SG3524 switcher problem

Without seeing your construction it is difficult to say for sure. The inductor is a good place to start. It must not saturate, up to perhaps

10 amps. Can you measure the waveforms with a scope? Can you measure inductance? Did you notice the note on the schematic by the inductor that says "air gap"? (An air gap will greatly reduce the tendency of an inductor wound on a ferrite core to saturate, though you need more turns to get the same inductance, compared with an ungapped core.)

Another item: be sure the current shunt (200mV@5A = 40 milliohms) is very low inductance, and the leads to it are vanishingly short. It must be able to dissipate 1 watt.

Suggest you get someone local to you who has experience with switching power supplies to have a look at it. It may well be obvious on visual inspection, and certainly should be using a scope to look at waveforms. In a suppy like this, layout and construction practice in some areas are critical.

Cheers, Tom

Well, it's certainly a hell of a lot more critical than "4 to 40". That's a range of A HUNDREDFOLD for inductance! Not even the broad side of a barn.

It sounds like you didn't read the schematic. It states "30uH 5A" for the inductor...

You can measure inductance by moving R15 in series with the inductor and watching voltage across the resistor. When terminal voltage (of the inductor) is known (20V by the looks of it), you can measure the slope of the increase in current and calculate inductance by dI/dt = V/L.

In addition, you can evaluate the maximum current capacity by noting at what current level inductance decreases, due to core saturation (current starts rising steeply).

Average (actually RMS) current capacity corresponds to the wire gauge. You'll need reasonably heavy (circa 16AWG or better) wire to handle 5A at that frequency.

Once you have the inductor in place and you are certain it isn't the problem, you can try evaluating the feedback circuit: voltage and current regulation.

For now, I would suggest you disconnect the error amplifiers and use a potentiometer to control duty cycle (route pin 1 to pin 9, remove R3 and disconnect pins 4 and 5).

Use a large resistor (say, 15 ohms 25W) on the output to provide a load.

Tim

-- Deep Fryer: a very philosophical monk. Website:

Unfortunately a switcher is one of these things where a schematic is only half the story. The other half is the layout / 3D build and the waveforms. You should have a few waveforms to compare to as well.

Rene

--
Ing.Buero R.Tschaggelar - http://www.ibrtses.com
& commercial newsgroups - http://www.talkto.net

Looks like a damned awful design to me....

I thought the main reason for using switch mode techniques for PSU's was to eliminate the need for heavy iron-cored transformers. Yet this design still uses a normal 15V/5A mains transformer and bridge rectifier to supply the rated output voltage of 15V at 4A....

The SG3524 appears to be used only as an ancillary device to control the duty cycle of the -VE return rail for voltage control and to switch the supply off for over-current conditions.

Surely, if you are going to use a standard full wave bridge rectifier and mains transformer to supply full current and voltage, then far simpler well tried analog techniques can be used for voltage and current control, without using a SG3524.

and

application?

If I'm not mistaken the schematic seems to be using a very funky weird means of driving the MOSFET gate. In this case, funky weird is decidedly not a desirable aspect. The SG3524 has a pair of uncommitted output transistors inside, but they are designed to be used to drive power bipolar transistors. They aren't designed to be used to make a totem pole driver for driving MOSFETs directly like shown in the schematic...

Due to this bizarre gate drive connection scheme, it appears to me the circuit cannot produce more than roughly half duty cycle. This would explain why you can't get more than around 6V output with 13V in. In an ideal buck converter operating in continuous conduction mode the output voltage is Vin*duty cycle = Vout. Since the gate drive connection scheme prevents large duty cycles, the maximum output voltage with any kind of decent current capability is limited to roughly half the input voltage.

To fix this, the SG3524 output transistors should be "wired ORed" (connect collectors together, emitters together, collectors to +12V, emitters to pull down resistor) and the output should drive a MOSFET gate driver IC such as the TC4427 by Microchip.

Given the nature of this conceptual error in the schematic, I wouldn't be surprised if there are other hidden problems that I have not noticed in it as well.