LM2576 -5V

[dan3460]

In the project that i'm working now I need to provide -5V, I used the LM2576 for the +5 rail and looking at the data sheet for the chip show the example of a LM2576-12 that provides with -12V. I thought about using the same diagram for my 5V chip, but after building the circuit I'm not getting any voltage on the output and the current limiter on the bench power supply is activated. I have set the current limiter at 300mA, with no load this thing should be only drawing a miniscule current. I have attached the diagram from the data sheet.
My first idea was ti build a regular 5V(lm2576) supply and put the + side to the project's ground and use the "ground" from the new supply as the -5V. Would this work?
The project is using a 12V 10A switching power supply, from there I have a 5V rail (LM2576) and a  3.3V rail(lm7803).

[bktemp]

This circuit needs a large current at startup. If you apply current limiting, it will not start up at all.
Try setting the current limit much higher, or keep the LM2576 disabled using its ON/OFF pin until the input caps are charged to the full voltage.

[mariush]

I'd suggest using a more modern regulator that doesn't force you to use big inductors and capacitors which would cause high current spikes at load

See for example TPS62175 (page 24 in datasheet) if 0.5A is enough, or LM2611 which can also do up to 0.6-0.7a if you feed it directly from 12v

[dan3460]

Thanks for the answers, I set the current limiter because at my first attempt I saw the current rising at over 1A for a lot longer that just a few mS, I was afraid of burning the chip so I switched the power off. To charge the caps should be just a temporary spike on the current, but I think in a few mS should be almost no draw if circuit in not under load. The setup that I have for the 5V does start under the same setup (using the bench power) and with the current limiter in the same position. Even do, I have checked the connections several times I have stared at it for quite a while, I will check them again. I thought maybe this doesn't work with -5V only with -12 

I was using this chip because was in my bin. For the prototype I have a power supply that have + and - 12V, I have some LM78XX that can regulate to 5V, 3.3V and -5V. I can use this for the prototype and explore Mariush suggestion for the final product. The chip suggested looks that will be sufficient for the application, I have to run the numbers again but I think my peak current consumption on the -5V rail is about 100mA

[bktemp]

Using a buck converter for generating a negative voltage turns it into a completely different topology. That's why it behave totally different at startup.
Even the datasheet mentions the high inrush current:

This type of buck?boost inverting regulator can also
require a larger amount of start?up input current, even for
light loads. This may overload an input power source with
a current limit less than 5.0 A.
Such an amount of input start?up current is needed for at
least 2.0 ms or more. The actual time depends on the output
voltage and size of the output capacitor.

It needs a couple of amps at startup for a short time. If you power on the circuit with all caps discharged, it will stuck at the current limit threshold of your power supply and will not start up at all.

A buck converter will charge the output cap during startup because all the current flows over the inductor into the load. Therefore it will start up even with a current limited supply. But the inverting regulator will not charge the output capacitors, because the inductor is connected to ground, so all the current will flow directly to GND until the regulator starts switching. To make it operate correctly, it needs enough input current to trip its overcurrent protection. That's why it needs upto 5A for a very short time.

[dan3460]

Thanks for the answers, for the prototype because I have a computer power supply that provides -12v i'm using a linear regulator to get the -5V. I'm going to have to design a new power supply if I get this project working.