1
\$\begingroup\$

I bought a DPS5005 unit in order to setup a cheap power supply. The module seems to be quite popular and has often been described as remarkably good for its low price of about 35 $. However, even under load (2 A flowing through a 0.1 Ohm resistor) I get short 450 mV peaks at a frequency of 35 kHz. These are independent of the power supply that is used as source for the DPS5005.

To smooth the output, I plan to implement an LC-Filter. The formula to compute the cutoff frequency is

$$f=\frac{1}{2 \pi \sqrt{LC}}$$

with the inductance L of the coil and the capacitance C of the capacitor. As cutoff frequency, I'd like to have something in the range of 10 kHz. According to the formula, I could realize this for example using a 47 uH inductor and a 60 uF capacitor.

Is it sufficient to simply use an electrolytic capacitor? Or should I use a combination of a small ceramic and a big electrolytic capacitor? And since electrolytic capacitors in the mF-range are available, is a bigger capacitance a better choice?

Any tips/hints/empirical values?

Edit: Here are two plots of my measurement: Overview of signal

The first shows an overview. The sample frequency is too low - therefore the variance in the amplitude of the spikes. But I hope it gives an idea of the problem. One of the spikes

The second is a closeup of one of the spikes. I marked the data points by dots.

  1. Edit: According to the post of EE_socal, I have modified the probing setup and indeed, the peaks are much lower. Here is an image of the setup: image of the setup And these are the new plots of the signal: Overview of signal One of the spikes

  2. Edit: EE_socal suggested in his answer to add a ferrite bead, bulk capacitor, and small ceramic capacitor. Following this advice, I tested a 80 Ohm @ 10 Mhz, 110 Ohm @ 25 Mhz ferrite bead together with electrolytic 47uF and ceramic 0.1uF capacitors. The peaks have vanished leaving a peak-to-peak amplitude of <0.1 V, which is sufficient for my requirements.

    Here is a plot of the filtered supply voltage:

Damped signal

\$\endgroup\$
  • 1
    \$\begingroup\$ Sounds like you are trying to solve a problem that might not exist? Are you confident in your scope probing technique? \$\endgroup\$ – Andy aka Feb 28 '18 at 15:34
  • \$\begingroup\$ Why do you think that the problem might not exist? Due to the rather high current, I doubt that it's simply electromagnetic coupling from some external source. And I measured the signal differentially using a calibrated 5,000 $ oscilloscope. \$\endgroup\$ – Jan Feb 28 '18 at 16:49
  • \$\begingroup\$ The cost of the equipment doesn't mean you know how to use it correctly. A rolls royce is just as easily crashed by a novice as any other car. Please explain what you see is a problem in what you have described. \$\endgroup\$ – Andy aka Feb 28 '18 at 17:44
  • \$\begingroup\$ Maybe you can post a photo of your probing setup and measurement, As far as the LC values more information is needed as to the frequency content of the noise. A scope picture would be helpful. You might be better off using a ferrite bead instead of an inductor. \$\endgroup\$ – EE_socal Feb 28 '18 at 17:48
  • \$\begingroup\$ Be careful with an LC output filter. Load transients near your resonant frequency can cause ringing, potentially causing damage to the supply and whatever is connected to it. 450mV peaks would be unusual, I suspect @Andyaka is correct with his comments about probing technique. \$\endgroup\$ – John D Feb 28 '18 at 18:01
1
\$\begingroup\$

The best way to probe would be with a very short ground spring tip, something like this:

enter image description here

Let's assume the noise spike you measured is really there or at least some it is there. The width is very short on the order of nano-sec. This is typical of noise originating from very fast edges on the switch node and coupling to the output through the parasitic capacitance of the inductor. The best filter for this is a ceramic capacitor and ferrite bead as they are effective at these frequencies and the ferrite bead is largely resistive which reduces the chance of ringing. However I don't know how you plan to add such a filter to the off the shelf power supply you are using.

\$\endgroup\$
  • \$\begingroup\$ Thanks, you helped me a lot! I modified the probing setup according to your suggestion. The measured spikes are much lower but still in the range of 0.3 V peak to peak (see my edited post). Since the DPS5005 is only a frontpanel, I have to build the housing myself. Thus, it's not problem to add some additional parts. Can you give me an example of a ferrite bead - capacitor combination that would be suitable for this specific case? \$\endgroup\$ – Jan Mar 1 '18 at 9:51
  • \$\begingroup\$ It depends on the voltage and current. In a design I did I used a ferrite bead that was 33 ohms @ 100 MHz and rated at 6A. The capacitor should be ceramic with the proper voltage rating. I would use a 0.1uF in parallel with a bulk capacitor like 47uF. The 0.1uF is good at high frequencies and bulk is to supply the load during transients since the ferrite bead will degrade the transient response. \$\endgroup\$ – EE_socal Mar 1 '18 at 15:46
  • \$\begingroup\$ I tested it (see 3rd edit of my post) and the results are quite good. Thank you! \$\endgroup\$ – Jan Mar 7 '18 at 13:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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