Sunteți pe pagina 1din 28

File :

Erstellt :
515
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The Future of the PCB
Arnold Wiemers
Introduction Preliminary information:
The CAD layouters and the assembly developers may practically
receive any PCB quality they demand.
The PCB manufacturers as well do not have to worry: The PCB will
also be in demand in future.
The tendency is evident: The PCB will become more complex and
more universal (figure 1, figure 2). However, already at this point,
the first drop of bitterness casts a cloud over the enthusiasm: The
PCB will also become more sensitive and more expensive.
To explain the technological developments, I would like to give you
an overview of the essential aspects by which the production of
multilayer boards is influenced. This overview will indicate which
variation ranges of materials, galvanic surface-finishes, mechani-
cal treatments and image structures are available. It will reveal in-
dispensable organization principles to be exactly considered, if the
construction and specification of a PCB by a layouter is to be suc-
cessful.
Figure 1:
Development of
the conductive
pattern structures
Year
300
250
200
150
100
50
C
o
n
d
u
c
t
i
v
e

p
a
t
t
e
r
n

s
t
r
u
c
t
u
r
e
s

i
n

m
89 90 91 92 93 94 95 96 97 98 99
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
516
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 2:
Development of
the pad and track
geometries
We do not talk about the really big sensations here, but the stand-
ard products are interesting which will define the daily work of the
layouter, of the PCB manufacturers and of the assembly providers
in one or two years.
These products will be 4, 6, or 8-layer boards as carrier for COB or
MCMs or simply as normal, real PCBs.
One interesting aspect is the indication of the actual possibilites of
the development of 200 m to 150 m, 100 m and finally 50 m-
structures (figures 3 and 4).
The reference is an assumed PCB, a multilayer board with a con-
ductive pattern structure of a track-width and of a track-distance of
150 m.
Due to the current discussions concerning high-technology in the
field of the PCB production, a 150 m-PCB should nowadays be a
common standard product for each PCB manufacturer. The tech-
nical specification of an assembly of such parameters as well
should not cause any problems for the CAD designers.
However, we will see: There are some questions which still require
some answers.
Pad and track geometries (m)
300 200 200 200 100 80
1600 600 500 400 200 100
SIL/DIL SMD Fine-Pitch MCM/COB
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
517
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 3: Carrier
module for COB /
flip-chip
Figure 4:
Switching module
with COB
MFT: Example: flip-chip
Flip-chip on carrier module
80m
MFT: Example: COB-module
Switching module with COB
80m
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
518
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Base materials The appropriate material for an electronic circuit has actually to be
chosen BEFORE starting with the layout.
The discussions regarding electromagnetic compatibility, high-
speed circuits and impedance-control has already been linked with
the technical features of the material assuming that the layouter is
familiar with material parameters like Tg-value (glass transition
temperature), eR-value (dielectric feature) and material composi-
tion (glass fabric, resin content, chemical classification of matter.
The layouter must also know the prices (figure 5).
Figure 5: Base
materials for
PCBs
The PCB manufacturer's logistics has to ensure a stockpiling in
time. This is not a simple task because the variety of material
types, material thicknesses and copper clads may lead to an ex-
ploding stock.
Group Composition Tg
r
Relative
costs
BT
Bismaleinimide triazine resin
with silica glass
180-
220
3.9-
4.9
5.3
CE Cyanate ester with silica glass 230 3.6 4.5
CEM1
Paper phenolic core
with FR4-outer layers
130 4.7 0.95
CEM3
Glass mat (or glass felt) core
with FR4-outer layers
130 5.2 0.95
FR2 Phenolic resin paper 105 4.7 0.73
FR3 Epoxy paper 110 4.9 0.85
FR4 Epoxy glass fibre laminate
135-
170
4.7
1
Reference
FR5
Epoxy glass fibre laminate
with crosslinked resin system
160 4.6 1.4
PD
Polyimide resin with
aramide-reinforcement
260 3.5 6.5
PTFE
Polytetrafluoroethylene
with glass or ceramics
240-
280
2.2-
10.2
32-78
CHn
High-interlaced hydrocarbons
with ceramics
300
4.5-
9.8
90
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
519
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The list of possible inner layer combinations makes this aspect
more obvious (figure 6). However, if these stocks can be reduced
due to future orders, will become more and more doubtful.
Whereas base laminates are durable for years, some prepreg
types can only be stockpiled for several months. If the prepregs are
not used within this period of time, they will have to be eliminated.
Over-stock prepregs would produce a delamination of the multilay-
er board and the product could therefore not be used.
A far more complex task for the PCB manufacturer is to adapt the
production processes to the different characteristics of the base
materials which are based on the altered combination of chemical
basic substances.
It cannot be taken for granted that the contacting of complex multi-
layer levels on FR4 (glass) is of the same quality than on PTFE
