Switcher Inductor Ripple Current

I need to design a constant current buck switcher. I need to power a Perkin Elmer krypton long arc lamp, a QCW12. At the middle of the lamp's V-I curve, near where I want to run, the lamp voltage is

155V DC @ 20A. I want to run at 80-100 Khz off rectified 220 line. I've serviced these systems since 1988, so I have a idea what needs to be done to support the lamp, but I never ripped out a customer's inductor and measured it.

For the curious who will ask, why do you want this?, the lamp drives a Nd:YAG laser that is frequency doubled to 3.5 watts of green light. Since the next question will be, jeeze, thats a tought design, why don't you just buy one? There is only one remaining maker of these supplies in the US and my lamp is too small for their unit, which is designed for a 40 amp 160V lamp. Everything else is solid state, pumped by laser diodes, these days.

These lamps start degrading from minute one, so the simple answer of using a series resistor is foolhardy, if you dip into the negative slope portion of the lamp I/V curve, the system oscillates and the lamp explodes, taking the laser rod with it. I've tried it, I didn't blow the rod or lamp, but the output power is all over the place.

As I look at various on line tutorials, I keep seeing a constant for ripple factor for picking the inductor. I've looked at three different tutorials and have seen three different numbers. The one that makes the most sense so far is .4 x the DC design current, or 8 Amps.

Anybody have a known proven rule for finding a value for this constant?

Thank you for your time,

Steve Roberts

Reply to
Loading thread data ...

Pull it out of your ass?

Seriously, there's a lot of different factors that go into choosing the number. The optimum is broad, and what's best and most economical for you may not be so for the next guy. So you're (a) free to choose from a broad range of options, and (b) under different constraints from everyone else anyway.

I'd try a number of different ratios and see what impact it has on my circuit cost and estimated reliability. Then I'd choose one, serene in the knowledge that no matter what, I've screwed up somewhere else in the circuit anyway, so it's not like I have to worry about destroying perfection with my inductor choice.

Reply to
Tim Wescott

I've got 4 150$ lamps and no money to replace them. I'd like two remaining when I'm done. One in the laser and one spare. That is the problem and why I want to get it right. They show up on ebay every once in a while for 50$, but I'd need to order 10 pcs or so to have a run done, so it does NOT PAY to blow lamps.

Everybody else, "pulled it out of their ass" with no justification.

So far it looks like 180 millihenry for one fet string, or 360 for two fet srings in parallel.

There is enough other fun for things to go wrong, the lamps need a cap charged to 1000V, limited by 40 ohms, to aid in starting and a 27 Kv ignite pulse in series with the lamp.

Reply to

Ripple current isn't a constant.

It's explained in most controller datasheets how you select an acceptable ripple and the impact the ripple has on the converter. Transient response,cost ,size,component stresses etc.

See here near the bottom of the page.

formatting link

and here TPS40200 VMC controller.Pg 21 (tps40200.pdf, 1143 KB)

formatting link

If it's still not clear check out more controller data sheets.

Reply to

Otherwise known as trial and error.

It took several iterations on a .3 inch toroid before we found that 3 was bad, and 4 was good, and 5 was bad (turns), and we had to step through a few different types of core media as well.

Litz wire made it operate better still, and it was only at 57kHz.

Reply to

Inductor ripple current will reflect in the output ripple, as current or voltage dependant on the type of regulator and the total filtering method.

It will also reflect in inductor flux swing and AC flux induced core losses contributing to temperature rise in this component.

Different ball-park figures reflect the different core materials and application voltages and operating frequencies that the figures are originally applied to.

Flux swing is a product of applied voltage and operating frequency.

deltaB = V . t / ( N . Ae )

delta B is peak to peak flux swing, in Teslas V = applied voltage, in volts. t = period of applied voltage in one switching cycle, in seconds. N = number of physical wire turns supporting the applied voltage. Ae = core cross-sectional area of the core structure, in meters^2.

The peak flux swing vs loss per volume, over a potential range of operating frequencies, is one common chart of material properties provided by core and core material sources.

Core losses

- reduce as turns increase

- reduce with volt-seconds-per-turn

- generally increase with frequency for most materials.


Reply to

Ah, now we get some decent math!


Reply to

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.