1
\$\begingroup\$

I'm currently looking for low noise isolated DC-DC power supply for analog components. There are different variants available, like SN6501 IC or integrated DC-DC convertors. But I wonder, what if I use a ZVS driver (like one used for tesla coils)? Will it have lower output voltage noise due to sinusoidal current waveform? Actually something like that is mentioned in Fairchild design document:

“The Conventional PWM technique processes power by controlling the duty cycle and interrupting the power flow. All the switching devices are hard-switched with abrupt changes of currents and voltages, which results in severe switching losses and noises.”

“Meanwhile, the resonant technique process power in a sinusoidal form and the switching devices are softly commutated. Therefore, the switching losses and noises can be dramatically reduced.”

Such driver doesn't need to be hi power or hi voltage, just +/-15V for several op-amps. It will be fed with fixed voltage and will have post-regulators, so it can probably have no feedback.

So, is there any sence in such idea, and if the answer is yes, what actual schematic should I use?

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ It would be much more instrumental to think about a ripple eater and post filter after the switching converter, than to compare various switching topologies (IMO). \$\endgroup\$
    – tobalt
    Dec 16, 2023 at 8:45
  • \$\begingroup\$ All those technics will be implemented as well. \$\endgroup\$
    – Tolik4
    Dec 16, 2023 at 8:50
  • \$\begingroup\$ then I guess just pick the DCDC converter that is the simplest one to fit your specs. The switching noise is typically at 100 MHz up into the GHz range. This is a) easily taken care of by filters b) further rejected by the EMIRR of the opamps. At low power the filter components are essentially "free", so keeping the converter simple is the way to go (IMO) \$\endgroup\$
    – tobalt
    Dec 16, 2023 at 8:54
  • \$\begingroup\$ That said, if you have the components in your BOM already, then the shown circuit is relatively simple already (I assume it's self-oscillating). So if it meets your specs, go for it. \$\endgroup\$
    – tobalt
    Dec 16, 2023 at 9:00
  • 1
    \$\begingroup\$ Please add a citation and link to the document you are quoting, and a source for the image. As for what schematic, we cannot tell you without more information: what noise level do you require? What voltage and current? Efficiency? Any other features, like short-circuit current limiting, voltage regulation, isolation (how much), etc.? \$\endgroup\$ Dec 16, 2023 at 11:08

1 Answer 1

1
\$\begingroup\$

Your diagram is well documented. It has been called a ZVS Royer and has been called a tuned collector oscillator. It is sometimes called a baxandall oscillator. BJTs have been used in earlier versions. I have used similar stuff getting DC by using a bridge rectifier on the output. For high currents I have used a choke and for low current high voltage I have used a diode pump. So yes it can work for you. Remember that you do not have much control of VDC out because there is no neg feedback loop controlling duty cycle. When Vin has a large prospecting fault current like say a car battery then you can place a resistor in series with the pos Vin to defend T1, T2 I have done this on simple low power where a resistor is not a problem. I have prebucked for high power and/or high accuracy.

\$\endgroup\$
1
  • \$\begingroup\$ FYI, the "Royer" moniker is a misnomer; his original paper discussed core saturation-commutated non-resonant converters without the series choke. This circuit is better called a Baxandall oscillator. (This name is often used in the community, so it's no surprise to see it repeated here; likewise, I'm just doing my part to correct that. Cheers!) \$\endgroup\$ Dec 16, 2023 at 11:05

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.