Transient response of half bridge

[grizedale]

Hello

Any tips for maximizing the transient response of an opto-isolated 350W half bridge converter?

Vin = 210VDC to 375VDC.
Switching frequency = 111KHz
Vout = +/-50V

The high Vout, and high duty cycle (0.8) at Vin Min will conspire to give this converter a low crossover frequency, so any tips for improvement much appreciated.

 

[FvM]

Are you referring to the overall continuous-time equivalent circuit frequency response of the converter? The circuit topology (apparently flyback), output filter dimensioning and transformer inductances should be known.

 

[grizedale]

sorry

i put the secondary the wrong way round

-my question still holds, yes , i am referring to what you kindly say

 

[mtwieg]

Are you using current or voltage mode control?

In general it's feasible to get a crossover frequency of 1/3 of the switching frequency with either, but it's a lot easier with CMC.

Does a half bridge have a RHP zero? I've never designed one myself.

 

[Orson Cart]

if you are tryig to regulate the voltage on the output of your converter and there is an ouput inductor and capacitor the resonant frquency of these places a limit of the dynamic response of your feedback loop - usually your loop has to be slower with gain <1 at the res freq of the output LC.

 

[grizedale]

it is using voltage mode control

the dc bus spans 210vdc to 370vdc so the range of duty cycles is enormous.

,.....current mode would need so much slope compensation at duty of 0.8 that its not worth doing current mode control

 

[Orson Cart]

Slope compensation is related to inductor size & ouput voltage only - the larger the inductor the less compensation needed - we build converters up to 3kW with input range 250-400VDC using current mode with slope compensation (half the downslope of the inductor current + a little bit) and we have sold thousands, all with stable control and good current limiting. Regards, Orson Cart.

 

[grizedale]

though i report that the feedback loop bandwidth would be far too slow if the crossover frequency was less than the resonant frequency of the output LC filter.

-so it must be the case that crossover must be above the LC resonant frequency.

 

[Orson Cart]

this is why most people use current mode as it removes the effect of the output inductor from the transfer function of the power stage - i.e. you are now controlling a current into a capacitor (& load). Unless you are extremely adept in designing voltage control loops - you will find that any attempt to control the converter faster than the output LC res freq will give you an unstable result (except for discontinuous operation at very light loads) Regards, Orson Cart.

 

[grizedale]

if you see pg 250 of "switch mode power supplies" by Christophe Basso, Basso states it is necessary to get the crossover frequency at least 3 x the LC resonant frequency, so that it is fast enough not to react at the LC resonant frequency.

 

[mtwieg]

if you see pg 250 of "switch mode power supplies" by Christophe Basso, Basso states it is necessary to get the crossover frequency at least 3 x the LC resonant frequency, so that it is fast enough not to react at the LC resonant frequency.

I'm quite sure he doesn't mean what you think he means. In fact, having a load that is resonant within the bandwidth of a controller is probably a bad thing, since it will cause the transfer function to vary greatly as you pass through the resonant frequency (you would go from having a pole in your transfer function to having a zero at high frequencies). I can't imagine such a system being unconditionally stable.

---------- Post added at 08:59 ---------- Previous post was at 08:56 ----------

this is why most people use current mode as it removes the effect of the output inductor from the transfer function of the power stage - i.e. you are now controlling a current into a capacitor (& load). Unless you are extremely adept in designing voltage control loops - you will find that any attempt to control the converter faster than the output LC res freq will give you an unstable result (except for discontinuous operation at very light loads) Regards, Orson Cart.

Is this really true for for half bridge (and other topologies like forward)? Is using peak current control on the primary really equivalent to using it on the secondary side choke (which is what you would ideally like to control)? I'm just wondering if the primary magnetizing current would screw with the control.

 

[Orson Cart]

Is this really true for for half bridge (and other topologies like forward)? Is using peak current control on the primary really equivalent to using it on the secondary side choke (which is what you would ideally like to control)? I'm just wondering if the primary magnetizing current would screw with the control.

Yes. Magnetising current in the Tx is actually beneficial (artificially adds slope compensation).

Don't forget you can't use peak current mode with a half bridge (unless you address the resultant volt imbalance on the DC blocking cap(s)) - you can have average current mode though (with a half bridge).

 

[grizedale]

Hi

So are you saying that the transient response of a half bridge converter is slower (worse) when one has smaller Hennry value output inductors?

Surely that cant be right?

 

[Orson Cart]

So are you saying that the transient response of a half bridge converter is slower (worse) when one has smaller Hennry value output inductors

I don't think any one has suggested this - Regards, Orson Cart.

 

[grizedale]

OK, my humble apologies, but is there agreement that if a voltage mode, half-bridge smps is wanted to have a fast transient response, then it is likely that we will pick low Henry value inductors for its output inductors?

 

[Orson Cart]

This is true for any topology that uses an output choke... Regards, Orson Cart.