   SIGNOFF TechNet 
################################################################
Please visit IPC's web site (http://www.ipc.org) "On-Line Services" section for additional information.
For technical support contact Hugo Scaramuzza at [log in to unmask] or 847-509-9700 ext.312
################################################################

################################################################
TechNet E-Mail Forum provided as a free service by IPC using LISTSERV 1.8c
################################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in the body:
To subscribe:   SUBSCRIBE TechNet <your full name>
To unsubscribe:   SIGNOFF TechNet
################################################################
Please visit IPC's web site (http://www.ipc.org) "On-Line Services" section for additional information.
For technical support contact Hugo Scaramuzza at [log in to unmask] or 847-509-9700 ext.312
################################################################

################################################################
TechNet E-Mail Forum provided as a free service by IPC using LISTSERV 1.8c
################################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in the body:
To subscribe:   SUBSCRIBE TechNet <your full name>
To unsubscribe:   SIGNOFF TechNet 
################################################################
Please visit IPC's web site (http://www.ipc.org) "On-Line Services" section for additional information.
For technical support contact Hugo Scaramuzza at [log in to unmask] or 847-509-9700 ext.312
################################################################