(glass/ceramics) or on PD (polyimide + aramide). In particular, if
the frequently very different and decisive resin proportions and
qualities of the materials are also taken into account.
Figure 6:
Laminates for
multilayer board
inner layers
Multilayer board laminates
Laminates are indicated
without Cu-clad
0.050
0.060
0.075
0.100
0.200
0.250
0.360
0.460
0.710
0.930
1.000
1.130
1.430
1.860
1.930
2.330
0.050
0.060
0.100
0.200
0.005
0.009
0.017
0.035
0.070
0.105
Laminate Prepregs Cu-clad
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
520
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
It can also not be taken for granted that these different materials
can be drilled without reducing the quality of the hole walls and
without reducing the reliability of the solder resists to adhere to the
surface.
The fact that base materials for PCBs can be produced in such an
individual form is very much a point in the favour of the PCB man-
ufacturers. The actual solution, however, was found by the part-
ners of the PCB manufacturers, the chemical and material
manufacturers. Their contribution to this success was the most im-
portant one.
Conductive
pattern
structuring
The miniaturization of the components also requires miniaturized
conductive pattern resolutions. The structure exposure therefore be-
comes a considerable challenge.
In the general opinion, if the problem is solved to expose track-
widths and track-distances of 80 m or even 50 m, all subsequent
production steps are inferior.
This is unfortunately NOT the case.
The laminating of the photo laminate (before) and the etching of the
structures (afterwards) belong to the production process of the con-
ductive pattern structuring.
Films (standard), glass master plates (scarcely) and laser machines
(even more rarely) are available as exposure tools. If diazofilms
(copies of the original plot) were used in the past, black films (origi-
nal plots) are now commonly used. The extraordinary quality of
these films allows structures up to 60 m. If this threshold range can-
not be reached, the common hardware (exposure devices) and the
ambient conditions (clean-room conditions) are often the reason for
this.
In case of laser machines, the limit is near 40 m. Decisive for the
better results of the laser exposure, compared with the film, is the
register accuracy to the drilling pattern (optical registration), the fea-
sibility of reproduction (no misalignments), the generally smaller me-
chanical tolerances (reception systems, no vacuum fixing), and the
fact that some production steps are no longer needed (figure 7).
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
521
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The photo laminate has to take up the image structure during the ex-
posure and must completely "store" it. Due to the exposure process,
the molecular structure of the laminate changes - it cures. It is nec-
essary that the curing is effected homogenously over the total thick-
ness of the photo laminate to receive a fine, sharp-contoured
conductive pattern which is also stable at the edges.
An increase in the light energy (scattering) is not sufficient. Thinner
photo laminates with a thickness of 20 up to 25 m (the standard is
38 m) will ensure these demands. Unfortunately, these thin lami-
nates are only rarely available at the moment (figure 8).
Figure 7:
Comparison
between
diazofilm, black
film and laser
direct imaging
Time 24h 10h 3h
Tolerance 0.1mm 0.1mm 0.03mm
Expose conductive pattern G G G
Register film G G
Laminate PCBs G G G
Plate PCBs G G G
Drill PCBs G G G
Retouch diazofilm G
Produce diazofilm G
Measure film G G
Verify film G G
Develop film G
Produce photo plots G G
Documentation G G G
CAM processing G G G
G required
not required
Diazofilm Black film Laser
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
522
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 8:
Comparison
between 38 m
and 20 m
laminate
Finally, the etching process decides, if the structuring of the con-
ductive pattern is successful. This production step is the most un-
popular one for each PCB manufacturer, if the layout demands
tracks and distances smaller than 150 m. The defined application
(spraying) of the liquid etchant on the PCB and the subsequent re-
moval (washing) cannot be carried out so precisely as demanded
even by computer-controlled machines. The losses of 10 up to 30
m regarding structures to be produced are definitly unpleasant in
superfine-line or in super-microfine-line technology.
Holes + vias The drilling of PCBs is THE current subject, i.e, the connection of
selective layers of a multilayer board (figure 9). The discussion in-
dicates the necessity to assign at least a rough classification sys-
tem to the circulating terms.
Functional classes
"Holes" are used for the reception of components or for the fixing
of the future assembly (either in the terminal device or in the as-
sembly machine).
A hole goes through every layer of a PCB.
A hole may take over the function of a via.
Structure exposure
38 m photo laminate
17 m copper clad
100 m inner layers
200ym 200ym
Exposure
Standard
Superfine-line
20 m photo laminate
5 m copper clad
50 m inner layers
100ym 100ym
Exposure
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
523
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 9: Different
via types for
PCBs

Vias ensure the signal flow over several layers of a PCB.


