Linear power supply undershooting !!!!!!

[pablo]

Hello guys !!!!

The story is this:

The last tuesday i received my brand new instek gds-1062a oscilloscope (my first oscilloscope ever),
so i started to play around with it  . Then i decided to hook up the oscilloscope probe to my little homemade
power supply.

The power supply has 3 voltage regulators so that i'm able to use a 5 volts output, positive variable output and
a negative variable output. Here is the schematic diagram:


The thing is that i set my oscilloscope in single shot mode and tried to capture any overshoot it had.
(quiet frankly i expected a smooth transition from gnd to 5 volts due to the linear nature of the power supply)
but i was surprised to find out that my trusty linear power supply was overshooting 2.5 volts above the 5 volts
positive rail and undershooting 2.5 volts below ground. 

I really don't know why is doing such a thing. Maybe i'm probing the circuit incorrectly. I 've heard that
the probe's ground terminal can affect or load the circuit due to inductance. Anyway here are some images from
the oscilloscope so you can see how the power supply is performing when is turned on.



As you can see there is no overshooting, that's because i performed several tests but i couldn't get
the overshoot again but i swear it was there the first time. It was about 2.5 volts above the positive 5 volts
rail. (7.5v)



In these pictures you can see how the 7805 voltage regulator ramps up the voltage in several steps but
before that it oscillates for a while and goes below ground reference. It means that i was applying
-2.5 volts directly to my circuitry. The weird thing it's that i never had any issues with it.
Maybe i was only degrading the life expectancy of components to a certain extent.



Is this a common thing to happen?

here is another shot:



And here is how i probed the circuit:





Any suggestions, ideas, etc.
Thanks

[pablo]

i also found this:


It's almost what i did in my power supply but they placed diodes across regulator's outputs.
Maybe there is instability issues with this design, otherwise why they used diodes???

thanks

[Mechatrommer]

i suspect it coming from your transformer, before and after the bridge diodes. try probing that area might give some clue. i can see your regulator lack of protection circuit (diodes), that also may explain why spike can sneek in. here the schema from 317 datasheet. and try to probe at peakdetect mode to catch that very quick spike. nice circuit though, thanx for showing us.

[pablo]

Thanks for your suggestions Mechatrommer 

Although i haven't done what you said, it's a pretty good idea, but i will have to solder
some wires directly from the solder joints because there is not where to probe properly.

In the picture i show how close the ground layer is to the AC tracks.
What a bad product design.
Any further suggestions will be appreciated.

thanks

[bilko]

I would use a dremel tool or scalpel and remove some of the copper ground adjacent to the AC input. There should be at least 5mm 1/4" gap to provide isolation.

Edit: Also noticed that you don't have ground wire. This means that at the moment any moisture between AC line and DC ground will conduct mains voltage to your DC ground.

I would use a 3 core mains lead with ground wire.

[Mechatrommer]

agree with yacht. imo, there should be clear visual separation between them. its hard to trace your circuit, esp where the main connected to your transformer primary. but i guess the blue dot is not it. you can probe on those 3 big circles in the middle (trasnf. 2ndary right?). or that 4 pins below (bridge IC?)

[bilko]

Mecha, the AC input is bottom left of the PCB, the 2 core flex supplies the transformer via a fuse (top- left PCB). Hardly any clearance even between phase and neutral.

[tecman]

Just a quick comment on your PCB.  For standards and good practice, you should open up the clearances on your AC pads/runs, from AC to AC and from both sides to the ground plane.  3mm/0.1" minimum for safety and leakage concerns.

paul

[pablo]

Thanks guys for your answers.

Edit: Also noticed that you don't have ground wire. This means that at the moment any moisture between AC line and DC ground will conduct mains voltage to your DC ground.

there should be clear visual separation between them. its hard to trace your circuit, esp where the main connected to your transformer primary.

For standards and good practice, you should open up the clearances on your AC pads/runs, from AC to AC and from both sides to the ground plane

I think woud be easier to discard that PCB and make a new one. But this time i'm gonna
reduce the ground layer to only cover the secondary circuit. Clearly i did something wrong with, and it seems to be major
concern. I just hope that 12 vac from the secondary does not couple into the DC ground causing some noise.

