# FR4 PCB material quality in Aus

G

#### Geoff C

Jan 1, 1970
0
I have recently had a lot of troubles with 2 PCB assy's which had been
working in products for 10 years or more, but about a year ago started
having problems. The symptom appeared to be leakage in the preamp, as these
boards have high gain and also modulated input signals of a few dozen Hz.
The problems began when we started using different suppliers, such as a
popular low-cost one in Queensland. I could not pin down a correlation with
a particular supplier, but all the bad boards used Chinese blanks. However,
we had also used Chinese boards for some years without problems. That is
what made it hard to simply blame chinese suppliers.

Circumstantial evidence points to US origin FR4 material being much better
than the Chinese. Perhaps Chinese makers were importing US FR4 for some
time but now make their own, but that is conjecture. When we switched to a
ceramic fibreglass composite by a particular maker, the problem
disappeared. Also, as a test, we tried some PCB's from a large reputable
(but pricey) Melbourne PCB maker. They were good too, but if Chinese FR4 is
a problem and they start to use it, then we would still have a problem.

I can't say what the nature of leakage is, but the ceramic PCB's do work,
with a few $extra per board. I will say that we have about 150 other board types with no problems with Chinese suppliers, but they are not so sensitive to leakage or dielectric variation. BTW, we tried baking, guard rings etc. D #### David L. Jones Jan 1, 1970 0 Geoff said: I have recently had a lot of troubles with 2 PCB assy's which had been working in products for 10 years or more, but about a year ago started having problems. The symptom appeared to be leakage in the preamp, as these boards have high gain and also modulated input signals of a few dozen Hz. The problems began when we started using different suppliers, such as a popular low-cost one in Queensland. I could not pin down a correlation with a particular supplier, but all the bad boards used Chinese blanks. However, we had also used Chinese boards for some years without problems. That is what made it hard to simply blame chinese suppliers. Circumstantial evidence points to US origin FR4 material being much better than the Chinese. Perhaps Chinese makers were importing US FR4 for some time but now make their own, but that is conjecture. When we switched to a ceramic fibreglass composite by a particular maker, the problem disappeared. Also, as a test, we tried some PCB's from a large reputable (but pricey) Melbourne PCB maker. They were good too, but if Chinese FR4 is a problem and they start to use it, then we would still have a problem. I can't say what the nature of leakage is, but the ceramic PCB's do work, with a few$ extra per board. I will say that we have about 150 other board
types with no problems with Chinese suppliers, but they are not so
sensitive to leakage or dielectric variation.

BTW, we tried baking, guard rings etc.

Haven't had any problems with our boards, some made by Entec, and
others get supplied to us so through a third party so I don't know the
origin.
We have 500Mohm input preamps on the Entec boards and the other is a
Gohm range switching matrix. No problems with either.

Dave

G

#### Geoff C

Jan 1, 1970
0
Hi ,
are you sure you should be pointing to the PCB material?
Yep.

Did you analyse the nature of the failures being observed and try to
determine the root cause? Sounds perhaps that the root cause was not
determined.

Yes, heaps of analysis. The one thing that always worked was wiring the
components connected to the opamp input node in the air without touching
the PCB.

Could it be a variation in component tolerances which was not
accounted for in the circuit design? Perhaps a pole or zero moving,
bias changes, etc as a result? Or perhaps components out of tolerance
and no inwards goods inspection?

Circuit is tolerant of component cganges as the gain is servoed.

Were all the test articles made from the same batch code components as
the problem units?

Not sure, the bare boards were in stock for some time. Incoming QC is
minimal. You know, costs and all that.
Were some of the components lead free and the problem a result of
whiskers?, not all manufacturers can guarantee you don't have lead
free parts (yet).

Whiskers would create larger currents than the few nA I've observed.
In your last comment you indicate that you know the circuit is
sensitive to dielectric variation, isn't that saying you've not
designed a robust cct or have not defined the minimum requirements of
the substrate to be used?

Well, I commented on dielectric constant but I don't really think it's an
issue, just something that a lot of people here suggest since it is more
of a problem on RF baords when trying to control stripline impedances
etc. It's true that if you take the worst case for FR4 into account, the
circuit could conceivably not work. FR4 specs are really not that great
but the typical performance is much better. That is a correct assertion
you've made. However, when the cct was first used, and for about 10 years
since, it has worked fine. Under these conditions, it would be not
desirable to use lower leakage PCB's since it FR4 is cheap.
If it is the material then that is easy fixed, specify on your
purchase orders the material to be used, or supply it FIA.

That is the solution proposed at present.

Greg
(deja vu or have i actually read this or a very similar post in the
recent past?)

I mentioned this some months ago before the buildup of circumstantial
evidence pointing to the bare boards.

A

#### Alf Katz

Jan 1, 1970
0
Hey Geoff,
Haven't had any problems ourselves, lots of boards from China, but
nothing nanoampish.
Have you megged out the bare boards at all?
Other possibilities besides the FR4 material, if you haven't already
considered them could be:
Solder resist (if any).
Residue from whatever is used to adhere the Cu to the board.
Residue from etchant and/or whatever is used to clean after etching
(possibly under the solder resist - if any).
Moisture trapped under the solder resist.

Most of these could be eliminated with careful use of a dremel between
tracks.

Cheers,
Alf

G

#### Geoff C

Jan 1, 1970
0
Hey Geoff,
Haven't had any problems ourselves, lots of boards from China,
but
nothing nanoampish.
Have you megged out the bare boards at all?
Other possibilities besides the FR4 material, if you haven't
considered them could be:
Solder resist (if any).
Residue from whatever is used to adhere the Cu to the board.
Residue from etchant and/or whatever is used to clean after
etching
(possibly under the solder resist - if any).
Moisture trapped under the solder resist.

Most of these could be eliminated with careful use of a dremel
between
tracks.

Cheers,
Alf

Hi Alf,

yep, megged boards with Keithley electrometer but measured >200 gigohm.
Have considered all of above. Moisture under the resist or some other leaky
substance is possible, but we can't control that. Baking overnight at 60
deg had no measurable effect. We need something manufacturable that
requires no intervention.

The boards that work use Rogers 4003 material, and this stuff is not too

A

#### Alf Katz

Jan 1, 1970
0
Geoff C said:
requires no intervention.
LOL. The dremel suggestion was intended to eliminate surface effects as a
culprit on known bad board, not as a manufacturing technique. At least that
will confirm the material is the cause.

Anyhow, good luck with it. Sounds like at those sort of current levels, any
excuse to use a better material is a good one.

Cheers,
Alf

D
Replies
32
Views
3K
Allan Herriman
A
E
Replies
0
Views
16K
Eeyore
E
J
Replies
12
Views
3K
John S
J
J
Replies
4
Views
923
John S
J
Replies
1
Views
696