A via connects at least 2, more than 2, but at best all layers of a
PCB.
A via should never take over the function of a hole.
If a via connects ALL layers of a PCB, then it is called a "through-
via".
If a via connects two or more than two layers but not all layers of a
PCB, it is called a "partial via".
There are two variants of partial vias: as "blind via" or as a "buried
via".
Blind vias ALWAYS connect one or several - but not all - inner lay-
ers of a multilayer board to an outer layer.
Buried vias connect 2 or more layers inside the multilayer board,
however, they NEVER produce any contact to an outer layer.
Via types in UTMs for MFT
1.0
0.8
0.6
0.4
0.2
0.0
BS
I2
I3
I4
I5
LS
Via 0.25
up to 0.10 mm
Standard Buried Vias Blind Vias
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
524
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Mechanical classes
The term "micro via" indicates that the mechanical diameter of this
via is considerably smaller than 100 m. The threshold range is ap-
prox. 50 m.
From 100 m on vias are called via.
Production technological classes
"Laser vias" are partial vias which are technically produced by a la-
ser. Two adjacent layers may be connected. This technology can
be applied for common base materials. Through-vias are not pos-
sible.
"Photo vias" are partial vias which are produced by the phototech-
nical structuring and the subsequent galvanotechnical build-up of
the laminate between adjacent layers. The isolation of the individ-
ual, electronically active layers is effected by the laminate. The
possible applications and the stability of this technology are subject
to intensive research.
"Plasma etching" in a plasma atmosphere produces partial vias, if
appropriate materials are used. This technique cannot be applied
for standard base materials.
Through-vias are not possible.
"Micro holes" are mechanically produced as usual. Available drill-
ing tools have diameters up to minimum 0.1 mm. Therefore, both
partial and through-vias can conventionally be drilled.
These technologies are altogether a treasure of alternatives at the
disposal of the PCB manufacturer and among which the CAD de-
signer may have the choice. -
Or so it seems at first sight.
In practice, the via technologies have only the value of the sphere
to which they are related. Regarding blind vias, a rule says that the
via-depth should not exceed the via-diameter. Otherwise, the gal-
vanotechnical connection of the individual signal levels is problem-
atic and not reliable. Due to this reason, the via-depth and the via-
diameter must also be adapted to each other in case of "through-
vias" (relation approx. 6 : 1).
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
525
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
In case of laser vias, photo vias or during plasma etching, only two
adjacent layers may be connected. Necessary vias over more than
2 layers or required through-vias must still be drilled mechanically
so that a combination of technologies becomes necessary.
Without doubt, all connecting processes contribute to a considerable
increase in the density of image structures on the PCB.
Surface-finishes Additional connection technologies regarding the classic soldering
process are nowadays the bonding and - relatively new - the glue-
ing of components on the PCB. For each individual technology,
there are very good solutions.
The requirements, however, increase, if different connection tech-
niques are to be combined on one PCB because the components
are not otherwise available or the reduced space requires such
measures. A variety of surface-finishes with different features are
available (figure 10). Due to SMD components with fine-pitch dis-
tances, plane surface-finishes are required. When choosing a sur-
face-finish, the costs should not be neglected.
The good, old "tin lead" surface-finish cannot be recommended an-
ymore, because the surface is too bended. This surface-finish is
not expensive, it should, however, only be used for THT compo-
nents or in combination with SMD components in a grid distance
not less than 1.27 mm.
"Hot-air-leveling" is still a favourable alternative as well for standard
SMDs (pitch-distance = 1.27 mm), with its process related surface
roughness of 10 - 20 m, however, too undefined.
"Nickel" is an alternative in case of mechanically loaded surface-
finishes, if switching functions are directly guided via the surface.
The soldering behaviour, however, is always temperamental.
"Copper" is generally not worth discussing. The advantage of the
OSP variant "Entek+" is that it is not expensive.
"Immersion tin" offers a plane surface and is inexpensive. The
process can be easily handled within the production, the results,
however, are not completely free of surprises.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
526
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
