Very strange standard voltage regulator behaviour

[Alex]

Hello everybody


I'm having serious problems with a simple +5/+12 V dual output power
supply, which I never had before. I already built many such supplies,
all working without any problems.
It would be great, if anybody could give me a hint/tip

First a description of the circuit:

Load to be connected to the psu: 5V @ about 200mA, 12V @ about 30mA

*The mains (230V 50Hz) enters a standard transformer (12V 0.833A
output rating)
*The secondary winding is connected to a B40C800 bridge rectifier,
then goes to a 10'000 uF filter cap, with a 4.7kohms resistor parallel
to it.
*This gives a measured unregulated voltage of about 19.1 VDC with no
load. A load which draws 400mA, directly connected to the filter cap
reduces the unregulated DC voltage to 16V.
*Connected to the filter cap are two positive fixed voltage regulators
(LM340AT +5V and LM340AT +12V). Each of them with 220nF low loss,
polypropylene foil caps at input and output, just beside the LM340
regulators. The regulators are equiped with apropriate heatsinks.
*At the outputs of the regulators are two LEDs with series resistors
as seperate power indicators for 5V and 12V.
The whole psu is built on a PCB carefully designed regarding the
needed gaps between mains and secondary side, no gnd loops etc.

Now the problems are the following:

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about
20s, the current goes back to 400mA and a "nice" oscillation of about
250kHz is seen on the scope screen. Soldering 220nF ceramic chip caps
directly to the leads of the LM340AT didn't help anything. Also
changing to different brands of V-regulators (7805/7812 from Motorola
etc.) didn't eliminate the problem. Disconnecting both regulators from
the unregulated voltage and testing them each alone didn't make the
strange phenomenons disappear neither.

It's really frustrating as I'm an electronics engineering student and
had just designed/built a properly working 1GHz frequency synthesizer
with uC-Control, UHF-PA-stage as a project work.

I've already built about a dozen or more of such standard supplies for
digital & analog circuits, which all work fine.

Many thanks for every help in advance

Alex

 

[Ian Stirling]

Hello everybody


I'm having serious problems with a simple +5/+12 V dual output power
supply, which I never had before. I already built many such supplies,
all working without any problems.
It would be great, if anybody could give me a hint/tip

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about

Hmm.
Wierd fault in the capacitor?
Does anything that shouldn't get hot?

 

[John Miles]

Hello everybody

...
It's really frustrating as I'm an electronics engineering student and
had just designed/built a properly working 1GHz frequency synthesizer
with uC-Control, UHF-PA-stage as a project work.

Wonder if you got some bad parts? It happens. Try some different
regulators from another batch or distributor.

The Mini-Circuits people were quite interested when I built a code-
practice oscillator with a defective 4 GHz ERA-5 MMIC a few months
ago....

-- jm

 

[Spehro Pefhany]

fOn 22 Jun 2004 15:45:22 -0700, the renowned [email protected] (Alex)
wrote:

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about
20s, the current goes back to 400mA and a "nice" oscillation of about
250kHz is seen on the scope screen.

Is that "4.7K" resistor getting hot?

If not, take off one of the regulators at a time and see where the
current is going. As I understand it from your description, it's
either going into the "4.7K" resistor or one of the two regulators.

Best regards,
Spehro Pefhany

 

[John Fields]

Hello everybody


I'm having serious problems with a simple +5/+12 V dual output power
supply, which I never had before. I already built many such supplies,
all working without any problems.
It would be great, if anybody could give me a hint/tip

First a description of the circuit:

Load to be connected to the psu: 5V @ about 200mA, 12V @ about 30mA

*The mains (230V 50Hz) enters a standard transformer (12V 0.833A
output rating)
*The secondary winding is connected to a B40C800 bridge rectifier,
then goes to a 10'000 uF filter cap, with a 4.7kohms resistor parallel
to it.
*This gives a measured unregulated voltage of about 19.1 VDC with no
load. A load which draws 400mA, directly connected to the filter cap
reduces the unregulated DC voltage to 16V.
*Connected to the filter cap are two positive fixed voltage regulators
(LM340AT +5V and LM340AT +12V). Each of them with 220nF low loss,
polypropylene foil caps at input and output, just beside the LM340
regulators. The regulators are equiped with apropriate heatsinks.
*At the outputs of the regulators are two LEDs with series resistors
as seperate power indicators for 5V and 12V.
The whole psu is built on a PCB carefully designed regarding the
needed gaps between mains and secondary side, no gnd loops etc.

