LM2576 Adj 'switcher'

[Davy]

Wonder  if anyone knows the regulation % of a LM2576- adj 'switcher regulator'. I get  1.1v drop on the output between no load and full load.

On a DC supply I can set the voltage to 11.6 and it draws around 2.14 Amps, on a switcher it drops to 10.5V, the circuit anf board layout is as shouwn on the datasheet.

This is for a uV LED light panel I've built, is this voltage drop normal, the data sheet seems vague to me, may be I need to 'up' the on-load voltage, rather than set it off-load. Thanks,

Dave.

[MagicSmoker]

Not enough info - post the input voltage and specific inductor you are using.

[T3sl4co1l]

1.1V drop sure, which is how many percent of nominal output?

How much of that is taken up in the first few percent of rated current -- does it need a "bleeder" load to meet spec?

As I recall, the internal error amp has unimpressive gain (about 80x), so the regulation is poor as regulators go, but more than adequate, and 1.1V would have to be out of a total 40V+ nominal output to be due to regulator operation.  So I suspect there may be a leakage current, or minimum duty cycle limitation, that's causing your output voltage to rise when very lightly loaded.

Or if it's because voltage is falling below nominal, at high currents, you might not have enough supply voltage (maximum duty cycle and dropout limitations), or you're using inferior components (inductor, diode, capacitors) that are costing voltage in a somewhat roundabout way.

Tim

[Davy]

Using a 200uH inductor wound with 55 turns of 24SWG, measured on a Peak LCR meter, SR560 diode with two 1000uf 25V capacitors Panasonic FR series, low ESR, the input voltage is 15V.

Now then let me explain.
I repaired many a Nikko Baby 10 TVs', these are small portable TVs in the CRT days, they used a custom block regulator that ran very hot and was often failing, they replaced it with one of these regulators that ran from a 12 to 15V supply the output being 10.7V if I can remember correctly..... thats where I got the inductor size and used the same  yellow core  with a white edge, that had 55 turns on only used a single 1000uF, can't recall the current consumption but was well over an amp, thats how I came to build this circuit since it would run cooler than a linear regulator,  from memory it used a 1K to ground and something like 7.8K to o/p side.

I might be able to get my hands on a spare Nikko Babay 10 'switcher' then I'll do a comparison.

The PCB tracking is as per regulator data sheet. Don't get me wrong it does work and very well and cool indeed just that the 1.1V drop set me wondering, once set the voltage stay's 'put'.

Dave

[MagicSmoker]

Using a 200uH inductor wound with 55 turns of 24SWG, measured on a Peak LCR meter, SR560 diode with two 1000uf 25V capacitors Panasonic FR series, low ESR, the input voltage is 15V.

Erf... #24SWG is way too small a wire for that amount of current, and 200uH is way too much inductance for this current and switching frequency (~52kHz). You need to triple up on the #26 and also cut the number of turns in half (resulting in approximately 1/4 the inductance). Also, the yellow/white core is almost certain powdered metal so expect high losses in it, too.

Finally, the guaranteed minimum peak current rating is only 3.5A for this device, so expecting 3A nominal output current from it might be a bit optimistic. I'm surprised you say it is running cool, since the "typical" voltage drop across the internal switch is 1.4V so that would result in ~4W of dissipation right there.

EDIT- strikeout applied to where I mixed up #24SWG with #26SWG

[Davy]

The inductor is only 24SWG on the Baby 10 version, I measured it a long time ago when I repired  Baby 10 telly's, I used a proper wire guage disc.

Soon I'll have my hands on a Baby 10 regulator, they regulated at 10.7V  for 12-15V DC in, don't knoiw what their original block was like, it was all epoxied in a metal box.... they ran quite hot.

Thanks MagicSmoker for the reply, got plenty of cores I can try. Many of the circuits in the data sheet seem to aim at about a 100uH, gotta admit I didn't look the data up for this yellow/white core, I just copied the design.... apart from the 'top resistor'. I'll have a play later.

Dave.

[MagicSmoker]

The inductor is only 24SWG on the Baby 10 version...

Oops... I had #26 on my brain when I wrote my previous reply because that is the color code for Micrometals iron powder mix #26 (yellow/white); #24SWG is not too far off the mark for 50kHz and 3-4A RMS current so the wire gauge is fine. The #26 mix core is okay for use at 50kHz as long as flux density swing isn't too great (which is typically the case for boost or buck inductors, and not typically the case for buck-boost inductors or flyback transformers).

Note that too much inductance isn't a problem in and of itself, rather, it requires a much larger core to be able to withstand the amp*turns, and if the core is saturating early because of too many amp*turns then that could cause the current limit to trip too soon which reduces the output voltage. See how this all ties together? Welcome to the wonderfully weird world of switchers.

[james_s]

Keep in mind there are a lot of fake LM2576 parts coming from Chinese sellers. They work reasonably well but they are not LM2576, don't even run at the same frequency and cannot be expected to meet the same specs.

[David Hess]

The National datasheet shows load regulation which is so good that they did not bother to specify it.  Figure 19 shows a change of millivolts from 0.5 to 3 amps.

[Davy]

I know about all the fakes kncking around... especially RF power transistors, some have been re-blacked and reprinted, infact I have 2 pairs of 2SC19171 & 2SC1972's  the printing is easily smudged as if not dried, their die is only 1/3rd of the originals..... they wouldn't event fit in a Kenwood TS2000, and for RF drivers they had gains in their 90's.... their rf capability was about 2 Watts at 28mhz. The 144Mhz drive in the TS2000 comes from the HF drivers, clearly these are NOT gonna' work at VHF.

I now only buy silicon slabs from reputable companies.... Ebay is a big no, no unless I know who it is I'm buying from.

No problem about the inductor, I'll have a play when time permits, just that I was curious about 1.1V, I could understand if it was 200Mv or so. Not going to revert to an original linear reggy yet.

The darn LED  with some 315 uV LED's gets warmer while the 'switcher' is almost stone cold, got em in groups of three with two SMD resistors in parallel in each group for comfort.

Let you all know how I get on..... if ya' see the moon with two faces give us a wave.

Dave

[Davy]

Eureka...!
I wound another inductor on the same core with 36 turns of 20SWG, this giving about 84uH, the o/p voltage increased by about 350mV BUT on load only drops funnily enough by about 300mV, not 1.1V as before this is with 13.5V input. Maybe a little more reduction in inductance could be had.... this is good enough for me and doubt the LEDs would complain providing I don't drive them 'hell for leather'.

Thanks to MagicSmoker for putting me in the ball park. I appreciate all your inputs.

Dave

[hesam.moshiri]

if you want to build an adjustable converter using LM2576, you can build this