"Immersion gold" and "galvanic gold" offer a considerably better
quality of the surface-finish.
The gold layer is 0.05 - 0.2 m in case of immersion gold and 1.5
m in case of galvanic gold. The gold is applied on nickel. Both var-
iants are easy to be soldered. Immersion gold is suitable for bond-
ing aluminium wires which are adapted ON the nickel surface-
finish. Gold will then be the corrosion protection.
These surface-finishes are extraordinarily suitable for the glueing
technique, if bright-copper is applied beneath the gold layer be-
cause therefore the surface roughness is improved from > 4m to
< 1 m.
"Immersion bond gold (reductive)" and "galvanic bond gold" are
especially adapted to the bonding technique. The thickness of the
gold layers is 0.3 - 0.6 m in case of "immersion bond gold" and 1
- 2 m in case of "galvanic bond gold".
The bonding is carried out by gold wires IN the gold surface-finish.
This gold as well is applied on nickel.
Figure 10:
Galvanic surface-
finishes for PCBs
In addition to the general definition of galvanic surface-finishes,
there are also several metallizations which mainly take over the
functions of solder depots completely replacing the solder paste
print before the assembly process.
Galvanic surfaces Connection technique Relative
costs Sol-
dering
Bon-
ding
Glu-
eing
Tin lead
+ - -
1.00
Bond gold (immer-
sion)
+
Au
+ +
1.50
Bond gold (galvanic)
+
Au
+ +
3.00
Entek+
+ - -
1.00
Gold (immersion)
+
Al
+ +
1.15
Gold (galvanic)
+
Al
+ +
1.70
Hot-Air leveling
+ - -
1.00
Copper
+ - -
0.90
Nickel
+ - -
1.00
Tin (immersion)
+ - +
0.90
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
527
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Surface-finish combinations, "tin lead" for example with "partial
gilding", are more rarely used at present. The reduced costs do not
justify the the complicated production process, especially in case
of smaller series.
All things considered, the scope of galvanic surface-finishes is suf-
ficient. In the opinion of the PCB manufacturers, the problems can
be found during the further processing in the assembly technology.
The variety of the surface-finishes consequently requires a variety
of assembly preparations and of soldering techniques to be used.
There is still a demand for harmonization.
Lacquers + pastes The lacquers used in the PCB production are wrongly in the shad-
ow of the ubiquitously discussed high-technology.
The solder mask is generally assumed as a kind of solder resist.
Its technical features, however, are considerably more impressing.
The lacquers have a high density, they are resistant to scratches
and have a sparkover voltage up to the range over 100KV/mm.
The dielectric characteristics of these substances are frequently ig-
nored. On an epoxy basis, they have an eR-value of approx. 4.5
which is a constructive quality regarding EMC requirements. The
phototechnical treatment allows a structural resolution with link-
widths about 100 m (figure 11) so that the sensitive spaces of a
400m-pitch of SMD-ICs may be covered which considerably con-
tributes to the prevention of short-circuits.
The limit of the silkscreen is now 1.0 mm for text height-sizes and
180 m for text line widths. An identification print would otherwise
be impossible for many SMD layouts.
Also printable on the PCB is the "peelable solder resist" for protect-
ing zones which are not allowed to absorb tin during the soldering
process, the "via filling mask" for stabilizing the depression during
the in-circuit test and the "carbon conductive lacquer" for the partial
conductivity of the PCB's surface-finish. The "solder paste mask"
is generally used for SMD-PCBs for applying the solder before the
reflow soldering process.
Less known is the print of "resistances" and "capacities" in discrete
values directly on the inner layers of multilayer boards.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
528
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Within permissible tolerances, component and assembly costs are
therefore saved. Moreover, additional space is gained on the outer
layers. The techniques are well-tested, however, there are only eco-
nomical in case of larger series.
Figure 11:
Threshold ranges
for the application
of solder mask
Testability The electronic test of the PCB without assemblies is now a weak
point in the production process since the image structures have
been reduced to 150m and even less.
The reason for this is not the continuously repeated discussion, if
and how it is possible to test against Gerber data. This would be
the second step anyway. The dilemma is revealed in the first step,
the mechanical adaptation of the PCB to ensure the contact be-
tween the PCB pads and the testing machine.
The "pin adapter" has been a good solution for a long time. Each
pad is contacted with the tip of a pin which basis is situated within
the test field of the machine.
