step-down question...

i've managed to get a working step-down going based on the maxim 1627 (MAX1626-7). for the coil and the output cap i've used the same zetex components (150uF tant cap and the bigger core with 22 guage wire and 2.5 turns). my question is, whenever i have more than minimum output (under 2.8V) it jitters pretty bad. i've skipped a couple of components like the R3 cap and the input cap, perhaps this is causing it? or perhaps i need a coil with more turns or a larger output cap? any ideas? thanks.

wow. according to my calculations i'm getting a whopping 30% efficiencey as well. arg...

here's a layout and list of components.



the IC in the middle is the MAX1627, the OUT pin has been cut off.
the transistor, to the right of the IC, is the Fairchild NDS9430A.
the cap bottom left is a .1uF tant.
the resistor bottom right is simply a jumper.
the diode top right is the Zetex ZHCS2000.
top center is the 150uF low esr tant cap from the 300 circuit, right now it's removed with no discernable change.
not shown are a 158K 1/4W resistor jumping FB (left side middle) to ground (top right, bottom left) and a 100K pot with 158K resistor in series jumping FB to OUT (top left). there is also a coil jumping the transistor output to OUT, the coil is now the large JW Miller ferrite bead with 10 wraps of 26 guage wire.

i've gone through all the soldering to make sure there are no cold joints, i've played with the coil a bit (different number wraps of different gauge wire) and tried a cap in parallel with the pot/158K R FB circuit to no avail. perhaps i made a bad choice on the transistor?

i suppose, if i can't get this working, i'll go back to the 65% eff linear regulator, but that's not an ideal solution.

any help greatly appreciated.

Hi there papasan,

1. First off, what are you using for a load?
You do need to apply a load to test most
switching regulators.

2. When you say "OUT pin is cut off"
what do you mean by that?

3. What are you using for a sense resistor,
and what value is it?

These questions should help answer your
original questions.
Im sure you will get it working one
way or another with a minimum number
of tests.

Good luck with your LED circuits,
Al

thanks for looking at this MrAl, your input is always welcome.

load is 17 nichias in parallel. with the resistor/pot set-up it should be an variable output of 2.6V to 3.4V (running the array at 3.6V draws almost an amp, i gues due to thermal issues with so many LEDs in parallel, overdring the array too much).

OUT pin has been cut from the IC. i didn't realize that the OUT pin was supposed to be not connected on the MAX1627 until after i had etched the board, so i just cut it off with a pair of mini-dikes.

i was using a short piece of 30 gauge wire for the resistor between the Vin of the IC and the transistor (similar in length to what i use in the ZCS circuit), but as of right now it's a piece of lead cut from a thu-hole resistor jumped straight accross. i.e., it's probably very close to 0 ohms. i can't tell or measure any difference with the resistor and without.

** edit **

i have managed to get rid of the jitter by lengthening the resistor wire between V+ and the transistor. even with no output cap it seems smooth now.

readings as of now:

the 6mA reading can't be right because you can still read to 2 feet or more. that's 1/3 of a miliamp per LED if it's right. perhaps mu fluke doesn't measure very small voltage very well.

Hello again papasan,


I see you've got FB bypassed, which is very good.

I also noticed that on the data sheet they devote enough
attention to the input capacitor(s) to warrent using one
or two. The reason for the input cap(s) is to bypass the
high frequency spikes that can appear at the V+ pin of the
ic chip due to inductance in the leads from the battery.
If the V+ pin goes low for even a short time period, it
could screw up the internal one shot timing.

I also noticed that they recommend some kind of what might be
called 'derivative' feedback, which comes in the form of adding
a small value cap across R2, maybe 100pf or so. Im not sure
if you tried this or not, but that would be a good idea too.
The range of values might go from as low as 20pf to as high as
1000pf. I would try stepping in values of:
20pf, 50pf, 100pf, 200pf, 500pf, and 1000pf.

The MOSFET is spec'd at 4.5 volts so that doesnt seem to be a problem.

If you continue to have the instability problems we might have to take
a very close look at the actual layout. If there are any exceptionally
long leads going to/from the circuit board that could be part of the
problem too.

This does look like a pretty simple step down circuit. I might be
tempted to try this one out myself

Good luck with your LED circuits,
Al

i tried a cap accross the FB resistor, just some very small tant i had. it had the effect of putting a delay on the FB adjustment and smoothing it out, but it also seemed to put a limit on the output and at full it was only at 3V or so. suppose i could get by this by adding more resistance. the circuit didn't get above the eff. in the last reading (which have been the best so far).

i will definatly try an input cap. need to tweak the layout a little to get one in, don't have any thru-hole caps.

i'll let you know how it works out.