PFC in CCM with NPC1654 for Audio Amplifiers 300 W

[2rock]

Hello guys,

I built a CCM PFC circuit using a NCP1654 IC from On Semi to get 390 V that I feed to an unregulated DC/DC-Converter that powers a Class D amplifier. It is supposed to operate in wide mains from 85 V up to 265 V and have a maximal output power of 300 W.

The DC/DC-Converter works with 100 kHz +-20kHz so I choose the switching frequency of the PFC to be 200 kHz so I won't get differential tones in the audio spectrum, as the other frequencies of the chip are 65 kHz and 133 kHz.
Additionally my choke becomes smaller.

My circuit is able to operate and provides a stable output off 390 V when I power the IC with an external supply. As component count and size is a matter, I would like it to operate from a bias supply using an auxiliary winding from the choke. But then my output voltage is not stable enough and the PFC is not able to operate in wide mains.

I've come to the conclusion, that this is connected to my other problem:
In audio applictions the power needed to drive the speaker is very low. With all the other devices in my audio amplifier I will need about 10 W in idle mode.
As far as all datasheets and other literature, CCM ICs are not able to work properly below 20 % of their rated output power. That is also the power level at which the IC transits between CCM and CrM/BCM.
With 300 W rated output power, 20% means 60 W, which is way above what I need, so at 10 W the IC is working in CrM/BCM or in the case of the NCP1654 it goes into burst mode. In this burst mode the supply voltage through the auxiliary winding to the IC is not stable enough, so that it is average 9.5 V. That means the IC shuts itself down to protect itself and so the output voltage is not well regulated and therefore the auxiliary voltage is also not stable. That is the reason why TI suggest powering up the IC with an external supply instead of my bias supply method.

When I draw 64 W from the Class D amplifer at the end of my supply chain at 4 R (measured at the load) at high line (230 V), the PFC transits to CCM and I get a very stable PFC output voltage with almost no ripple voltage.
At 180 V I am able to get into CCM at just 32 W.

All the literature I read suggests, that I might have choosen the wrong topology for my project. But when I look at the datasheets and eval boards from TI, ST, Fairchild and On Semi and other topologies like CrM/BCM, FCCrM, interleaved and bridgeless designs, I come to the conclusion, that they will also fail when the power drawn is very low compared to the rated maximum output power.

Does anybody have some suggestions on how to proceed further? A new design? Another topology? A low power supply for idle mode and a high power PFC?

In addition, my PFC disturbs the mains heavily or radiates, as I can hear the pulses on the radio in the next room.

Thank you for your help and advices.

PS: The primary reason for the PFC is the ability to draw the same power from the line either in 100 V or 230 V mains. Only a voltage doubler for 115 V to 230 V would mean less output power, when the device is connected to 100 V mains.

 

[MicrosiM]

For that low power you don't need a PFC in reality.

Unless your application is critical, then you need one. but in audio applications you don't need

It will be difficult to obtain the same result without AUX supply feeding the PFC controller.

PFC is an advanced project and requires lot of testing, perfect parts selection, proper topology, perfect inductor selection + proper calculations.

Hope that helps

 

[2rock]

Thank you for your reply.

Unfortunately this project is not based completely on reality. I need to do it to get my degree. Neither me nor my professor and my supervisor comprehended the difficulties regarding power factor correction beforehand so now I am stuck with a half-working circuit board and I don't know how to proceed further.

I found this design from TI, that is also using a CCM PFC IC: http://www.ti.com/tool/PMP5327#0
It operates down to 10% Pout, so I should also be able to get my design working, as long as it does not depend on the downstream DC/DC converter.

I thought about something like the NCP1129 to provide a stable supply voltage to my IC.

I will also try some new magnetic designs with higher inductance to get the ripple current as well as the DCM / CCM transition border down to a lesser required output power.
For my initial design I followed the calculation sheets from On Semi referring to the NCP1654 and despite the unstable IC supply, the EMI issues and the light load problems it is stable when externally supplied.

Any suggestions regarding my EMI issues? Unless I get my pcb I will have to deal with them. Surely they are to a big part caused by my circuit board layout but I didn't expect them to be so massive. I tried to minimize loops as good as it is possible with through hole parts. Could there be any other causes for the EMI issues except the layout and the burst mode / DCM behaviour of the IC at light loads? Even in CCM my design is radiating EMI. When in burst mode, I only hear the burst every few seconds through the radio.

Best regards

 

[blasphemy000]

The NCP1654 is a fairly simple, easy to implement, and fairly flexible PFC controller. As you have found out though, at very low loads, when operating in burst mode, generating the VCC from an auxiliary inductor winding doesn't work. The current pulses through the inductor aren't enough too keep the storage capacitor for the VCC rail charged. One simple solution, since the PFC is non-isolated, is to employ a simple buck converter off of the rectified mains, or a simple flyback supply using something like a TOP/TNY IC to generate the VCC supply. Your auxiliary winding's return path should also be connected to the main return point(since this is where the IC gets it's GND). You have yours connected to the line side of the current sense resistor. This could effect the reading the IC is getting when sensing the negative current sense signal and could be causing some of the EMI issues, but I'm unsure about that. I've never had any EMI issues when using this IC.

 

[JGalt]

the NCP1654 datasheet shows an EMI filter in the block diagram..maybe that would help?

for the VCC supply, cant you generate something directly from the output of the full bridge? add a resistor + cap and make a simple rectified power supply? The Iq of the chip is only a few mA. Use high value resistors to limit the current, and OR them with what will take over once the supply is up, which according to the datasheet is I guess:

"It is not recommended to make a
low−voltage bias supply voltage by adding an auxiliary
winding on the PFC boost inductor. Alternatively, it is
recommended to get the Vcc biasing supply from the
2nd−stage power conversion stage."

 

[JGalt]

also, is it possible the inductor you chose for 200khz is saturating or somehow not performing as expected and passing very large current spikes which create the EMI?

 

[2rock]

As my schematic shows I also integrated an EMI filter but I might be able to improve it with changed parameters and eventually an added inductor for blocking differential mode interferences.

For the power supply I will use a Traco integrated AC/DC converter with 5 Vout and 3 W.

I don't think that my inductor is saturating as my inductor is quite big (ON Semi used the same core size for their 65 kHz reference Design) and I have added sufficient air gap. But otherwise I still have to look at the current to make sure.