Add an x-tra output to an excisting fly-back off-line SMPS 1

[walker1]

Hi
I have an off-the-shelf fly-back SwitchMode power supply giving off 5V, 10 A.

I only need about 5A, but wouldn't mind having a few x-tra volts.
No output adj. is possible, as it is all done in some sealed off thin film circuit.

However could I simply add a negative forward converter circuit to the same transformer output,
or will I by doing so destroy the stored magnetic energy needed for the fly-back part ?

The forward output will be quite unregulated, but could non the less be helpfull.

 

[melone]

Why not just put a zener in the ground leg of the supply?

 

[Lewish]

Yes, if I understand what you say correctly, you can do that. It is commonly done on supplies for systems with backlit displays and LCD displays. Your total power out capability must not be exceded, however, so be careful how much current you draw from the additional source.

 

[walker1]

If I understand the thing with the Zener correctly, then I need to cut the transformer GND output
and insert a Zener of say 3.3 to 4.7 Volts, thus 'fooling' the regulator to think the output is only 5V ?
An interesting idea !!

This zener might need an additional capacitor in parallel, to cary the current spikes from the output filter.
And/or maybe a Zener boosted by a transistor ?

As the supply is able to work from 100-240 VAC input, I should have plenty of voltage headroom
when running at 230 VAC.
The 5V output is rated 10A, and the 12V, which is nearly unused is rated 3.5A,
so also powerwise (100W total) I should have enough headroom.

Except for mere mA I need to pull 4A from 5V to get anything out of 12V anyway.
The internal fan is running nicely, though

 

[melone]

One piece of advice, place a resistor from the output of your rergulator to the Zener. You want to make sure that your Zener is fully broken down otherwise your output might oscillate a little. The amount of current flowing through the ground leg of your regulator is usually pretty small, and you just want to help it out a little. If you can guarentee about 1-10 mA through the zener, then you can be confident that your zener voltage is stable.

Good luck and keep us posted on your progress!

 

[walker1]

OK. I cut the transformer GND wire (all 0.75 inch of it) and just for a start inserted 3 diodes
in series shunted by a good capacitor (3300 uF). (The capacitor was needed !)

When pulling 1 Amp from the old 5V, I had 6.5V. However I could not pull much power from the 6.5V !
System shut down until load removal and a power off/on cycle.

Instead I tried to raise the GND of the regulator. Same circuit as above. (Again the capacitor was needed)
Nearly same result, however. System 'died' at 2 Amps.
(I have previously had the 5 V loaded with 9 Amps with no problems at all. (Heat in the load excepted )

I am now turning towards a secondary switcher, boosting the 5V/10A to something like 8V/4.5A

 

[melone]

Your diode aren't forward biased enough. THERE IS VIRTUALLY NO CURRENT FLOWING THROUGH THE GROUND LEG OF A REGULATOR!!!!! That means you have to supplement it. Try putting a 4.7K resistor from the output to the top of your diode string and see if that fixes it.

 

[walker1]

Yep. Done that already. 1.5 k in fact.
I did notice that the voltage over the diode (and on the output) was a bit low.

The only part of the low-side regulator that is visible is a TL431, but it is controlled from the blacked out circuit.
On the high-side I have a Samsung KA3842.

Except for that "reverse engineers hands off" circuit, nothing seems to be out of the ordinary.

Could the KA3842 be connected in some way, so it does some 'forward voltage control' ?
That is: "This duty cycle is too high given this input voltage so I shut down"

I think not, but ... ?

 

[melone]

Does it have to be a switcher?

 

[walker1]

As a matter of fact some x-tra circuit around the KA3842 does shut the thing down,
if the duty-cycle becomes too high !
(I did some reverse engineering over the weekend

The supply voltage (not at upstart, but after that) comes from an x-tra output on the transformer
(standard fly-back output, thus sensitive to transformor duty-cycle),
and if this voltage becomes too high, an SCR shorts the ref. output of the chip to GND.
I might have used some less drastic methode involving the current sense input, but ...

With that knowledge, I may now be able to create something a little more 'cunning'