1
\$\begingroup\$

I am following up on my earlier question. I am trying to build a very clean 5V and 3.3V supplies. I have asked this question and got several very good answers, we designed it and it will come back soon, I will let you know.

The new question came from a colleague who suggested using two switchers one for 3.3V and other for 5V. He indicated this way, the efficiency would be much better. I wasn't sure since this would also introduce a second noise source to the board. What do you guys think? If cost and space is not an issue, where the only thing you care is noise, would you go for 2 switchers?

\$\endgroup\$
3
\$\begingroup\$

I've used multiple SMPS in these situations. They're called for when power efficiency is the main concern, noise a lesser concern, and the power needed for each voltage is within a magnitude.

Your question, however, says that noise is the main consideration, in which case I wonder why you're bothering with SMPS at all. If noise is more important than efficiency, then don't use an SMPS at all and live with the high power requirements.

Now, if you want to do this right, you need to start from your load, and the signals within, and do a proper noise analysis, with the goal of defining an acceptable level of noise. Then work this back through the PSRR (use the graph vs. frequency, not the headline numbers) to see how much noise your power supply can generate. If the acceptable power noise level is greater than or comparable to the switcher's output ripple and switching noise, then you can use the SMPS.

If you need less noise, look at using linear regulators: my favorites are the TI TPS7A4900 and TPS7A3000 series, because they provide a lot of PSRR (50 dB or so) out to hundreds of kHz. Then see what the acceptable noise level on their input is, and if the SMPS is less than that. Switching noise (the spikes at the corners of the ripple voltage) can be dealt with by inserting ferrite beads in series with the current before each filter capacitor (not in series with the capacitor itself).

\$\endgroup\$
2
\$\begingroup\$

Your question supposes that both 5 V and 3.3 V power is really needed? Why? If I remember right, this is a battery operated device so presumably battery life and therefore efficiency is a issue.

My first reaction would be to design the circuit to run directly off the battery. Many microcontrollers and opamps are specified over voltage ranges suitable for direct battery operation. Why can't you use these? If some component you can't control requires either 5 V or 3.3 V, can you design the rest of the circuit to run from the same power voltage?

You should step back and explain more what the overall problem is. It seems to me you made some high level choices early on and are forgetting these can be revisited now that you're working on the details.

\$\endgroup\$
2
  • \$\begingroup\$ Well said. Our lack of trust to 3.3V opamps, bias of photodiode with 5V and driving lasers with 5V are the main reasons for 5V. Don't know we can eliminate all of them and move the system to 3.3V where I can use multiple batteries and LDO. \$\endgroup\$
    – Ktc
    Oct 3 '11 at 0:54
  • \$\begingroup\$ BTw, we end up sticking to single SMPS since we didn't want to take risk. \$\endgroup\$
    – Ktc
    Oct 3 '11 at 0:55

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.