The test field contains pins in a grid of 1.27 mm. If the distance
from pad to pad on a PCB smaller than 1.27 mm, the pins will have
to be aligned to the adjacent grid point within the test field.
Solder mask
0
.
6
m
m
100ym
470ym
0
.
3
m
m
1.27mm
0.635mm
50ym
235ym
0
.
2
m
m
50ym
100ym
0.4mm
Standard Superfine-line
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
529
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
In case of a pad distance smaller than 0.8 mm or in case of densely
packed high-performance ICs, the excursion of the pins is so im-
mense that the contact between PCB and test field is interrupted.
Or, the pins are so dense that they touch and therefore provoke in-
correct error-messages.
A pin adapter for a high-dense PCB is really a little piece of art. The
test time of 1 - 2 seconds is unequalled. The production costs of
1.000 - 3.000,- DM are, however, too high for prototypes and small-
er series, and too much time as well, one or several days, is need-
ed for installing the device.
An alternative for the complete and simultaneous test of all con-
nection points is the "translator". The adaptation is effected via a
foil "translating" between test piece and machine. The foil is elec-
trically neutral, it becomes however conductive due the the pres-
sure in the Z-axis.
Since the pads on the PCB are 20 - 30 m above the PCB level,
the required conductive compaction is partially produced. The con-
tact to the testing machine is mechanically established via an inter-
mediate adapter. This method is well-tested, good and quick.
However, it is too complicated and not economical in case of pro-
totypes and smaller series.
An elegant solution would be the "flying-probe" or "finger tester".
One or several motor pairs each control a contact pin and place it
on the PCB pad. The pads to be tested are activated one after the
other and measured against each other or against a fixed ref-
erence. The adapter exists only virtually in this system as a
software programme, no mechanical installation is required. The
time needed for preparations is 1 - 3 hours and fine-pitch compo-
nents can be reliably tested. However, a relatively long test time is
required: For complex boards 30 minutes or even more may be
necessary. And even this system must be correctly aligned if the
areas to be tested are below 100 m because the tolerance of the
positioning accuracy will then lead to misleading error messages.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
530
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
In order to recognize production errors even in advance, "AOI-test-
ers" are increasingly used. These devices compare the image of a
PCB or of an inner layer of a multilayer board with a stored refer-
ence or with stored data by means of optical record systems. The
final product, the complete multilayer board, can therefore not be
tested.
The PCB manufacturer has to cope with the deficit that the testing
technology is not perfect. However, to be able to offer a sufficient
testing in spite of this fact, the strategy of combined tests is used.
This may signify that high-density zones are tested by a finger-test-
er and zones of lower densities by a pin adapter. It may also be ap-
propriate to test dense zones by an automatic device and high-
density zones manually and optically by a camera system.
It is not very pleasant if zones which cannot be tested are ignored.
Special PCBs Representing numerous unspectacular innovations, the efforts in
the fields of the sensor and cooling technology should be men-
tioned.
There are different possibilities to control the heat development on
an active assembly. The heat transfer via a loop of liquids WITHIN
the PCB is the general option. In combination with a micro pump, the
heat is absorbed by the liquids in the heat-generating zones, trans-
ported to cooling zones and carried-off to the ambient air by control-
led thermovias.
For measuring and generating defined electromagnetic fields, sen-
sor coils on thin laminates are appropriate.The production of these
coils and sensors can be realized with considerable smaller toler-
ances by means of PCB technologies than by the conventional
winding technique.
It is also remarkable that PCBs will have quite a different quality in
this case. They are not only carriers for component but a compo-
nent themselves.
Tolerances The official (acc. to DIN) tolerance of 100 m is permissible in the
PCB production. The conductive pattern and the solder mask may
have this misalignment to the drilling pattern reference.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
531
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Many elementary design-rules are based on this tolerance defini-
tion.
"100 m" do not seem very impressing. In connection with other
rules for the qualitative judgement of a PCB, however, the result may