Now the problems are the following:

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about
20s, the current goes back to 400mA and a "nice" oscillation of about
250kHz is seen on the scope screen. Soldering 220nF ceramic chip caps
directly to the leads of the LM340AT didn't help anything. Also
changing to different brands of V-regulators (7805/7812 from Motorola
etc.) didn't eliminate the problem. Disconnecting both regulators from
the unregulated voltage and testing them each alone didn't make the
strange phenomenons disappear neither.

It's really frustrating as I'm an electronics engineering student and
had just designed/built a properly working 1GHz frequency synthesizer
with uC-Control, UHF-PA-stage as a project work.

I've already built about a dozen or more of such standard supplies for
digital & analog circuits, which all work fine.

Many thanks for every help in advance

 

[Spehro Pefhany]

Could it be possible that the footprints of the regulators got printed
in reverse? That is, are the inputs and outputs reversed?

That sounds like an excellent possibility.

Best regards,
Spehro Pefhany

 

[Tam/WB2TT]

Disconnecting both regulators from
the unregulated voltage and testing them each alone didn't make the
strange phenomenons disappear neither.

Alex,
Are you saying that with both regulators removed, you still don't get the
19V? I would look for a shorted diode, or bad filter cap. Since you are able
to measure external current, doesn't seem like it could be a shorted
transformer. Capacitor, more than likely.

Tam

 

[John Larkin]

Hello everybody


I'm having serious problems with a simple +5/+12 V dual output power
supply, which I never had before. I already built many such supplies,
all working without any problems.
It would be great, if anybody could give me a hint/tip

First a description of the circuit:

Load to be connected to the psu: 5V @ about 200mA, 12V @ about 30mA

*The mains (230V 50Hz) enters a standard transformer (12V 0.833A
output rating)
*The secondary winding is connected to a B40C800 bridge rectifier,
then goes to a 10'000 uF filter cap, with a 4.7kohms resistor parallel
to it.
*This gives a measured unregulated voltage of about 19.1 VDC with no
load. A load which draws 400mA, directly connected to the filter cap
reduces the unregulated DC voltage to 16V.
*Connected to the filter cap are two positive fixed voltage regulators
(LM340AT +5V and LM340AT +12V). Each of them with 220nF low loss,
polypropylene foil caps at input and output, just beside the LM340
regulators. The regulators are equiped with apropriate heatsinks.
*At the outputs of the regulators are two LEDs with series resistors
as seperate power indicators for 5V and 12V.
The whole psu is built on a PCB carefully designed regarding the
needed gaps between mains and secondary side, no gnd loops etc.

Now the problems are the following:

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about
20s, the current goes back to 400mA and a "nice" oscillation of about
250kHz is seen on the scope screen. Soldering 220nF ceramic chip caps
directly to the leads of the LM340AT didn't help anything. Also
changing to different brands of V-regulators (7805/7812 from Motorola
etc.) didn't eliminate the problem. Disconnecting both regulators from
the unregulated voltage and testing them each alone didn't make the
strange phenomenons disappear neither.

It's really frustrating as I'm an electronics engineering student and
had just designed/built a properly working 1GHz frequency synthesizer
with uC-Control, UHF-PA-stage as a project work.

I've already built about a dozen or more of such standard supplies for
digital & analog circuits, which all work fine.

Many thanks for every help in advance

Alex

The regs may oscillate with just 220 nf low-esr output caps. Try a few
uF of electrolytics instead and see what happens.

Cute problem, whatever it is.

JOhn

 

[Mr TUBEAMPS]

i had a strange prob with a mosfet amp
that i bort from dick smith years ago,
it was the first mosfet amp kit, 50 watts p/ch.
when i fired it up one of the filter caps went off
like a rocket, the pint on the outside of the cap
was reversed.

john

 

[Jan Panteltje]

Replace filter cap.

