Ralph Hersey wrote:
> 
> Mail*Link(r) SMTP               FWD>DES:ISOLATION
> 
> Jack inquired about differences in high-voltage (electrical spacing) isolation
> requirements between components and printed boards.
> 
> Date: 5/30/96 10:47 AM
> From: Jack Olson
> 
> >I'm not sure if this is the best place to ask electronics engineering
> >type questions; so if anyone knows of an engineering-related forum,
> >please let me know...
> >
> >I am in the process of designing a power board to military specs, and
> >one of the requirements is "2500V isolation between primary and
> >secondary". Using the formula that says 0.12 mils/volt, I am looking
> >at a 300mil clearance wherever appropriate, right?
> 
> You're correct for a type "B4" permanent polymeric coated printed board and
> for a type "A7" conformally coated printed board assembly per Table 3-1 in
> IPC's-D-275.
> 
> >One of the components in the design is an opto-isolator in a DIP8
> >package, and the data sheet guarantees 3000V insulation between
> >primary and secondary. Since it is packaged in a DIP, and I am
> >required to use 65 mil pads, I am only left with 235 mils between
> >primary and secondary, which works out to (235/0.12) 1958.3 volts.
> >
> >Am I misinterpreting the requirements?
> 
> IMO - No
> 
> >Is their 3000V claim based on a different dielectric or something
> >internal to the component? If so, how is that meaningful knowing the
> >DIP package must be mounted to something with a lower value?
> 
> IMO, it's the differences in the requirements and ratings between component
> manufacturers and printed boards.
> 
> Your component is "guaranteed 3 kV" but I suspect it's not "rated" for
> continuous operation at that voltage.  I'm not sure about you MIL SPEC
> requirement for 2500 V, but I suspect it is for some limited time duration, I
> suspect, typically in the range of 3-10 seconds.
> 
> To my knowledge, the component manufacturer's requirements use essentially
> pristine "new" parts, not remembering for sure, but off-hand, the components
> may??? be tested for isolation after some humidity aging, then again I suspect
> maybe not.
> 
> A quick check of one of the optical isolator application manuals gets out of
> very easily, their "rated/guaranteed" is 2.5 kV for up to 4 seconds, then in
> the application notes, they cop-out and state the continuous voltage depends
> on the cleanliness, humidity, etc of the application and you should
> test/qualify to your environment.
> 
> The D-275 requirements are based on of long-term environmental testing in all
> kinds of environments, salt air, desert, altitude, dust, dirt and atmospheric
> pollutants.
> 
> >Is the 0.12 mils/volt spec over-estimated to be on the "safe" side?
> 
> Yes it's very conservative to ensure a very low risk.
> 
> >Are the terms "insulation" and "isolation" interchangeable?
> 
> IMO no, without going to the dictionary, insulation is generally used with a
> term like resistance, meaning that ites not perfect, there are some
> losses/conduction.  Isolation generally means that something on one side of
> the isolator has no effect on something on the other side of the isolator.  In
> some ways, an isolator would be a perfect insulator for something.
> 
> >Can I use this component (and/or) Am I splitting hairs?
> 
> IMO, yes, I don't think you're splitting hairs, but:
> 
> *  You just need to keep as much separation (isolation) as you can between the
> land patterns, then use the increased spacing for the remaining condutor
> spacing.
> *  Make sure the printed board manufacturer "really cleans" the surface of the
> printed board prior to the application of solder resist (applicable)
> *  Make sure the printed board doesn't get dirtied up during assembly due to
> handling
> *  Make sure assembly residues (low residue fluxes or cleaning residues) are
> suitable for the application and don't create future paths for electrical
> leakage
> *  I have no experience with low residue fluxes and high voltages, so from the
> old school, clean it really good, then conformally coat with a good adherent
> coating.
> 
> >thanks in advance,                                     Jack
> 
> Hope this helps
> Ralph Hersey
> e-mail:  [log in to unmask]
> 
> ------------------ RFC822 Header Follows ------------------
> snip-sniped it all_____________________________________________________________________

Good evening all:

Assuming that the component is suitable for the specified application 
and that the pad to pad spacing between the primary and secondary is 
substandard, the board can be designed with an "air gap" between the 
primary and secondary pads.  This is accomplished by cutting a 
non-plated slot between the pads.  Narrow (.031") slots have been used 
to increase the effective separation between adjacent pads.  Wider slots 
have been used between opposed rows of DIPs to increase the effective 
separation.  This slot can be whatever width your design can tolerate 
but usually .062 inch to minimize costs. In cases where the slots would 
weaken the board, zig-zagged slots can be used.

 P   P   P         P   P   P           P   P   P
zzzzzzzzzzzz     zzzzzzzzzzzzzzz     zzzzzzzzzzzzzzzz
           z                   z                    z
           z                   z                    z
           zzzzzzzzzzzzzzz     zzzzzzzzzzzzzzzz     zzzzzzzzzzzzzz
 P   P   P         P   P   P           P   P   P

                   where z = slot & P=  IC pad

In this manner, the surface distance would be measured by following the 
shortest distance without crossing a slot.  The air gap across the slot 
is generally sufficient to create an acceptable isolation gap.

Sorry for the length of this explanation.  If I haven't been clear 
enough, e-mail me your FAX number and I'll send your a pencil sketch of 
the configuration.

Don Vischulis
[log in to unmask]