be some remarkable restrictions.
Example:
"Annular rings of vias". Basically, an annular ring around a via has
to be closed and has to be 100 m at the thinnest section. To en-
sure that in case of a misalignment of 100 m, the annular ring is
still 100 m wide, the annular ring must be defined with 200 m.
Since the ring is circulatory, the result for a pad to a via is that it has
to be 400 m larger than the via diameter to be drilled (figure 12).
Due to these defaults, there is an inappropriate loss of space for
the tracks in case of vias < 0.4 mm. This tolerance must be reduced
at least to the half. Therefore, the standards have to be changed
from 100 m to 50 m and the register accuracy during the produc-
tion of PCBs must be improved from 100 m to 50 m.
Figure 12:
Tolerances for the
annular rings of
vias
Fit-tolerances for vias
Via
0.5 - 0.6mm
Pad
0.9 - 1.0mm
Via
0.2 - 0.3mm
Pad
0.4 - 0.5mm
200m
100m
100m
50m
Standard Superfine-line
Tolerance 50 m Tolerance 100 m
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
532
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Second example:
The "pressing of multilayer boards". The tolerance is 10 % of the
total thickness and does not determine how these 10 % are to be
distributed on the laminates and prepregs. According to this defini-
tion, a multilayer board with an intended thickness of 1.5 mm will
be o.k, if the final thickness is in the range between 1.35 and 1.65.
This may be accepted from the mechanical point of view, but there
will be considerable restrictions, if impedances are taken into ac-
count because the optimum operation range is exceeded from a
thickness tolerance of 7 % on.
Therefore, the pressing tolerance must not exceed 7%.
Filing systems The technical discussions being in the centre of attention usually
conceal the requirement regarding strict, reliable, reproducable fil-
ing systems in the background.
The PCB may NOT follow its way, if the data storage, the documen-
tation, the design-rules, the operating instructions, the file-logistics
or the multilayer board construction instructions are incorrect.
Without any systems, the necessary arrangement and coordina-
tion cannot be successful neither between customer and manufac-
turer nor within the production process.
Many manufacturers have an archive of 20,000 jobs with approx.
400,000 individual data sets. Each data access has therefore to be
unequivocal. There are different possible solutions to create a file
system to meet this requirement (figure 13).
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
533
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 13: File
syntax for CAM /
data processing
archives
In case of multilayer boards, the freedom to combine the material
in nearly any desired stack-up has caused a variety of construction
types for the design (figure 14). There is no clear registration of
these construction types, there are no official rules and hardly pub-
lished catalogues. The PCB manufactures are mostly surprised by
the speed of this development. However, it is mainly their job to in-
form their clients about possibilities and impossibilities regarding
the individual case.
ILF5D044. MB
File name . Extension
ILF5D044.MB
ILF5D044.I2
Both files belong to the same layout
(solder mask and inner layer).
WIE6B124.LS
ABC4H069.LS
The files belong to different layouts,
however, they both describe the
conductive pattern for the solder side
Extension Contents Format
AB
VB
DB
MB, MBN
BS, BSN
I2, I2N
I3, I3N
I4, I4N
I5, I5N
I6, I6N
I7, I7N
I8, I8N
I9, I9N
LS, LSN
ML, MLN
DL
VL
AL
Z1/Z2
UM
ZZM
DRI/NDK
Peelable sold. resist comp. side
Via filling mask comp. side
Silkscreen comp. side
Solder mask comp. side
Conductive pattern comp. side
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern inner layer
Conductive pattern solder side
Solder mask solder side
Silkscreen solder side
Via filling mask solder side
Peelable sold. resist solder side
Drilling programme
Outline plan
Dimensioned outline plan
Drilling programme
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
Gerber 3.2mm
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
534
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 14:
Example of a
multilayer stack-
up
Conclusion No. 1 If you have been involved with PCBs for years and if you have wit-
nessed its technological evolution, you are fascinated if you look
back.
Once called simple and being very unpopular as an additional cost
factor within the calculation system of an assembly, the PCB has
made its way.
It does practically not exist anymore as a pure component carrier.
The PCB itself became a component of complex electronics a long
time ago and its technical features cannot be exchanged at will any
longer. The PCB requires and deserves our attention, our sympa-
thy, our understanding and, last but not least, our respect.
Multilayer stack-up plan
Pressed 1.18 - 1.33 mm

