Re: Feed Forward / Slop compensation

Thanks Jim, most of what you say makes sense to me, however I have a few questions.

I assume this is current mode control.

By modulation ramp do you mean RT/CT the oscillator input (pin 4 UC3842)?

If I understand what you mean by modulation ramp and am not mistaken, does this signal not vary with line voltage? Thereby not being feed forward but slope compensation?

If you check out the datasheet for the UC3842 page 7 (fig 9) has a diagram for slope compensation.

Is this done by gapping the core to increase the magnetising current and there by putting an extra thin slice on top of the primary current waveform?

From reading your post, what I think is that feedforward has to be governed by a change in a parameter (eg the line voltage) that the circuit corrects before the output changes. This definition could be applied to the inner fast current control loop and something that detects a change in input voltage and compensates the current sense pin.

Slope compensation applies achieves the same thing as feedforward but is not dependant on a moving signal, it's a fixed compensation technique what I called a graphical technique earlier.

Does this make sense?

Reggie.

In this example, the trailing edge of a PWM output is being modulated within a fixed frequency and limited maximum pulse width. Other forms of modulation are possible - leading edge, frequency, period etc, each could have a different method of introducing a feedforward influence.

That's a good example of a ramp circuit, but it is only used by the

3842 to determine frequency and minimum deadtime. Circuits like the 3524 and 78s40 compare an error amp output to asimilar ramp timing signal.

Normally a pullse width modulation ramp signal is constant in a basic modulator. No current signal is used. No slope compensation is used. No feedback or feedforward path is present.

This affects the magnetizing current contribution. If the secondary is coupled (ie forward converter), then output choke contribution is also present and usually dominates.

Slope compensation, if of fixed amplitude, has no feedforward influence what-so-ever.

If made variable, it could produce a feedforward influence.

RL

Sorry more questions,

When you say if the secondary is coupled do you mean that when the fet is on the current flows' in the primary winding and the output choke; thereby the slope of the primary current will be affected by the magnetising current of the transformer and the inductance of the output choke?

Generally output chokes have a larger ish inductance to make sure their currents don't go discontinuous (improves multiple output cross regulation)

Why does the inductance of the output choke dominate?

So feed forward is not the same as slope compensation, because slope compensation is of "fixed amplitude" as you put it. Is this in essence the difference?

I think/ hope thats it!!!

Reggie.

In a flyback or Cuk converter there is no coupling to the output during primary conduction, so there will be no contribution to primary current slope by the output filter inductor. In other

As a percentage of load current, 10% ppk ripple is not abnormally high, nor is a 5% minimum load on one output of a mmultiple output converter. It would have to be a special topology to allow similar levels of primary magetizing current in a forward converter.

Magnetizing energy is lost energy in a simple single-ended forward converter. Two transistor, constant voltage recovery, switched snubber or other circuit elaborations can recover some of it, but more commonly it just gets burnt up in clamps or the switch.

RL

Wow, you know your stuff.

I take it that magnetising current is much higher than 5%. Typically what would it be without a gap?

Is this correct?

Reggie.