You are measuring with very little load (10 M? of a 10x probe or 1 M? of a 1x probe). Add a reasonable resistor as load.

I haven't tested that but the problem is that most of the time i only use the power supply to feed only a microcontroller
so it should be stable with very little load, otherwise it would be suitable for that purposes.

BTW, your question doesn't get more interesting just because you add multiple exclamation marks in the subject line. This just makes you look stupid and impatient.

Well, i didn't typed multiple exclamations marks just because i believed that would look more attracting to people.
I did it because i was surprised to see how my power supply was undershooting.

Well anyway, maybe next time i will avoid using them, although IMO those characters harm nobody. Sometimes people
overreact.

Thanks to all of you for replying, including you BoredAtWork !!!!!!! .

[Bored@Work]

I did it because i was surprised to see how my power supply was undershooting.

Oh sure, pull the other one. Do you attention seekers really think we are so stupid to believe this?

[Mechatrommer]

dont mind BoredAtWork pablo, he always there on interrupt pin 0x08 . no need to trash that pcb, just do as yacth's advice (cut the track a bit) will do, thats is one fine pcb, unless worst thing happen as mentioned. mains dont just can kill pcb, but human too. report your next probing when you found out something. cheers.

[bilko]

I'd get to work with the hobby grinder, dremel ? and do something like this

[qno]

I am missing the small 330n capacitors on the input of the regulators in the schematics.

Read the datasheet carefully.

[pablo]

just a quick update:

I've been performing some test and here is what i've found...

I probed the three outputs and they where doing the same thing. undershooting or oscillating before the
voltage started to rise.

I also probed regulators inputs with same results. So i opted to remove the transformer from the PCB
and soldered some wires directly between PCB and transformer's secondary. Unfortunately that didn't help.

Finally I replaced every component and built it again but this time on a breadboad with no luck.
My last test consisted of a bridge rectifier and a capacitor with same results.

WTF is going on !!!!

Maybe there is nothing wrong with my circuit so I'm afraid that the oscilloscope's probe
could be causing this due to some weird tank circuit behavior while being connected,
like some inductive circuit loading.

What do you think?

[jahonen]

It is very probably common mode noise transient coupled into measurement. So what you see is a measurement artifact. You can verify this by connecting the scope probe ground and tip to the same point (for example by probing ground), and checking if you still see the noise (I guess that you do).

Regards,
Janne

[pablo]

Janne...

I tested what you suggested and it turned out to be that.
I simply connected the probe to the ground terminal (where it was connected before) and
i got the same results.

But then, what causes this...?
It happens when i plug the AC cord to the wall outlet to feed the transformer.

Does this mean that:
-that my oscilloscope is ok?
-has nothing to do with probing (loading) the circuit?
-this could affect my circuitry?

Well anyway i really appreciate your help.

[pablo]

here is another image of the common mode noise.
The probe's ground and tip are tied together in the same node (circuit's dc ground).
I got the same results in the positive rail.

thanks.

[am2pgs]

The common mode noise measurement artifact is a very common. It is to do with noise coupling through the mains earth connection. There is nothing wrong with your scope as such but some scopes deal with it better due to better earth isolation.

often people recognise it and then simply igonre it.

If you want to rule it out for sure you can do a few things:

- Do what you just did
- use a battery operated scope
- use differential probes with decent CMMR
- play around with you mains connections (the scope and the DUT)

[pablo]

I bought one of these AC outlet wiring testers so i know that ground is open or floating.

Do you think this could be causing the common mode measure artifact?

[LEECH666]

http://sound.westhost.com/project44.htm

If you take a look at Rods Page he states:
D5 and D6 are recent additions, suggested by a reader (and as used in P05). These help protect against external reverse polarities.

The other diodes (AFAIK) protect against the various capacitors discharging through the output pins of the regulators. The regulators might get damaged if the output voltage is higher than the input voltage. The diodes form an alternative current return path to protect the regulators.

Cheers from Germany,
Florian

//EDIT: Whoops, sorry for the late post. I see the topic has moved onto other topics already.