Questions about LM2596

I need to create different power supplies at 3.3V, 5V and adjustable dual 12V. By using some linear components I would need some big heatsinks on top of the voltage regulators, so I'm going to use swithing voltage regulators. I found the LM2596 (**datasheet link**) to be very simple because it has just what I need: fixed 3.3V and 5V versions and adjustable voltage versions. The required current for each power supply is 1A max. Now questions:
1) These power supplies will not always run with a load attached, meaning that the output leads would be left floating. Is that OK or will this break the IC somehow?
2) In the datasheet the schematic of the fixed voltage versions is only 1. Does that mean that the same schematic can be used with both 3.3V and 5V versions ? Shouldn't the schematic change due to the different voltage? I would have expected at least a component value change, but seems there isn't any (or I probably missed some important notes).
3) I would like an adjustable dual 12V power supply, but the LM2596 can only output positive voltage. I need a complementary IC that would allow me to get a -12V power supply, do you know any? Or is there any IC that would output both -12V and +12V supplies?

EDIT
For the -12V I found this website that converts a LM2575 from a step-down positive 12V regulator into a step-up -12V negative regulator:
**broken link removed**
To you, is this safe or would you recommend otherwise?

1) These power supplies will not always run with a load attached, meaning that the output leads would be left floating. Is that OK or will this break the IC somehow?

Click to expand...

There is absolutely no harm in this (circuit being on with no load). You just have to avoid overloading i.e more then 3A load (rated value).

2) In the datasheet the schematic of the fixed voltage versions is only 1. Does that mean that the same schematic can be used with both 3.3V and 5V versions ? Shouldn't the schematic change due to the different voltage? I would have expected at least a component value change, but seems there isn't any (or I probably missed some important notes).

Click to expand...

In fixed voltage versions, you will have to use the schematic and components values of fixed voltage version schematic, same for both 3.3V and 5V.

For the -12V I found this website that converts a LM2575 from a step-down positive 12V regulator into a step-up -12V negative regulator:
**broken link removed**
To you, is this safe or would you recommend otherwise?

Click to expand...

Looks like a fine way. But you have to be care full about the input voltage. Input voltage to the regulator is also negative.Which means you will interchange the ground and Vcc.

What does it mean "interchange the ground and Vcc" ? I know that I'll get a -Vcc voltage between GND (+) and Output (-), is there anything else I should be aware of?
Also, is it safe to connect the -12V circuit GND with the +12V one making a common GND? I mean, would they interfere with each other? Would I still have 24V between +12 and -12V if I use a common GND?

p.s. The input voltage will not exceed ±17V (17V per IC), and I will use a 9V transformer for the 3.3V and 5V. I think I'm keeipng the whole safe

Also, is it safe to connect the -12V circuit GND with the +12V one making a common GND?

Click to expand...

As i said interchange the Vcc and Gnd........See....+12V net is your Gnd net for -12V. +12V and Gnd (for-12V) is the same wire in this case.
So that means you have different Gnd reference for -12V signal and +12V signal.

That's a shame... I actually need a common GND... is there anything you can suggest me?

Well, a simple solution involves a dual transformer that can step down to near 12V and near 24V. Put the grounds together and it will be your -12V wire. the +12V transformer output will be your reference Ground and +24V will be your desired +12V Supply Output with respect to your reference Ground(+12V).

Yes, that's correct, but that's a linear power supply. I would like to reduce both space and heat by using switching power supplies.

This may provide some ideas: **broken link removed**

Using buck converters as voltage inverters is addressed in an application note. https://www.ti.com/lit/an/snva022e/snva022e.pdf

There are stability caveats to be considered. Generally the method is working fine. I've implemented it in a number of commercial products with National/TI converters.

LM2596 is a first generation simple switcher with bipolar output transistors, resulting in not so mind blowing performance and low switching frequency. You might consider newer memebers of the "Simple Switcher" product family (LM26xx and other more recentchips) as well.

Thanks FvM, I'm actually considering many more solutions. For positive voltage there is plenty of switching IC solutions, but I'm having a hard time right now in finding a dual supply solution; it seems there's no 'easy' way of doing it... I'm now reading the link provided by bking, I hope it'll enlight me a little.

EDIT
I read bking's link and yours too FvM. bking's link gave me some englightment, but your link FvM gave me the final solution. As I watched the schematic, it looks like I can use two 2673 in positive and negative configurations with a common GND.

Thank you all guys!

- - - Updated - - -

Sorry for double-posting, I have another question.
The LM2673 adjustable schematic reports two resistors (R1 and R2) on the output for the feedback pin. If I replace the two resistors with a potentiometer, is it necessary to use two more protection resistors in series, so the feedback never gets to Vcc or GND ? (see attachment)

The resistors on either side of the pot will limit the max-min range of your voltage adjustment, which may be a good thing for a particular design, but on the adjustable version grounding the FB pin is how you get the min of 1.2V out. I've never tried directly connecting the output to the FB pin on the adjustable version, but based on the Vout equation I guessing you'll get whatever the input is (no regulation). Or the part burns up. ; )

See page 7 of the data sheet ( https://www.ti.com/lit/ds/symlink/lm2673.pdf ) for the FB pin configuration for each version. Note that you can adjust the 'fixed' versions as well with a series FB resistor, but only higher than the 'fixed' value. This may be useful to compensate for mixing diodes in a simple multiple regulator set-up.

I got it now

Thank you again for your help!