DC on the mains???

[Ed_Dinning]

Hi Folks, a lot of the present range of SMPS that use a PFC front end only rectify (usually bridge) at this point with no smoothing. This feeds a chopper circuit operating at usually over 50KHz and pulse width modulated, such that the integral of the current pulse approximates a sine wave.
This meets waveform purity standards as well as allowable harmonic content.
This will give good power factor but at the expense of RFI that then needs more bits to filter out.
This PFC stage feeds into a DC link with a large storage cap(hold up, ride through etc) and then on to the convertor.
A simplistic overview of an SMPS that mostly meets present standards

Ed

[Mark Hennessy]

That's right. Basically, an existing design is cut in two, between the output of the bridge rectifier and the bulk smoothing cap. Between them is where you insert a boost converter - a separate switched-mode converter in addition to the main one which doesn't get altered as part of this.

At first glance, the boost converter aims to turn the varying (at 100Hz) output of the bridge rectifier into something like 380V DC at the input to the existing converter. The reality is a bit more complicated, as the error amplifier incorporates an analogue multiplier that samples both the incoming 100Hz waveform and the DC voltage at the bulk smoothing cap. It's this action that changes the boost converter from something with a negative input resistance into something that is approximately resistive.

Boost converters boil down to just a handful of parts: an inductor, a switching transistor and a diode - plus the controller IC. And there's normally a high quality film cap (typically 1uF) at the output of the bridge/input to the boost converter.

The fact they can be easily retro-fitted to an existing design - literally by cutting the PCB track joining the rectifier to the big cap if you were so inclined - means that they are easy to add to a proven design. They were often a small daughter-board in PC power supplies in the era when "active PFC" was a premium option. Today, PFC is required on anything bigger than 75 watts, and it is almost always active PFC (using a boost converter).

Passive PFC was sometimes seen, and is basically an inductor acting as a filter. This was a bit of a cheat, and didn't give brilliant results. Can't remember seeing one in the last 15 years.

Another nice feature of boost converters: no need to worry about input voltage switching. Some supplies could work over a wide range natively, others had the selector switch that turns the input bridge into a 2 diode voltage doubler. No need with PFC - the boost converter doesn't really care if the input only reaches 160V at the peaks rather than 320V - it can still put a nominal 380V at its output.

Another thing that simplifies the design of the boost converter: it doesn't provide DC isolation from input to output, so the inductor is cheap because it doesn't have to be a transformer with the requisite isolation and breakdown performance. All that is coming from the main converter, which has probably already been designed and tested.

For viewing current waveforms, current transformers are pretty good: https://www.markhennessy.co.uk/mdu/ - might not have the BW required for serious work EMC, but not bad for the cost and simplicity