RMS or Average ?

[ferrari10]

Hello, i have a deceptively simple question, which i don't feel is that simple....

I am writing with regard to calculating the input power to an SMPS,
-that is calculating the input power from the DC input voltage and current.

The problem i have is

For a continuous mode boost converter PFC, they use V(rms) * I(rms)

...but for say a discontinuous mode flyback, they use V(average)* I(average)

I wonder when to use average and when to use rms ?

Here is a picture of the input power calculation dilemma.

I am very grateful for any replies.

 

[VEBill]

Averages can, in some circumstances, be very deceptive.

Imagine that the current waveform on the right side of your image is, for some reason, both positive and negative. An average current would have the positive and negative canceling out, leading to false results.

So you can only use averages if you're very, very careful.


RMS is more reliable and could be used on both sides of your image.

 

[ferrari10]

Thanks VE1BLL,

Before seeing the ML4821 datasheet, I thought RMS was just for AC. In fact, i still think it is only for AC.

I always thought averages were right for DC.

 

[zeusfaber]

RMS works fine for dc too - as long as an RMS figure is what you want.

If you don't want to caught out by oddities of phase and waveform, you might want to fall back on averaging the instantaneous product of I and V - much easier than it used to be in the days before you could export the data from a scope to a PC and do calculations with it there.

A.

 

[waross]

Before digital meters the most common meter movement was the D'arsonval meter movement. This is a DC movement. AC was measured with a rectifier circuit. BUT, the movement responded to the average of the DC current or voltage.
There is a 1.1 ratio between the Avg and RMS of a sine wave. This factor, called the form factor, was/is canceled by offsetting the scale 10% so that although Avg was measured, RMS was indicated. This worked so well that many persons were unaware of the subtle difference. I have several times seen distorted waveforms where the form factor was no longer 1.1 and seen the confusion that resulted.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[stevenal]

AC or DC, Pav=Vrms*Irms

For the waveform on the right, Vav=Vrms and Iav=Irms.

 

[stevenal]

Okay, where is the edit button? I sure messed that one up. Let me try again:

The general rule is Pav=Vrms*Irms*pf= average of V(t)*I(t). Rule works with sinusoidal and non-sinusoidal quantities.

The case on the left with sinusoids in phase, the power factor (pf) is 1, so the equation is correct.

The case on the right is not sinusoidal, so the pf (distortion pf in this case) does not =1. In this case, V=Vav=Vrms=V(t)for all t. However, Irms is not equal to Iav. The ratio of the two would be the pf. The given equation is correct. If the rms equation is used, use a pf of sqrt(2)/2.

Sorry for my incorrect answer.

 

[stevenal]

Further correction needed: I considered only the conduction period when I derived the pf given above. The off period duration will lower this number.

 

[VEBill]

Is that the average pf over each cycle?

;-)

 

[stevenal]

I guess you could call it average, since it is the ratio of Pav/VA. No such thing as an instantaneous pf. And if you don't have a periodic waveform, there is the problem of coming up with a suitable period. Simply called pf.

 

[stevenal]

Average pf is generally used to describe pf over many cycles, such as the average pf over a month's time.

 

[ferrari10]

hello,

i can appreciate that if the above right flyback waveforms (picture in 1st post) were of a flyback stage occuring after a simple full wave bridge then the power factor would indeed be low.

of course, forgetting about the AC mains for a minute, with DC waveforms from a DC source, the power factor is always unity.

zeusfaber has given an answer which i believe to be correct....

"averaging the instantaneous product of I and V" gives the actual power.

its just that this is not always a practically convenient way of doing it (as zeusfaber said)

the problem is that the product of V(rms)*I(rms) for the above waveforms (in each individual case) would not be the same as V(average) * I(average).

...and i was wondering which one is the correct one?

 

[stevenal]

DC waveforms? Pure DC has no waveform, just a flat line. If you have any kind of periodic waveform, the pf is not necessarily equal to unity. Low pf can be caused by displacement, distortion, or a combination.

The given equations are correct for the waveforms shown. If you want a more general rule, you will need to use instantaneous values as zeusfaber suggested.

 

[ferrari10]

thankyou stevenal,

"If you have any kind of periodic waveform, the pf is not necessarily equal to unity."

.....May i just pose a question on this?

If i run a flyback SMPS from a battery , then are you saying that somehow there is a power factor to be taken account of?

 

[stevenal]

If it creates a periodic waveform and you have rms quantities, yes. Nothing to account for if you use instantaneous values.

 

[ferrari10]

stevenal, thanks for getting back.

the thing is, none of the semiconductor companies sell PFC chips for DC battery applications

with DC your instantaneous V*I product is always positive, at any instant, and for that reason, the power factor is always unity.

 

[stevenal]

Never heard of such a rule. My analysis of your positive only sawtooth above shows a less than unity pf. If you didn't follow; where are you lost?

Power factor correction capacitors work only to increase a low lagging displacement pf for in an AC system. I know of no similar product for distortion pf, and don't know why it would be desired.

 

[ferrari10]

stevenal,

to be honest, i find this is a "trick" question at intefrviews that i've had.

-they speak of a battery application and then ask you what sort of power factor correction you would use....

the correct answer is that you don't need power factor correction for battery applications.