Testing PCBs

[Amanda Robin]

I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

One of the attenuator PCBs has ceased to function, and both of the
regulators get hot enough to burn me as soon as power is appleed. I am
testing components to find the problem. My power supply appears OK, or
at least one of the circuits thinks so.

I run into this a lot: how does one test components when they are
soldered in place? What I generally do is use the continuity and
capacitance functions of my multimeter and look for "reasonable"
values. However, capacitors obviously don't always have their rated
value when they are plugged into a circuit board.

It's nice to have a twin circuit that actually works; that way I have a
benchmark for reasonable values. However this time it hasn't helped me
find the problem (yet).

Earlier today I suspected that the 7905 (negative) regulator was fried
and replaced it and its capacitor, but my problem continues. Since both
regulators heat up, I assume that ground must be compromised, but
haven't found the problem yet.

If anyone has any suggestions, either for my immediate problem, or for
the more general problem of testing PCBs, they would be most welcome.

Thanks,
Amanda

 

[Andrew Howard]

Maybe it's the design of the circuit itself that is the problem? I dunno,
just checking...

Andrew Howard

 

[Apollo]

"Amanda Robin" <amanderr@yahoo.com> wrote

I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

7905's may sometimes oscillate if incorrectly bypassed.
You need a capacitor from the -5V rail to Ground, and on
the input to the regulator to Ground. Actually, the same
should be present on the 7805, but 7805's usually don't
actually go into oscillation when the capacitors are
missing - whereas 7905's usually do.
That may be [a] problem.

One of the attenuator PCBs has ceased to function, and both of the
regulators get hot enough to burn me as soon as power is appleed. I am
testing components to find the problem.

Is one of the chips in backwards?
Or, atleast, are the supplies connected to the chip backwards?
Are the power supplies shorted elsewhere?

My power supply appears OK, or
at least one of the circuits thinks so.

Nothing beats putting a 'scope on it and checking.

I run into this a lot: how does one test components when they are
soldered in place? What I generally do is use the continuity and
capacitance functions of my multimeter and look for "reasonable"
values. However, capacitors obviously don't always have their rated
value when they are plugged into a circuit board.

The best, first, thing you can do is check everything
thoroughly. Before you apply power (too late!).
A very thorough visual can do wonders.

It's nice to have a twin circuit that actually works; that way I have a
benchmark for reasonable values. However this time it hasn't helped me
find the problem (yet).

The other best, second, thing you can do, when you don't
have oodles of confidence in successful results, is to
build up the circuit a stage at a time - power supplies
first, check - first chip - test - second chip - test etc

Earlier today I suspected that the 7905 (negative) regulator was fried
and replaced it and its capacitor, but my problem continues. Since both
regulators heat up, I assume that ground must be compromised, but
haven't found the problem yet.

Another method is to photocopy the circuit, sit down and
test every connection on the circuit with a meter, drawing
over the circuit in coloured pen as you do it. You could
also mark an photocopy of the circuit board tracks.
Anything that is left uncoloured is then suspect.

If anyone has any suggestions, either for my immediate problem, or for
the more general problem of testing PCBs, they would be most welcome.

The way we do this sort of work, the main index is:
http://www.airborn.com.au/method/index.html
Hope all this helps.

Regards
Steve

 

[Bob Masta]

On Thu, 25 Mar 2004 01:46:13 GMT, Amanda Robin <amanderr@yahoo.com>
wrote:

I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

One of the attenuator PCBs has ceased to function, and both of the
regulators get hot enough to burn me as soon as power is appleed. I am
testing components to find the problem. My power supply appears OK, or
at least one of the circuits thinks so.

I run into this a lot: how does one test components when they are
soldered in place? What I generally do is use the continuity and
capacitance functions of my multimeter and look for "reasonable"
values. However, capacitors obviously don't always have their rated
value when they are plugged into a circuit board.

It's nice to have a twin circuit that actually works; that way I have a
benchmark for reasonable values. However this time it hasn't helped me
find the problem (yet).

Earlier today I suspected that the 7905 (negative) regulator was fried
and replaced it and its capacitor, but my problem continues. Since both
regulators heat up, I assume that ground must be compromised, but
haven't found the problem yet.

If anyone has any suggestions, either for my immediate problem, or for
the more general problem of testing PCBs, they would be most welcome.

Thanks,
Amanda

To the good advice others have given, let me add one
more bit: It's handy to have a good bench supply that
you can substitute for the board supplies. You can
unsolder board components to disconnect the board
supplies, or even cut traces. On a first-time board
that you want to be extra-careful with, you can start
off without the board supplies present, and just
run from the bench supplies. (Keep the board
supply output filter caps, however.)

With the supplies separate like this, you can easily
insert an ammeter, if your bench supply doesn't
already have one. Bench supplies also tend to be
more robust and may have adjustable current
limits that can be helpful. Even if you build your
own bench supply, just being able to isolate
it to insert ammeters may be useful.

Since you have one baord already working, you
might want to cut the supply connections on
both and use the known-good supplies to drive
the bad board, one supply at a time. If the good
supply makes the bad board work, you've solved
your problem. If having all good supplies doesn't
help, then you at least know it's not a supply
problem.


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

[CFoley1064]

Subject: Testing PCBs
From: Amanda Robin amanderr@yahoo.com
Date: 3/24/2004 7:46 PM Central Standard Time
Message-id: <240320041946135487%amanderr@yahoo.com

I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

One of the attenuator PCBs has ceased to function, and both of the
regulators get hot enough to burn me as soon as power is appleed. I am
testing components to find the problem. My power supply appears OK, or
at least one of the circuits thinks so.