Final thickness Tin lead 1.26 - 1.42 mm
(incl. solder mask) Hot-air 1.29 - 1.45 mm
Gold 1.25 - 1.41 mm
mm Material File Mounting
(0.050 HFPrepreg type : 106)
(0.060 Prepreg type : 1080)
(0.100 Prepreg type : 2125)
A1
A2
B
0.017 Copper *.BS
0.017 Copper *.LS
0.050 FR4
A3
0.060 Prepreg
0.100 Prepreg
0.050 HFPrepreg
0.017 Copper *.I4(N)
0.250 FR4
0.017 Copper *.I5(N)
0.017 Copper *.I6(N)
0.017 Copper *.I7(N)
0.100 Prepreg
0.060 Prepreg
0.250 FR4
0.100 Prepreg
0.060 Prepreg
0.035 Copper *.I2(N)
0.035 Copper *.I3(N)
Multilayer construction type 8M13FR4I5I25K17K35
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
535
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
If you nowadays design an electronic circuit without showing this
respect towards the PCB, you will act with negligence or, in any
case, not very clever.
You will ignore the possibilities resulting due to the numerous vari-
eties of the materials, surface-finishes, conductive pattern structur-
ings and of the connecting alternatives.
The miniaturization with tracks and vias to 100, 80 or even 50 m
regularly produces speculations: What will come next? - 20 m? 10
or even 5 m?
Maybe!
I do not guess so.
The PCB is not the missing link to the hybrid circuit or to the micro-
chip of the 70s. It will always do what has marked its character: to
keep together autonomous components as a whole and to create
connections.
This will also be the case, when its status as a mechanical-electronic
precision component is generally accepted by everyone. This will
probably be the last milestone of its evolution.
Afterwards, different things will come with different names.
The question regarding "the future of the PCB" has therefore found
several answers.
However, there are obviously still two further questions to be an-
swered.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
536
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The first question is:
What is the future of the PCB manufacturer?
The PCB manufacturers do not welcome the present technological
development with arms wide open. The lean years seem to be over
but even the best enterprises did not escape unscathed. The de-
mand of the market for variety, quality and short deadlines creates an
incredible pressure.
Some manufacturers still guess, others already know: To be in the
position to offer all variants of the new PCB, investments will have
to be made which seem completely absurd regarding the (almost)
empty tills and the moderate profits. It is not only necessary to re-
place existing, old machines against new ones which do the same,
of course better, quicker and less expensive (figure 15).
This cycle is not new and the PCB manufacturers are quite used to
this process.
Now it is also necessary to invest in machines which have not been
used and which will become necessary IN ADDITION to the other
machinery.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
537
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Figure 15:
Machines and
investments
A simple example is the contour treatment. It is common to mill
contours. Now, the scoring of contours is also a variant. The scor-
ing maching, however, may not replace the milling machine which,
some years ago, replaced the parallel shears. Now, a scoring ma-
chine AND a milling machine is necessary. Only then, the custom-
ers demand for a combined scoring-milling assembly panel can be
satisfied.
Second example: A pin tester is required for the electronic test.
But, IN ADDITION; a flying-probe tester is necessary in case of mi-
crofine-line PCBs.
These examples depend on the investment capacity of an enter-
prise.
Production technology
to be used for
Standard
Superfine-
line
Super-
microfine-
line
CNC without Z-axis control yes Partially no
CNC with Z-axis control yes yes yes
X-ray drilling machine yes yes yes
Film exposure yes Partially no
Glass master exposure yes yes Partially
Laser direct imaging yes yes yes
Clean-room technology yes yes yes
AOI-tester yes yes yes
Standard multilayer press yes Partially no
Process-controlled multilayer press yes yes yes
Standard electroplating yes Partially no
Process-controlled electroplating yes yes yes
Etching (alkaline) yes Partially Partially
Etching (acid) yes yes yes
Solder mask (screen printing) yes Partially no
Solder mask (film/foil) yes yes yes
Pin adapter, resolution 1/10" yes no no
Pin adapter, resolution 1/20" yes Partially no
Electronic test (translator/probe) yes yes yes
Machines and technology for super-microfine-lines
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
538
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The requirements on the galvanotechnical surface-finish variants
are totally different. Galvanic baths are very susceptible to rest pe-
riods. They cannot be activated or deactivated within a longer pe-
riod of time to quickly carry out a gilding or a tinning. If these baths
are only rarely used, their balance will be upset. They cannot be
used any longer and economical deficits will be the consequence.
If the PCB manufacturer intends to follow the way of the product,
there will be only one alternative: Cooperation!
This is very hard for an industry which was able to produce inde-
pentently for decades and which procured base materials but nev-
er needed considerable industrial services. But that is not all: Many
services cannot be obtained neutrally but only by a competitor who
was a fierce rival in the fight for customers not long ago. These
service enterprises must now be logistically integrated in the inter-
nal organization. Many company policies will therefore be turned
upside down.
Other companies take another appropriate measure as a result of
these changes and start to specialize. They either produce only
single-sided or double-sided PCBs, or only multilayer boards or
only 6 or 8-layer board or only rigid-flexible circuits or only up to a
number of 100 pieces or from 100 pieces on.
This is very economical because the machines are used in the best
possible way.
It is obvious that a specialist for single-sided PCBs is always less ex-
pensive than a company which also produces multilayer boards
which require a perfect plating, measuring places and laser direct
imaging systems which have to be sufficiently used and correctly
maintained and serviced.
Parallel to the investment and specialization concerning machines,
the most urgent problem ist the training and further education of the
personnel. The technical sequences change, it would be fatal, if the
staff did not. The requirements on the CAM data processing are con-
