Driving feedback pin with negative voltage. Does it work?

[VEGETA]

Hi there friends,

I know that linear regulators like LM317 can be made to output true 0v by feeding them with -Vref voltage in their set or fb pin. Now I want to do the same with switching buck regulators (or any topology) like the famous LM2576 or any similar one. I like MAX17503 part but it only goes down to 0.9v which is not suitable for a lab supply that I intend to make.

There is also controllers like LT8710 SEPIC controller which is capable of either positive and negative controlled outputs, but each internal gain amplifier has its own voltage reference.

My setup will be the traditional OP-AMP with positive rail and a negative rail to be able to output necessary voltages.

Can this be done with these traditional bucks that I mentioned? notice that LTSpice simulation of this method on certain buck converters didn't give me the required results for some reason.

thanks!

[David Hess]

I am not sure about the LM2576 but integrated regulators like the LT1074 which bring out the output of the transconductance amplifier for frequency compensation certainly can be.  You might be better off using an integrated pulse width modulator like the SG2524 or one of its current mode descendants to drive an external power switch.

[VEGETA]

I am not sure about the LM2576 but integrated regulators like the LT1074 which bring out the output of the transconductance amplifier for frequency compensation certainly can be.  You might be better off using an integrated pulse width modulator like the SG2524 or one of its current mode descendants to drive an external power switch.

Well, making a buck converter out of mosfet switching is not a bad idea but I am afraid it will be unstable and stuff like that. I tried simulating it and didn't get what I aimed for. LT1074 that you mentioned seems neat and easy to use, but it is a whooping 15$! while the full-featured LTC3649 (http://www.linear.com/product/LTC3649) is only 12$. I want something around 5$ max.

Do you think driving an n-channel mosfet as a switch will go down to true 0v? will it be stable?

I wanna search for such designs, but I need to make sure current limiting and current source capability is achievable.

thanks

[David Hess]

I did not know the LT1074 was so expensive these days.  Check the LT1076 also which is the same part with a lower current limit.

The reason I do not know if the LM2576 is suitable is because the datasheet is vague compared to the LT1074/LT1076 datasheet.  The LM2576 can probably be used that way but it would need to be tested.  Read the LT1074 datasheet for information about what happens at the very low duty cycles needed to support a 0 volt output.

These regulators can provide a 0 volt output without relying on a negative voltage.  They are not like the LM317 which always maintains 1.25 volts between its output and adjustment pin; instead, their feedback pin is referenced to ground and *raising* it above the regulators reference voltage will bring the output voltage down to 0 volts.  So you want to add a current into the feedback pin instead of subtracting a current to bring the output below the reference voltage.

[VEGETA]

Well I have tried LT1076 in LTSpice and I could not get any good results. I tried feeding the FB directly with a voltage source. I tried different values but non seemed to work.

Can you point out how to make output voltage from 0-30v out of this? with using voltage of course because I don't want to use POTs.


thanks!

[David Hess]

Do not drive the feedback pin with a voltage although you could drive the Vc pin with a voltage in a closed loop application which bypasses the built in error amplifier.  Instead, use the normal resistive divider to set the output voltage and then drive the feedback pin with a current.  Here is an excellent example of a 0 to 5 volt supply.