Hi,

I'm building a CV/CC power supply. I have the linear regulation part working already, but i want to use a switching pre-regulator to keep the voltage feeding the linear regulator close, as to avoid huge power dissipations.

My problem, is that i have never built a switching regulator, can someone point me in the right direction? I readed around the net about this, and i think i'm going to use a wall adapter to supply the input voltage (as to not have to mess with rectified AC directly, and try to dodge some of the EMI problems), followed by a buck stage, but i'm open to suggestions on the switching mode, and frankly everything . The good news, is that using a linear regulator means that the SMPS doesn't need to have low ripple.

I doubt i can find a pre-built inductor for this kind of use, i'm probably going to need building it myself, but hopefully the materials for that are available. It's another thing i never did before, however: making my own inductor.

Any tips are welcomed.

The needed parameters for the SMPS pre-regulator are:
Vout = 4 to 35v
Iout = 3A (continuous)
Ripple = not very important

Thanks for any info.

 

Go to www.national.com

You can find many switching regulator ICs there and also you can find inductors for them (RS, Farnell, etc), not need to build one.

 

National's Web Bench is a pretty good way to get a design quickly that will meet your specs, they even give you part numbers and suppliers, and they can even put kits together for a number of designs.

Linear Technology has their SwitcherCad software; a part of LTSpice.

You could try to design your own, but be prepared to spend a LOT of time on it.

If you want to experiment with some low-power switching supplies, check out Ronald Dekker's page:
http://www.dos4ever.com/flyback/flyback.html
The supplies on his page are more educational than anything else, but they are cheap and easy to build, not to mention rather fun.

 

Thanks, i'm going to check the National site, maybe they have official distribution here, and i can skip a lot of trouble.

I had seen their simple switcher IC's, and i must say that i like the simplicity of the circuits.

 

You may find that the more simple the circuit is, the less efficient it is.

If you're going to use this supply a good bit, you want it to be efficient.

Don't forget that your linear regulator will add quite a bit of inefficiency to the back end of the supply.

 

You may find that the more simple the circuit is, the less efficient it is.

If you're going to use this supply a good bit, you want it to be efficient.

Don't forget that your linear regulator will add quite a bit of inefficiency to the back end of the supply.

Thanks for the advice. It's for a bench supply, so it's isn't really that important, as i don't intend to leave it connected constantly, and they guys at national claim very good efficiency over 90%, which should be good enough, even if i throw some of that down with a linear regulator.

I think it's that simple, because it's a buck regulator, and the IC has the power mosfet integrated.

 

You might find the attached simulation interesting.

It's a 2.5A, 1.5v to 30v buck supply using an LT1171 switching regulator. Better than 90% efficient when stepping down from 34v to 8.25v. The lower in output voltage you go, the less efficient it gets, but it's still far more efficient than a linear regulator could approach.

It could be upgraded to a 5A switcher by substituting an LT1170 5A switcher (Digikey stocks them for around $10) a 50uH inductor capable of 10A peak, and a 3.9uH inductor rated for 10A peak, along with similar cap/diode upgrades.

I've also been experimenting with a negative buck/boost supply from the same positive source; I've managed to get about the same efficiency from it; much better than the example schematic in the datasheet.

 

I'm surprised to see that kind of output from a simulation. I thinked the noise in the output of a switching regulator would look much more like a repetitive signal. It's very low amplitude, however.

I will ask the guys at Elko, which are supposed to be distributors of LT here, but one time i asked them for an LTC3490 led driver, and they didn't have it.

I'm still gathering information on who has the inductors. Sadly, it's that difficult to find them here. The best i could find by now is an enterprise that makes inductors, and switching supplies, but they don't sell in low quantities, and they even say that their ferrite cores are good up to only 100 KHz. Oh, well, at least i'm not in a hurry.

 

The reason that the output isn't a repetitive signal is due to L2/C6/C7 being used as an output filter, and the additional delay in the feedback loop. Without the output filter, there is quite a bit more noise.

My favorite local shop, "Skycraft Parts and Surplus" here in Orlando FL has a fair assortment of toroids. I've been picking up a few each time I stop by, winding on 10 or more turns and testing their response.

I've found some there which seem to be quite good candidates; 38 turns of AWG22 magnet wire gives me 50.6uH and fills the toroid nicely in a single layer; they have an AL of 34 to 36, and they are 16.9 OD x 9.9 ID x 0.69 H (mm).

Just because toroids might seem to be hard to get does not mean your dead in the water. You can salvage usable toroids from junked computer power supplies, and re-wind them to suit.

See Ronald Dekker's site:
http://www.dos4ever.com/flyback/flyback.html
The particular page is called "Flyback Converters for Dummies", but that's only in jest. It's really a great resource for those interested in learning more about switching power supplies, and experiments they can actually perform using pretty commonly available materials. Fair warning though; a decent O-scope is a must.

I made a different version of an inductor tester from a 74HC04 hex inverter and some caps & resistors; quite useful to get the basic uH of the inductor without any saturation info. Ronald's setup will give you the saturation information as well as the uH value; if you have a 'scope with a reasonably accurate cursor.