2
\$\begingroup\$

I Designed a 5v,7A power supply from 24v input for my new project. It is giving 4.9v, but have noise issue. When I increasing the load its noise level is also increasing.

I connected 1A load and a rapberry pi at the output. Is it an acceptable level of noise.

I am attaching the schematics and output waveform, Kindly check it and suggest me a solution.

Waveform with raspberry pi + 1A load

Schematics

\$\endgroup\$
3
  • \$\begingroup\$ What's the switching frequency? Can you zoom in in time on your noise? How much dead time do you end up with, with those transistors? Show L1.1 waveform too. How long from C6 to Q1? \$\endgroup\$
    – winny
    May 31, 2017 at 14:30
  • 2
    \$\begingroup\$ Probing is important here. Take a look at this little guide first. Also, please widen Time/div so that we can see the waveform better. \$\endgroup\$ May 31, 2017 at 14:31
  • 2
    \$\begingroup\$ Don't ask us if it's acceptable. We don't care. If it suits your overall design then it's fine. \$\endgroup\$ May 31, 2017 at 14:38

1 Answer 1

1
\$\begingroup\$

Noise seems high although as some have suggested if it works, why fix it? The Pi will have it's own bypassing capacitors that mitigate this noise. You're clearly using a commercial /professional switch chip rather than some Frankenstein 555 timer creation. The chip's data sheet has layout recommendations towards it's end. These are pretty important due to the large switch currents flowing around (and beneath) the chip. Do these recommendations match yours?

I'm assuming that your circuit is lifted straight from the data sheet. So you might expect the same ripple /noise that it gives. I can't read the screen sufficiently on my device to ascertain the noise, but is it ~30mV @ 20 MHz? I'm not sure I believe this as data sheets can offer representative information. The data sheet extract below is from this commercial PSU outputting 5V @ 6A. Ripple is 120mVp-p, higher than your chip's datasheet, but still significantly lower than your results.

enter image description here

If you can't reduce the noise to an acceptable level though layout improvements, you might consider anther stage of LC filtering. You'd only need uH stuff as you'd be looking to smooth frequencies in the order of 100kHz - MHz.

I can't see from your 'scope grab, but do you have 20 MHz bandwidth limiting on? You need to have that set to usefully compare noise readings between PSUs.

I also suggest you watch Dave's video regarding correct oscilloscope probe technique for measuring PSU noise. You really should do it differentially as in the video. It is likely that you'll see improved results with a better probing technique as you should be able to compensate out at lot of extraneous noise.

Personal statement: I never build switch mode power supplies unless I'm specifically trying to learn how. If I ever need a switcher, I always buy off the shelf. There's a lot of analogue /EMI /ground loop theory to absorb and digest to minimise the noise they're notorious for. Once you factor in the components, especially those weird specialist high frequency transformers and a commercially manufactured PCB, the cost differential is small. Plus you can consider any extra cost as an insurance premium for ruling out odd behaviour from an uncertain DIY switcher. Just sayin'.

\$\endgroup\$

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.