Improving stability and transient response?

[eem2am]

Hello,

We are doing an offline, 260W , isolated, voltage mode , full-bridge smps. (Vout = +/-50V/2.6A)

We need a fast transient response, but we know that this will be difficult because of the output LC filter. (68uH and 940uF).

This has a severe 180 degree phase drop at the LC resonant frequency (630Hz).

...this in turn makes feedback compensation very difficult.


However,

If we add a series resistor of 0R27 in series with the output capacitance, then the ESR zero frequency becomes 627 Hz.

........Thus we avoid the severe 180 degree phase shift of the output, and turn the power stage into a flyback type stage, which means a faster transient response is much easier to get. (Because there is less delay going through the power stage)


Do you think this is correct?

(The resistor will dissipate 0.6W at max load but this power supply is for audio guitar amplifier and not often at max load.)

[VivaVee]

Originally Posted by eem2am

Hello,

We are doing an offline, 260W , isolated, voltage mode , full-bridge smps. (Vout = +/-50V/2.6A)

We need a fast transient response, but we know that this will be difficult because of the output LC filter. (68uH and 940uF).

This has a severe 180 degree phase drop at the LC resonant frequency (630Hz).

...this in turn makes feedback compensation very difficult.


However,

If we add a series resistor of 0R27 in series with the output capacitance, then the ESR zero frequency becomes 627 Hz.

........Thus we avoid the severe 180 degree phase shift of the output, and turn the power stage into a flyback type stage, which means a faster transient response is much easier to get. (Because there is less delay going through the power stage)


Do you think this is correct?

(The resistor will dissipate 0.6W at max load but this power supply is for audio guitar amplifier and not often at max load.)

No, because your 940uF capacitor did not start with zero ESR...

[eem2am]

i agree, but surely the main thing is that the 180 degree phase shift has been reduced to a maximum 90 degree phase shift....meaning less delay going through the power stage, and ease in getting a fast transient response.

i will just add enough resitance so the esr is always above a certain value

[sawreyrw]

Use a current mode topology. This makes it much easier to stablilze and get a wider bandwidth.

[eem2am]

..i cannot use current mode because:

The Input voltage is from around 180V to 375V due to universal mains input...(there is a voltage doubler link)

...this Vin range means current mode is impractical.....so much slope compensation is needed that your virtually in voltage mode anyway, and the primary overcurrent limit is not as tight when you have slope compensation.

so i need voltage mode

Also, voltage mode is less noisy in lighter load conditions

[sawreyrw]

A full bridge is just a form of a buck converter with transfomer coupling. Yes, you need slope compensation (assuming the duty cycle is over 50%), but so what if the input varies 2 to 1? It's still current mode.

[eem2am]

its not pure current mode though.

When you have slope compensation, you are in a combination of voltage mode and current mode.

When you have slope compensation, you have to make the current sense resistor smaller because you have to make room for the added slope ramp.

...so your short circuit protection is not as good as it would be with pure current mode control.

Also, with Bridge converters with dual outputs, you get better cross regulation with voltage mode than you do with current mode.

If you use current mode, you usually have to couple the output inductors.....you dont have to with voltage mode.

[sawreyrw]

Duh, I didn't notice that you had dual outputs. This does change thing a bit, however, I would still go with coupled inductors and current mode control.

I agree with most of what you said, but I'm not aware of what you said about voltage mode cross regulation being better, or why coupled inductors are only needed for current mode. Could you explain that, just for my information.

Finally, if you use a type 3 regulator with voltage mode control, you should be able to improve the performance without adding the resistor to get the zero. The resistor will also increase the output ripple voltage.

[eem2am]

http://www.ti.com/lit/an/slua119/slua119.pdf

...current mode vs voltage mode.

As Basso states in his book, when you are using TL431/Opto feedback, the type 3 compensation can often not be achieved.

I would stick to voltage mode, and put the series resistor in to artificially increase esr.

I am sure you would agree that getting rid of the 180 degree sudden phase change is a welcome situation to be in.

And im sure all would agree that slope compensated current mode designs have degraded short circuit protection since the sense resistor has to be downsized to make room for the slope ramp.

[megajocke]

Originally Posted by eem2am

This has a severe 180 degree phase drop at the LC resonant frequency (630Hz).

...this in turn makes feedback compensation very difficult.

Not if you can set your crossover frequency sufficiently above the ESR zero. And the prerequisite for that is that the switching frequency is sufficiently above the ESR zero.

Then your compensator only needs to be an integrator + pole/zero pair (or integrator + zero) as shown in the attached diagram.