1
\$\begingroup\$

I'm trying to design a cheap way to power my DIY Eurorack modules. I'm trying to use two 15V wall warts to that effect but I'm getting a lot of noise. I've seen a few questions related to noise induced by PSU, but I think mine is different. So here are a few pics of my design, what am I doing wrong? enter image description here

enter image description here

Here are some scope measurements:

enter image description here

enter image description here

enter image description here

\$\endgroup\$
22
  • 1
    \$\begingroup\$ Does the noise have a distinct frequency? What do the Eurorack modules do? \$\endgroup\$ Commented Mar 8 at 19:03
  • \$\begingroup\$ Is it possible that what you describe as noise, others would describe as signal? Why not? \$\endgroup\$ Commented Mar 8 at 19:06
  • \$\begingroup\$ @AndrewMorton By noise I mean a signal added to the signal I send \$\endgroup\$
    – brunobhr
    Commented Mar 8 at 19:44
  • 1
    \$\begingroup\$ @AndrewMorton The modules work fine when powered with my Rigol DC lab supply (no 'noise'). The modules are fresh built with new components \$\endgroup\$
    – brunobhr
    Commented Mar 8 at 20:12
  • 1
    \$\begingroup\$ I built a Eurorack module with a linear (transformer, bridge rectifier) PSU. Simple and not particularly costly. Generic wall warts are not going to give you a low noise solution. \$\endgroup\$
    – Ian Bland
    Commented Mar 9 at 5:57

2 Answers 2

3
\$\begingroup\$

You seem to be using switch mode wall warts as power supplies.

These output a lot of electrical noise, both differential mode and common mode.

One solution is to buy anothet set of wall warts that are not switch mode power supplies. Another solution is to make your PCB to be more complex and expensive to filter or smooth away the noise.

For this purpose, the regulator input capacitors are too small. You need few tens if not a few hundred microfarads, preferably after a resistor or inductor to have an RC or LC filter to remove noise.

As electrolytic caps are quite poor at high frequencies, having one larger electrolytic and one 100nF ceramic for filtering high frequencies would be much better. Unless you buy polymer electrolytics.

The PCB wiring is also quite narrow, wider tracks would pass more current with less voltage drop. It could be a lot better. It's two-sided, with both sides mostly etched empty from copper that could have been used to make a proper ground plane and wide tracks to easily carry a lot of current.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks justme! I will try changing caps values and try it out. I'll let you know! \$\endgroup\$
    – brunobhr
    Commented Mar 8 at 21:23
  • \$\begingroup\$ I tried with different combination of caps values, but couldn't get to a satisfying result. So as suggested by @Ian Bland, I might just try the linear and bridge rectifier way. Thank you again \$\endgroup\$
    – brunobhr
    Commented Mar 11 at 18:48
1
\$\begingroup\$

Wall-warts are classified as SMPS(switch-mode power supply). These are different to linear PSUs in that they waste very little power as heat(more efficient) however this introduces a LOT of noise into the output. Your 78/79xx linear regulators will attenuate some of this noise, however higher frequency noise will go straight through. As a general example, the STMicroelectronics 7805 has the following ripple rejection plot: enter image description here

This may look pretty good, however is only good for low frequency noise(e.g. 50/60 Hz powerline) up to about 100khz(10^5 Hz), at which the attenuation starts dropping. Unfortunately, SMPS supplies typically have a lot of noise in the MHZ region, and are not filtered out by the 78xx regulators very well.

A solution to this is to add a LC power filter between your regulators and the wall-wart to filter out this HF noise, with a ferrite bead instead of an inductor as the L element in the filter(beads have the property of having a higher resistance at increasing frequencies, starting at around the 1 MHZ range up to 100MHZ). Rigorous testing of this filter is required with some sort of simulation program(LTspice, etc) for optimal frequency response, it depends on how clean you want your supply.

\$\endgroup\$
1
  • \$\begingroup\$ If you look at the scope, the ripple voltage has a frequency of around 500 kHz as measured by the scope.and based on the pulses on the waveform, around 1 MHz. Please explain how a ferrite bead will remove such a ripple, as they are generally rated with impedance at 100 MHz, which is 100x higher. Which kind of ferrite bead you mean, maybe you mean a different component called a ferrite bead than what I think when someone says ferrite bead. \$\endgroup\$
    – Justme
    Commented Mar 9 at 10:25

Your Answer

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

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