8
\$\begingroup\$

I purchased a 5 volt power supply off Amazon. It runs from 120 VAC and produces DC power. It's a switch mode power supply. If I put a multimeter on the output it always shows approximately 5.0 VDC.

I ran into extreme difficultly in using this power supply for any sort of actual project. The output is extremely noisy.

I connected some small capacitors with a value of 10 pF and 10000 pF to the output of the power supply. I would think the power supply would have small capacitors of this type in any case, but apparently not. These eliminate lots of HF noise that is coming out of the power supply. Unfortunately this noise is not really the issue.

Here is what the power supply looks like on my oscilloscope with no load.

enter image description here

Adjusting the time scale and voltage scale I saw this

enter image description here

The channel in blue is the output of the power supply. The channel in yellow is the output of a filter network I built. I used the low voltage side of a mains transformer and a large electrolytic capacitor. Here are pictures of those, although I doubt it matters at all. The inductor is wired in series with my load(if any), and the capacitor is parallel with the power supply.

enter image description here

enter image description here

I decided to test just the power supply with a resistive load. I selected a 10 ohm resistor. This should provide a load of approximately 500 milliamperes.

enter image description here

The filter network deals with some of the oscillations, but there is still an almost 1 volt spike on the output of the filter network. I tried moving the capacitor around, but it makes little difference. In fact, even with the capacitor disconnected the output does not change much.

Here is a small transformer removed from a switching power supply. I connected the 5 volts in series with the primary of this transformer.

enter image description here

And the view from my oscilloscope:

enter image description here

It seems that almost any inductor filters out the oscillation with the period of approximately 3.5 microseconds. But that huge spike the precedes the oscillation remains. In this case the power supply jumps around by over 2 volts. 2 volts on a power supply meant for 5 volts is 40%.

The interesting thing about this is the capacitor seems to make no difference. It is old, but I've tried several and gotten the same result. They all have some capacitance, although it may be slightly diminished over time.

Given the fact the voltage still swings all over the place with the capacitor, my only theory is that the circuit inside the power supply is actually shorting its own output. If the regulator on the output of the power supply just turned off, the voltage would just taper downwards because the capacitor would slowly discharge. It's almost as if the power supply is internally shorting for a brief period of time, then the regulator goes a little nuts and "rings" as it tries to find 5 volts again.

Why is the regulation of my 5 volt power supply so poor and how to address it?

Although I cannot imagine it would help, here is a picture of the power supply with its case off

enter image description here

Update:

I performed an additional test with the mains transformer as a filter wired in series with 4 resistors. One of the resistors was the 10 ohm resistor, the other three were 6 ohm. This should give a resistance of 1.66 ohms for approximately 3.125 amps of current. This does not change anything significantly in the observed output. I reversed my probes in this test, so the colors in this screenshot are reversed as well.

enter image description here

Here is a closeup shot of the "spike" as I called it.

enter image description here

I also tried connecting a 1 microfarad capacitor across the power supply while it was driving the 10 ohm load. Here is what that looked like

enter image description here

enter image description here

\$\endgroup\$
4
  • \$\begingroup\$ 1. Your ripple frequency at about 1 MHz is probably above the resonant frequency of that ginormous cap. 2. I have no idea if this applies here but some older SMPS designs were not very nice at low loads, maybe try adding a load of at least, say, 25% of its rating. \$\endgroup\$
    – The Photon
    May 24, 2016 at 2:03
  • \$\begingroup\$ A valid point, I'll add an enormous load and see what explodes. \$\endgroup\$
    – Eric Urban
    May 24, 2016 at 2:03
  • \$\begingroup\$ @ThePhoton I do not have much in the way of dummy loads for this circuit, but I did add additional test point to my description. \$\endgroup\$
    – Eric Urban
    May 24, 2016 at 2:33
  • \$\begingroup\$ RE: the tall, fast spikes, another thing to watch out for is whether your scope probes are just picking up radiated noise from the switching circuit, and the waveform on the scope doesn't represent what the load is seeing at all. What kind of scope probe are you using? \$\endgroup\$
    – The Photon
    May 24, 2016 at 16:14

3 Answers 3

2
\$\begingroup\$

A 39000uF cap and a transformer winding is way too much for a filter.

I suspect your power supply is either defective, poorly-designed, or being operated way out of spec. The spikes are happening at a rate of ~85-90 kHz, which could be the switching frequency. The higher-frequency ringing afterward is clearly due to the spikes. If you can zoom in on the spikes with your scope, it might tell you (and us) more. A link to the Amazon page or a datasheet would also be helpful.

Regardless, your options are:

  1. Try a more reasonable filter -- maybe a few microfarads ceramic capacitor.
  2. Return or replace the supply.

The Photon's suggestion of trying a larger load is also a good one.

\$\endgroup\$
7
  • \$\begingroup\$ I may not understand, but wouldn't a ceramic capacitor with microfarads of capacitance be enormous? They are after all usually just a ceramic disc with metallized plates. In any case I added a better shot of the "spike" that precedes the ringing. \$\endgroup\$
    – Eric Urban
    May 24, 2016 at 3:31
  • 1
    \$\begingroup\$ @EricUrban you can get 10uF ceramic capacitors in 0805 SMD packages - for reference that is about 2mm x 1.2mm, so not that big. But even going with something like a 10uF electrolytic in parallel with a 470nF ceramic would certainly help. \$\endgroup\$ May 24, 2016 at 3:33
  • \$\begingroup\$ Alright, I'm just not educated on the subject then. In any case I do not have ceramics in that size. \$\endgroup\$
    – Eric Urban
    May 24, 2016 at 3:34
  • \$\begingroup\$ You could try a 4700uF cap designed for SMPS with low ESR. Add a 1mH inductor rated for a few amps and that should clean the power rail up. Several ideas to try. I would say the basic power supply is ok. If the snubber diode went bad, there would be a lot of noise spikes. \$\endgroup\$
    – user105652
    May 24, 2016 at 3:36
  • \$\begingroup\$ @Sparky256, can you point out a 4700 uF cap with SRF above 2 MHz? \$\endgroup\$
    – The Photon
    May 24, 2016 at 16:12
0
\$\begingroup\$

Eric Urban, With respect it sems to me thay you do not fully understand that useful abomination the switcher? In particular do yo know how dangerous it can be to go inside.?

They do by nature have a very poor regulation because every rule is bnroken in the inrerest of small size. The best saolution for you is to purchase a switcher with (say) more than 10-volts output and follow it wibh a voltage-regulator chip that will given you 5-V at very-low o/p impedance.

That uis what is inside the larger and more expensive items.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Your first paragraph should be a comment rather than part of an answer. Your second paragraph requires some more explanation, particularly how to deal with the fact that typical linear regulators' line regulation falls off at frequencies above a few kHz or 10's of kHz. \$\endgroup\$
    – The Photon
    May 25, 2016 at 1:05
0
\$\begingroup\$

This looks like a typical AC to DC converter. 350W maybe and could be any two switch topology given two switches shown, hard to say for sure. The transformer provides isolation from primary to secondary in any case. I recommend using a differential probe when looking at the secondary side waveforms or cautiously float the scope through an isolation transformer or cheater plug. What you are seeing might be a ground loop issue (?). A agree with others that supplies typically need a load to regulate to else they may operate in bust mode or some other mode to try to generate feedback to regulate to the set voltage.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.