0
\$\begingroup\$

I'm building a sensitive analog circuit powered by a wall wart. I'm using a DCV01 family DC/DC converter with dual outputs (+/- 6.5V) followed by linear regulators. The SYNC pin of the DCV01 converter unintentionally generates a signal that transmits wirelessly...the package's pin acts as antenna I suppose. Bringing a wire near it allows me to see the SYNC pin timing on an oscilloscope. While I have filtered the output of the DCV01 as suggested (I think), some of this periodic ripple (looks like a wavepacket every 1/400MHz) remains. The amplitude is millivolts--way over my noise budget. My circuits operate on microvolts.

I'm attaching an image of the circuit design. It's pretty trivial. How can I get rid of this signal? Is there a standard filtering approach that I should have impelmented?

enter image description here

\$\endgroup\$
  • 2
    \$\begingroup\$ Beware of EMI and use a low noise shielded linear supply with low noise Instrument Op Amps (INA) with balanced twisted shielded pair and active Guard. Get rid of EMI by identifying radiated noise and isolate it or shield it with scope shorted loop and measured conducted noise on supply and fix all sources of noise. \$\endgroup\$ – Sunnyskyguy EE75 Oct 20 '16 at 0:21
2
\$\begingroup\$

It looks like you have chosen not the best part for the job.

First, the converter is UNREGULATED, so you will have substantial output variability when load changes. It may or may not be important however.

Second, you definitely have no use of SYNCOUT signal. From the block diagram it looks like its amplitude is huge, and there will be very difficult to avoid EMI. Even if you will use heavy L-C filters on power rails (since the ripple level is 50mV by specs), it will be difficult to shield the susceptible electronics from over-the air EMI.

I am afraid you need either to make another selection, or shield the entire DCDC on all sides, with all care for ground layout. In any case you will likely need to spin the PCB.

\$\endgroup\$
  • \$\begingroup\$ Okay, thanks for the tip. Any ideas on good methods for generating very low noise +/- rails from a single wall-wart? \$\endgroup\$ – nick_name Oct 20 '16 at 4:06
  • \$\begingroup\$ One usual idea is to separate switchers from linear regulators and sensitive electronics by a long cable (shielded of course). Also get a better ripple rejection with LC filter and better LDO. LM317 and LM337 are a little old style. \$\endgroup\$ – Ale..chenski Oct 20 '16 at 5:33

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.