siderable. They cannot be fulfilled, if there are no instructions based
on generally accepted and respected design-rules.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
539
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Conclusion No. 2 Gently but firmly, the PCB takes along on its way its own manufac-
turers and designers. Its sensitivity, differentiation and variety is
transferred to the people who have to deal with it and to the enter-
prises which are connected to it.
The necessity to cooperate, however, offers the chance of a new
enterprise culture. The PCB manufacture changes from a product
to a project which can be only carried out successfully, if it is based
on a partnership.
No, it would definitely not be easier; neither for the PCB manufac-
turer, nor for the customers. The current calculations DM*dm2 are
now invalid. This kind of specification is too superficial.
Even the wish that PCB types, if single or double-sided, in any
quantities, if protoype or series, are available from one manufac-
turer cannot be fulfilled any longer on a long term basis. The nec-
essary specialization will not allow this anymore.
In many fields, the specialization itself will require new coopera-
tions. The series manufacturer has to find an arrangement with the
prototype manufacturer because it is appropriate to have the pro-
totype data as a basis as these data have been tried and tested in
the pre-production. Therefore, a data exchange is necessary. This
requires though a clear coordination during the formal transfer and
an agreement regarding common design-rules and a common
data format.
(I will not comment the subject "common data format". Everyone
knows that I prefer the "Gerber" format.)
Of course, the PCB will become more expensive. The investments
in expensive machinery which can not be fully used because there
are too many product variants hardly leave no other choice.
However, there will also be no reduced delivery times. The time
saved by technology and rationalization get lost on the way to and
from the external service companies.
Regarding these aspects, that's all for the PCB and the manufac-
turers. However, one authority is still missing.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
540
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
The second addtional question:
"What is the future of the CAD layout designer?"
The conflict exists. On its way, the PCB has taken along the lay-
outer more than he/she is actually aware of.
Basic decisions, for which there are partially or completely no cur-
rent experience values, must be made with every new layout.
The combination between mechanics, electronics, EMC, function
and costs will be so individual that the original task, the layout, will
almost be inferior.
The most important support is surprisingly the component manu-
facturer. They have ensured the constant reduction of the pin dis-
tances for the last years. However, the escape to MCMs and COB
indicates that the development of the mechanical adaptation has
come to an end. The bonding of chips on carriers makes the limits
obvious. The opposite force of the possible measures are the ap-
propriate measures.
Regarding the layout, the problems are concentrated on the selec-
tion of the correct material appropriate for the surface-finish, appro-
priate for the track-width, appropriate for the vias and appropriate
for the multilayer construction type. The freedom of the layouter is
determined by the function, on the one hand, and by the costs of
the assembly, on the other hand.
The layouter will not be in the position to decide independently. The
decisions will depend as well more than ever on the constructive
cooperation with the PCB manufacturer and with the assembly pro-
ducer and on the internal organization.
The layouter will be in charge of the classification of layouts and of
the precise technical specification of the PCB.
The layouter has to adopt the sensitivity of the functional compo-
nent "PCB".
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
541
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Conclusion No. 3 Paradoxically, the layouter is restricted by the variety of options.
There is finally the material with the correct
r
-value, but it cannot
be used because it is too expensive.
There is eventually enough space due to signal lines of 150 m.
But there is also a power supply on the same layer always placed
on 70 m of copper, which now cannot exceed 17 m.
Now, the optimal multilayer stack-up is available for the layouter
with 50 m-laminates, which offers an excellent broadband decou-
pling, however, just his/her supplier is (still) not able to process
these laminates.
More?
Multilayer boards out of favourable CEM material?
It does not work because it is a compound laminate.
Bondpads which are tested by the manufacturer by means of pin
adapters?
It does not work because the surface-finish of the pads will be dam-
aged by the adaption.
Tracks with widths of 100 m in 35 m copper?
It does not work because it cannot be etched.
A 10-layer board, 1.8 mm thick with through-vias of 0.1 mm?
It does not work because it cannot be reliably plated.
Then, is there at least some help, a training, further education?
This is also not possible, there is none.
But that is a completely different story.
WgdL
23.08.1999
The Future of the PCB
File :
Erstellt :
542
Publications


C
o
p
y
r
i
g
h
t

b
y

I
L
F
A

G
m
b
H

/

E
d
i
t
i
o
n

5
.
9
.
a
Final word The way the PCB, its manufacturers and its designers will have to
go, will be paved with all technical extras. Many of them are neces-
sary, progressive and rightly successful.
However, the question "how many varieties are actually neces-
sary?", is legitimate.
It must be possible to discuss if the faith in an increasing technical
complication as the best solution, will sometimes unnoticed turn
into a superstition.
Therefore, the miniaturization of the components is generally indi-
cated as the cause for the miniaturization of the vias and of the
conductive patterns.
That is not completely true!
Decisive is, how the connections between the individual compo-
nents are made. This task is often completely left to the autorouters
of the CAD system.
The machines are certainly quicker than we are, but, I do not be-
lieve that they are more intelligent.
It is absolutely important to talk about technolgy.
However, do not let us completely ignore the strategies, the ideas
and the creativity which - also - distinguishes us as human beings.
WgdL
23.08.1999

S-ar putea să vă placă și