4
\$\begingroup\$

In my circuit, I use a R-78E3.3-0.5 to generate the 3.3V supply voltage for a microcontroller from a 24V rail. The total load draws about 10mA. I'm using a 47uF decoupling capacitor at the regulator output and some smaller decoupling capacitors at the supply pins of the uC which already reduce the noise quite a bit.

There is still a significant amount of noise left which I would like to further reduce and thought about using a second stage LC filter for that purpose (a first stage LC filter is included in the R-78E3.3-0.5 if I understand correctly.)

I'm getting a switching noise at 570kHz (consistent with the R-78E3.3-0.5 datasheet) and a ripple noise at around 7.7kHz as can be seen in my oscilloscope measurement. Channel 1 (yellow) is the same signal measured at full range.

Supply Voltage measurement

All the instructions I can find talk about attenuating the switching noise with a LC filter set at a cutoff frequency of around 1/10 the switching frequency which would be 57kHz in my case. That would surely help with some of the noise but the ripple at 7.7kHz would be largely untouched, right? An LC filter with a cutoff frequency of 1/10 * 7.7kHz would require a really large inductor.

So my questions are:

  1. Am I overall on the right track to tackle this problem?
  2. Is perhaps the R-78E3.3-0.5 a bad choice in my application because I only draw 10mA of its potential 500mA? Would the R-78E3.3-0.5 produce a higher (and thus easier to attenuate) ripple frequency in an application with a higher load?
  3. Is it common to attenuate low ripple frequencies like my 7.7kHz from a switching regulator?
\$\endgroup\$
5
  • \$\begingroup\$ For what it's worth, it may be worthwhile to use a 4V switching regulator, then use a LDO regulator after it to drop 0.7V. 0.7V*10mA = 7mW loss. \$\endgroup\$
    – rdtsc
    Feb 2, 2022 at 13:08
  • \$\begingroup\$ I thought about that as well. 5V with 1.7V LDO drop, since 5V switching regulators seem to be more common. I'm still unclear whether the LDO output would be relatively noise free despite its input being affected by the burst mode ripple from the switching regulator. Would that be the case? Sounds like it's definitely worth experimenting with though. \$\endgroup\$ Feb 2, 2022 at 13:21
  • 2
    \$\begingroup\$ If you take a look at something like the MC33375, look at the PSRR (power supply ripple rejection) for a dB value. For this device, it will be ~75dB for 120Hz. Page 7 shows ripple rejection for other frequencies; looks like 7kHz will be about 55dB. Digital-only circuitry likely wouldn't need the LDO at all. \$\endgroup\$
    – rdtsc
    Feb 2, 2022 at 13:46
  • \$\begingroup\$ Sounds great. I have a bunch of LD1117's lying around. They seem to perform similarly at 120Hz. There is no graph in the data sheet for PSRR though but I will try them. My circuit includes an ADC reading the value of a pot. So it might make sense to reduce noise. \$\endgroup\$ Feb 2, 2022 at 14:11
  • 1
    \$\begingroup\$ This question may also be helpful. TL;DR: If the ADC has a stable reference, PSRR is irrelevant. :) \$\endgroup\$
    – rdtsc
    Feb 2, 2022 at 14:45

1 Answer 1

6
\$\begingroup\$

Is perhaps the R-78E3.3-0.5 a bad choice in my application because I only draw 10mA of its potential 500mA. Would the R-78E3.3-0.5 produce a higher (and thus easier to attenuate) ripple frequency in an application with a higher load?

You can see this from the waveform - it's going into burst mode and the burst repetition rate is 7 kHz. The little spikes on the rising ramp of the voltage are your actual switching waveforms then, when the output voltage reaches the regulation point, given the low current draw, the device shuts down into burst mode operation.

This is fundamentally a problem of that device and your low load current. You should be able to find many designs that can overcome this but, it's less likely you'll find an off-the-shelf device that avoids burst mode on such a low load current.

Is it common to attenuate low ripple frequencies like my 7.7kHz from a switching regulator?

In my opinion I would say not. You design the switching regulator (or choose the device) so that burst mode is avoided.

\$\endgroup\$
0

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.