UCC28950: Lmag

Hello KK

The reason for the 1-D term is explained in the attached word document. /cfs-file/__key/communityserver-discussions-components-files/196/2845.1_2D00_D.docx

If it's not clear then please let me know.

If Lmag is less than the value calculated by the Excel calculator then what will happen is that the system will behave more like a it is operating in Voltage Mode rather than in Peak Current mode. The reason is that the current sense signal contains contributions from both the transformer magnetizing current and the output inductor current. As the proportion of magnetizing current increases the output inductor current becomes swamped and becomes less important. In the limit, if the ramp were completely dominated by magnetizing current the proportion of the ramp contributed by the output inductor could be ignored - and this is effectively Voltage Mode Control.

So - either you increase the magnetizing inductance - larger core cross section or more turns - or select Voltage Mode control. VMC will require a different control compensation network but that is a once off design task. It will also require that you add a DC blocking capacitor to the transformer primary.

Please let me know if you have any further questions.

Regards

Colin

Hello KK

There is a sub-harmonic instability in Continuous Conduction Mode (CCM) converters using Peak Current Mode control(PCM) operating at greater than 50% duty cycle. The normal way to prevent this instability is to add a compensating ramp to the current sense signal. To achieve stability at all duty cycles up to 100% requires that the compensating ramp be 50% that of the output inductor down slope. This is really the minimum ramp that is needed and in some regards a 100% ramp slope is better. I'd recommend you look at the section on slope compensation in slup113  pp16 https://www.ti.com/seclit/ml/slup113/slup113.pdf 

Ramps greater than 100% will cause the system to behave more and more as if it were operating in voltage mode control but the transition is gradual rather than sudden.

I haven't spoken to the designer of the PMP8740 but it looks very much like he chose the magnetizing current slope to be close to 100% of the output inductor downslope. Note too that transformer magnetizing inductance will have a fairly wide tolerance band, ±20% would be typical so that would have entered into the calculation too. Finally, there may have been practical limits on the amount of magnetizing inductance that could be achieved on that core.

Here are some summary notes.

1/    Without slope compensation the loop is unstable for D > 50% but it starts to show an under damped behaviour even for lower duty cycles.
2/    Second, if the slope compensation ramp is 50% of the inductor down slope, the system is stable for duty cycles out to 100%. This is the usual recommendation because it keeps the peak – to average ratio constant and so the current limit point does not depend on duty cycle.
3/    A slope compensation ramp somewhere between zero and 50% of the inductor ramp will stabilise the system up to some duty cycle limit – maybe up to 70% for example. This can be done if the designer can guarantee the loop will never operate beyond the duty cycle limit
4/    A slope compensation ramp which is equal to the output inductor downslope gives the fastest recovery to a perturbation in the CS signal, see P17 in slup113 but the peak to average ratio changes with duty cycle. If the slope compensation is 50% of the output inductor current downslope then the average current does not change with the duty cycle and the current loop dynamics are better.
5/    In practice, the designer usually ends up with a compensating ramp around 70% of the inductor down-slope
6/    The current sense signal will be made up of a component due to the output inductor current and a component due to the magnetizing current. The magnetizing current ramp forms part of a slope compensation ramp.
7/    If the magnetizing current ramp becomes too much (Lmag too small) the CS signal (output inductor current + I_mag) becomes dominated by the magnitizing current and the system starts to behave more like a voltage mode controlled system than a Peak Current Mode controlled system. This is the main reason why we set a minimum value for Lmag. We also don’t want to have too much magnetizing current simply to avoid the associated I^2R losses.
8/    Non-linear slope compensation methods are available – but are rarely used.

Regards

Colin

Hello KK

It's been a while since your last response so I assume that we have resolved your problem and I'm going to close this thread. Please note that you can open a new linked post if you wish.

Regards

Colin