I run into this a lot: how does one test components when they are
soldered in place? What I generally do is use the continuity and
capacitance functions of my multimeter and look for "reasonable"
values. However, capacitors obviously don't always have their rated
value when they are plugged into a circuit board.

It's nice to have a twin circuit that actually works; that way I have a
benchmark for reasonable values. However this time it hasn't helped me
find the problem (yet).

Earlier today I suspected that the 7905 (negative) regulator was fried
and replaced it and its capacitor, but my problem continues. Since both
regulators heat up, I assume that ground must be compromised, but
haven't found the problem yet.

If anyone has any suggestions, either for my immediate problem, or for
the more general problem of testing PCBs, they would be most welcome.

Thanks,
Amanda

Hi, Amanda. Assuming nothing else on the board is getting hot (test by touch),
it kind of sounds like you either have a design rule violation (board layout is
unmanufacturable because traces/pads are too close, &c) or have a short in a
poorly etched PCB trace from V+ to V-. I've frequently been surprised by how
much current can pass through a tiny sliver of copper without fusing. If this
is the case, sometimes the board itself will have a small hotspot where the
current is flowing across the short. Assuming you don't have a design rule
problem (check with the board house you're working with on their real limits),
use an educated touch across the whole board area to feel for hotspots on the
circuit board itself. If that doesn't do it, in order to start tracking this
down, you'll need a fast-reacting voltmater (a standard handheld with 2 updates
on the reading per second won't do here). Set things up so you can apply power
with a footswitch, and then apply power momentarily, looking for voltage drops
across the board trace. Where the voltage is dropping, current is passing
through. By repeatedly applying momentary power, and checking for voltage
drops across the trace, you can nose your way toward the point where the traces
are shorted together.

It isn't as easy if you've got the power traces on an internal layer, but you
can get an idea from that. Unless you have copper pour for GND, you can also
sometimes put the board in front of a bright light, and see the internal
layer(s). Look for other evidence of poor etching, such as cloudy areas in
internal layers where traces weren't completely etched. Also inspect for
internal layer alignment -- do the pads on internal layers line up well with
the top and bottom? Alignment issues are a frequent cause of these problems.

Possibly you need a better board house. By the way, try to think American
here. There's a lot of excess capacity out there, and a local company is
easier to talk with and will deal with feedback in a more effective manner.

Good luck
Chris

 

[John Jardine]

Amanda Robin <amanderr@yahoo.com> wrote in message
news:240320041946135487%amanderr@yahoo.com...

I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

One of the attenuator PCBs has ceased to function, and both of the
regulators get hot enough to burn me as soon as power is appleed. I am
testing components to find the problem. My power supply appears OK, or
at least one of the circuits thinks so.

I run into this a lot: how does one test components when they are
soldered in place? What I generally do is use the continuity and
capacitance functions of my multimeter and look for "reasonable"
values. However, capacitors obviously don't always have their rated
value when they are plugged into a circuit board.

It's nice to have a twin circuit that actually works; that way I have a
benchmark for reasonable values. However this time it hasn't helped me
find the problem (yet).

Earlier today I suspected that the 7905 (negative) regulator was fried
and replaced it and its capacitor, but my problem continues. Since both
regulators heat up, I assume that ground must be compromised, but
haven't found the problem yet.

If anyone has any suggestions, either for my immediate problem, or for
the more general problem of testing PCBs, they would be most welcome.

Thanks,
Amanda
Sounds like one of the chips (connected across the supply rails) has gone

short circuit.
With power switched off, monitor for a short circuit across the plus and
minus rails. Unsolder a power supply pin on each chip and watch if the
short clears.
regards
john

 

[Amanda Robin]

In article <c3vhrl$3c6$1@news5.svr.pol.co.uk>, John Jardine
<john@jjdesigns.fsnet.co.uk> wrote:

Amanda Robin <amanderr@yahoo.com> wrote in message
news:240320041946135487%amanderr@yahoo.com...
I have two PCBs that I fabricated that attenuate an audio signal and
implement a low-pass filter. They have 7805 and 7905 voltage
regulators. They use an LTC 1164 8th order LP filter and 3 AD711 op
amps. The filter cutoff is set by a clock signal from another PCB.

snip my own description of voltage regulators both overheating

Sounds like one of the chips (connected across the supply rails) has gone
short circuit.
With power switched off, monitor for a short circuit across the plus and
minus rails. Unsolder a power supply pin on each chip and watch if the
short clears.
regards
john

Thank you all for the great suggestiong in this thread. I followed
Steve's advice and did some isolation stuff. I found a potential
problem (loose leg on the LTC1164) and fixed it. I applied voltage to
the re-assembled circuit and everything was fine.

But then--I connected the input and output lines and clock source. The
overheating comes only when the clock signal(about 6.4 kHz) is
connected. (So the loose leg on the chip might not yet have been a
problem.)

I don't know yet if this means I have a problem with the LTC1164 or
with the clock circuit. I followed the rules in the LTC's documentation
about connecting the clock signal, and about the clock and the lp
filter not having the same power supply.

Anyway, tomorrow I'll do some more isolation. I'm inclined to think I
have a problem with the clock circuit or its connection to the
attenuator circuit. The only thing that has changed lately is that I
removed the clock chip from the clock circuit and replaced it again.
Maybe I joggled something. On the other hand, the same clock circuit
serves 4 other PCBs in a star configuration and they are not heating
up, and this PCB heat up no matter which connection (in the star) I
switch it to. Maybe John is right and I have a bad chip.

By the way, this is not a new PCB. It has worked fine for a year. Also,
I made it myself, so unfortunately I can't blame a board house. Too
bad!

Again, any suggestions are much appreciated.

Amanda