 

[Fred Bloggs]

Hello everybody


I'm having serious problems with a simple +5/+12 V dual output power
supply, which I never had before. I already built many such supplies,
all working without any problems.
It would be great, if anybody could give me a hint/tip

First a description of the circuit:

Load to be connected to the psu: 5V @ about 200mA, 12V @ about 30mA

*The mains (230V 50Hz) enters a standard transformer (12V 0.833A
output rating)
*The secondary winding is connected to a B40C800 bridge rectifier,
then goes to a 10'000 uF filter cap, with a 4.7kohms resistor parallel
to it.
*This gives a measured unregulated voltage of about 19.1 VDC with no
load. A load which draws 400mA, directly connected to the filter cap
reduces the unregulated DC voltage to 16V.
*Connected to the filter cap are two positive fixed voltage regulators
(LM340AT +5V and LM340AT +12V). Each of them with 220nF low loss,
polypropylene foil caps at input and output, just beside the LM340
regulators. The regulators are equiped with apropriate heatsinks.
*At the outputs of the regulators are two LEDs with series resistors
as seperate power indicators for 5V and 12V.
The whole psu is built on a PCB carefully designed regarding the
needed gaps between mains and secondary side, no gnd loops etc.

Now the problems are the following:

If I switch on (no load, only the indicator LEDs) the psu, first
everything is fine: The LEDs are lit, the output voltages are correct
and the current consumption of the regulators is as stated in the
datasheet. Suddenly the current (measured between regulator and filter
cap) begins to rise with about 2mA/min first, then faster and faster
(exponential like behaviour). The quiescent current goes up two 1.7A,
then the unregulated voltage is about 2.5V, the LEDs off. After about
20s, the current goes back to 400mA and a "nice" oscillation of about
250kHz is seen on the scope screen. Soldering 220nF ceramic chip caps
directly to the leads of the LM340AT didn't help anything. Also
changing to different brands of V-regulators (7805/7812 from Motorola
etc.) didn't eliminate the problem. Disconnecting both regulators from
the unregulated voltage and testing them each alone didn't make the
strange phenomenons disappear neither.

It's really frustrating as I'm an electronics engineering student and
had just designed/built a properly working 1GHz frequency synthesizer
with uC-Control, UHF-PA-stage as a project work.

I've already built about a dozen or more of such standard supplies for
digital & analog circuits, which all work fine.

Many thanks for every help in advance

Alex

For one thing, you are overstressing the B40C800 which is not rated for
filter capacitors as large as 10,000u without a series current limiting
resistor. It sounds like you may be using a miniaturized board layout,
and since the description sounds heat related, the problem is most
likely a clearance problem with some small copper filing or solder ball
that is shorting the circuit at working temperature. It is somewhere in
the vicinity of the regulator heat sinks.

 

[legg]

The whole psu is built on a PCB carefully designed regarding the
needed gaps between mains and secondary side, no gnd loops etc.

Perhaps you used the wrong regulator puppet in your layout, this time.

LM320 series isn't the same as LM340. Perhaps you've missed routing
the ground line through the regulator, in the new layout.

Check the layout on the actual board, anyways, because it's one of the
things that always changes, whether you intend it to change, or not.

RL

 

[Alex]

Perhaps you used the wrong regulator puppet in your layout, this time.

LM320 series isn't the same as LM340. Perhaps you've missed routing
the ground line through the regulator, in the new layout.

Check the layout on the actual board, anyways, because it's one of the
things that always changes, whether you intend it to change, or not.

RL

Hello,


many thanks for all your answers. As there were no errors in the
layout/pcb (I always make multiple checks before I etch a pcb), it
must have been in one of the components. And finally the fault was in
the 220nF bypass caps, surrounding the voltage regs. These are 100V
MKS caps and they had a leakage current of several milliamps, which
increased over time until the psu was overloaded. Very strange, I
never experienced this before. Usually I get all components from my
local electronics supplier. But this time I took the components from
our lab as this is a semester project. Probably someone killed this
caps and put them back in the labs component inventory.

Now the psu is working fine with "my" caps. An further I replaced the
filter cap with a smaller one (4700u) and a 1ohm series resistor to be
withing the specs of the B40C800 rectifier bridge.

Again many thanks for your help

and kind regards,

Alex