Switching Lab supply design

[David Hess]

4. And, you can design a filter network for additional filtering with OC/SC damping, either by RC damping an intermediate node (the middle C in a CLCLC filter), or damping the L (probably not with ESR because of DC losses, but L || R or L || (R+L) can be used).  Or making a constant-resistance filter.

A constant resistance filter?  A t-coil?

[T3sl4co1l]

Nah, just a regular filter with compensation networks to present a constant resistance -- which by extension, eliminates sensitivity to loading (or not) on that port, since, after all, there's no reactance to react against.  Example:
http://jeroen.web.cern.ch/jeroen/reports/crfilter.pdf

You can also build one by making the dual of any given filter, and connecting them in parallel or series; this isn't always obvious, because the general dual may require as many components as the filter proper.  But as can be seen above, for the lower-Q prototypes at least, a simple network gets very close.

The special case for a Butterworth (or for Linkwitz-Riley if you're into that sort of thing) is self-dual (same values, I think?!, but obviously transforming one to HPF and the other LPF), and it's just a simple diplexing filter.  Which of course you can terminate the HP port for this use.

Tim

[David Hess]

But a t-coil can provide for a constant resistance load and source.  The disadvantage would be that it requires a tapped inductor, so not a standard part.  It just never occurred to me to use use as the filter for a switching regulator output, and I have never seen it done.

[T3sl4co1l]

I'm not sure what application a T-coil has to filtering; usually it's used to extend rather than restrict bandwidth, lol.  The classic case being, a capacitive load on an otherwise-resistive bus (e.g. op-amp input from an analog video line), or a resistive load inbetween a capacitive source and load (ye olde T-coil vertical amps).  But I'm not very familiar with the wide space of tapped/transformer coupled filters; I'd need to see an example.

Dual coils of various ratings, values, and limited ratios (1:1, 1:2, etc.), are not uncommon these days; not usually with the kind of coupling factors you might use for filters (i.e. 0.2-0.8 say?), but maybe that can be designed around, or of course one can add "leakage" externally.

Tim

[David Hess]

Extending bandwidth is the usual application for a t-coil, but it is not the only way it could be used.  I suspect nobody bothers because it involves extra design complexity and understanding, where multiple CLC sections are sufficient and easy.  The only reason it occurred to me is the similarity with multiple LC sections.

[elektryk]

I'm also interested in building similiar CC CV PSU.
I've got already a piece of FW which drives 12bit SPI DAC and now I'm looking for power stage which can be externally controlled with DAC.
After familiarizing myself with mentioned circuit diagrams from Elektor and Linear Technology, I think they look promising. But can anybody recommend something similiar but without specialized ICs or HV opamps?

[Vovk_Z]

I use the bench supply (made by myself) with a dc/dc pre-regulator (xl4016e1) and linear output regulator. Works ok. Output noise is <1.0 mV RMS (5 Hz ... 5 MHz range). Primary supply is a typical 50 Hz transformer with a Shottky rectifier of cause, because a typical hard-switching HV AC/DC really generate a powerfull common mode noise.

[5U4GB]

I suggest you have a look at the EEZ BB3 supplies; they have a switch mode pre regulator and then a linear constant voltage / constant current stage.

Didn't Silicon Chip mag (Australia) have a design like this within the last year or so?  I've had a quick look and can't find it but I remember they used that configuration.

In general, going with an electronics-magazine design and modifying it if and as required will be a lot easier than doing a greenfields development yourself, particularly since, at least for Silicon Chip, they tend to go into quite a bit of detail about design decisions and benefits and drawbacks of various options.

[elektryk]

I also thinked about chinese DC-DC module as a pre-regulator, XL4016 is indeed one of them.
Vovk_Z can you recommend any solution to linear part of the PSU?
I played a bit with npn transistors driven by opamp but even in LTSpice simulation oscillations are present.

[Vovk_Z]

I used a more powerfull versioms of LM317 - LM338 etc.
You may tie up a pre-reg with a post-reg using a small MOSFET or 817 optocoupler.
For example, look at LT3080 'Typical applications' in a datasheet.

[Vovk_Z]

Here in attachment is a not very accurate design of my switching Lab supply, which I use several years by now.
Upd.: This circuit has too much inductors (ferrite beards were intended mostly) but a real prototype showed at least half can be eliminated or almost all except main ones. Basically, with a good PCB design you need only two inductors: one main inductor for DC-DC of cause, and one more L <=5-10% of a main as a post-filter.

[dobsonr741]

What is the minimum current limit you’ve achieved, and how quick is it? Asking as limiting by the switcher is unusual, and I would assume slow.

[Vovk_Z]

Minimum limit current is somewhere 10-15 mA. Current limiting speed is average, not too slow. Of cause it will burn sensitive parts like LEDs etc because you still have to have some amount of capacitance at output (because you need filtering)..

[5U4GB]

Didn't Silicon Chip mag (Australia) have a design like this within the last year or so?  I've had a quick look and can't find it but I remember they used that configuration.

While looking up something else I found the most recent one, September 2023, but it's a conventional LM317-based linear supply.  Still, interesting reading because they give several pages of design details and performance figures, and its fanless via the giant heatsink out the back.