Is this PSU noise level acceptable for my needs ? How can I reduce it ?

[ItsMike]

Hello everyone,

I'm working on an RGB led strip controller, So I've got a 12v 6A "brick" (laptop like) power supply for it from ebay ( ).
I tried measuring the voltage directly on the end of the connector (with a "springy" ground lead) - See the attached waveforms.

One set is with a 2A load (5 meter rgb led strip fed from only one side with all leds on) and the other set is without any load.

There seems to be noticeable ripple at 50Hz and 52KHz, which are the ac power line ripple and the PSU switching frequency (i'd assume).

My controller would have an MCU at 3.3v, and I plan to use the internal ADC on it to sample an electret microphone so I'm planning on using an opamp to amplify the mic, I wish to power the opamp from a single supply of as high voltage as I can (up to 12v).
Is the power supply too noisy for this ?
If so, How should I filter it ?

One more thing I noticed, the power supply has a 3 lead power connector, but when I took it apart I noticed the ground on the input connector isn't soldered nor connected anywhere.
When I probed the supply output voltage, I noticed tiny sparks when the scope's ground lead got really close/touched the power supply's ground. Is that normal ?

Thanks in advance,
Mike.

[David Hess]

Some of that noise might be common mode noise through a ground loop.  You can test this by shorting the probe tip to the probe ground at the measurement point and looking for residual noise.

The AC ripple implies that either the power supply is worn out or always suffered from poor line regulation.  The high frequency switching noise can be removed with an LCR differential and probably common mode filter.

[Simon]

If this is simply meant for LED's it may not have been designed that great. Is it a constant voltage supply ?

[ItsMike]

Yes, it's a constant voltage supply, kind of like this one here.

I'll drive the led strip though this supply and I would also like to filter it enough to run all my sensitive electronics from it (About 100 mA of current).

[Simon]

It could probably use a decent input capacitor as it seems to struggle from one cycle to the next. Speaking of which you have a 50Hz noise not 100Hz which you would get from a proper diode bridge so it looks like it's really cheap and nasty with half wave rectification. try a bridge and smoothing cap of your own and see what happens - but don't kill yourself on the mains if you do.

[David Hess]

It could probably use a decent input capacitor as it seems to struggle from one cycle to the next. Speaking of which you have a 50Hz noise not 100Hz which you would get from a proper diode bridge so it looks like it's really cheap and nasty with half wave rectification. try a bridge and smoothing cap of your own and see what happens - but don't kill yourself on the mains if you do.

I missed that.  Either the power supply has failed in some way or is incredibly cheap.

[ItsMike]

The power supply was cheap.
It seems it does have a diode bridge and a common mode choke on the input.

Can anything be done to kill the ripple and noise to an acceptable level ?

[Simon]

So how many pins does that "bridge" have ? to be a proper bridge it needs to have 4 pins, or maybe one half of it has already failed. Maybe the capacitor is of very poor quality, something is up though.

[ItsMike]

The diode bridge has 4 pins, I have checked every diode on it with a multimeter and it seems fine.

[David Hess]

Are you measuring the output noise and ripple with no load?  Some switching regulators perform poorly without a minimum load.

[Simon]

For LED's the amount of noise is fine. If you want to power something from nice smooth power then use a diode followed by a nice filtering capacitor to isolate and filter the power for your control circuitry, you could even use an inductor too.

[ItsMike]

As I said before, I need the supply to power both the led strip and some 3.3v mcu with adc and analog circuits at 12v or 9v.
The total current for the mcu and analog stuff is slightly under 100mA.
I tried using a diode and than a couple of caps and a 7805 regulator (that's what i had on hand) with a 150mA load,  that didn't really help.

Any other suggestions for a filter ?
You mentioned an inductor, what kind of filter should I make ? Pi ? LC ?

[Simon]

You could use an LC low pass filter, inductor in series, capacitor in parallel, you can use more than one stage if necessary. You will need large capacitors after your diode to make sure they hold a charge while the input drops below the capacitor voltage and stops charging. It would be interesting to see waveforms of what you got.

[ItsMike]

I tried filtering the power supply with an random LC filter to no avail (it did nothing at all).

Next, I have removed the filter and added a Common Mode Choke (Stripped from an old ATX PSU), with a diode, 220uF and 100uF caps and a 7805 regulator.

The attached waveforms were measured across the regulator's output capacitor.
Clean as a whistle.

Thanks everyone.

[Simon]

i think the diode is doing most of the trick, try a diode with you lc filter.