SMPS controller compensation

[marginal]

Hello, regrading SMPS, the Peakswitch family of SMPS controllers require no compensation components and use simple ON/OFF control.

They have excellent regulation and transient response.....

http://www.powerint.com/sites/default/files/product-docs/pks_family_datasheet.pdf

.....what i can't understand is.....given they are so good and simple, why are these type of controllers only used up to ~120 Watts ?

 

[cabraham]

At the level of integration in these compact devices, more than 120 watts is difficult. The internal FET would have to be larger to carry higher currents, and removing heat from the package becomes an issue. This family of parts is popular because it is integrated and compact. I have product designs in the field that use the Power Integrations TOPSwitch family.

When used with a TL431 opto-coupled feedback, they do indeed have tight regulation. The Zener feedback is not bad. But, the simple transformer winding feedback functions well, but the regulation is not too good.

I use them when I don't have a lot of space (which seems to always be the case) as they save room. Also, less effort is required since much of the engineering is already done.

At higher power levels, discrete parts are needed to handle the power and heat safely.

Claude

 

[GonzaloEE]

I've worked with a few parts from TinySwitch and TOPswitch series. The internal MOSFET RdsON is fairly high (over 1 ohm), this may help the internal drain current sensing, but limits power output.
Unfortunately, PowerInt parts doesn't include any external MOSFET interface.

 

[marginal]

Thankyou very much indeed for your responses.

Though I think i may have confused the issue by making reference to a particular controller family.

Rather than asking about integrated controllers and their power particular power handling capabilities, i am alluring to the fact that ON/OFF control could be done with an external MOSFET and MOSFET driver etc......

....and wondering why this type of control (ON/OFF control) is not used at higher powers than say 120 Watt?

 

[marginal]

...this is the sort of ON/OFF controller circuit that could be put on an IC, for example:-

...i am wondering why this type of ON/OFF control is not done at higher powers say 200W + ?

...it works fine in the Peakswitch and they do up to ~120W

 

[GonzaloEE]

I guess we find loop compensation more often in lower power ICs, because manufacturing tolerances are less significant at these levels. Any variations in loop performance could be handled by external filtering, heatsinking, etc.

I'm not sure if the embedded loop comp. stage of a PeakSwitch (or any PWM controller) would work equally well at 1A and 80A, if I only replaced power ratings of FETs, xformers, etc.

 

[treez]

ON/OFF control works ok for discontinuous mode flybacks and a lot of current mode SMPS's.

However, when you have say a continuous mode boost, or a voltage mode forward converter, you cannot get away with on/off control.

this is because of the double output filter pole in the forward converter, and the right half plane zero in the continuous mode boost.

..these effectively put delays in your loop, and you then (paradoxically) need to put further delays in the loop so as to correct the phase so that you dont get the condition where you end up with gain >1 at 180 degrees phase shift, -giving oscillation.

..basically with these two modes, you know all smps's have some oscillation on there output....you just have to make sure that you don't exaggerate this oscillation by poor feedback compensation. -ie don't unwittingly make the downward output voltage excursions go even further down, and likewise for upward output voltage excursions.

with on/off control in these cases, you can get such oscillation.

higher power converters don't tend to be discontinuous flyback, and also tend to have an output L,C filter.

 

[Fluorescence]

also, those ON/OFF controllers are OK for small power SMPS, but they are too noisy when you go to higher powers.

At higher powers you need a nice smooth PWM or otherwise noise can mess up everything.