Power factors when dealing with non-sinusoidal voltage waveforms are a bit more complex to understand. A rectified AC signal when fed to a reservoir cap will have a current waveform which doesn't look anywhere near a sinewave so its hard to estimate the power factor which is normally just calculated from the phase difference between the current and voltage waveforms.
The general principles are the same though - the larger the cap is, the lower the ripple voltage and therefore the shorter the time the rectifiers will be conducting. During this short time, the charging current into the caps must be high enough to recharge the caps for the next mains cycle (10mS with full-wave).
Take a simple example - if the rectifiers are conducting for 20% of the time (2mS), the current pulses through the rectifiers will need to be 5X bigger than the DC output load current. Fitting a bigger cap will decrease the 2mS time and therefore increase the size of the current pulses.
With me so far?
The general principles are the same though - the larger the cap is, the lower the ripple voltage and therefore the shorter the time the rectifiers will be conducting. During this short time, the charging current into the caps must be high enough to recharge the caps for the next mains cycle (10mS with full-wave).
Take a simple example - if the rectifiers are conducting for 20% of the time (2mS), the current pulses through the rectifiers will need to be 5X bigger than the DC output load current. Fitting a bigger cap will decrease the 2mS time and therefore increase the size of the current pulses.
With me so far?
A rectified AC signal when fed to a reservoir cap will have a current waveform which doesn't look anywhere near a sinewave
OK
The general principles are the same though - the larger the cap is, the lower the ripple voltage and therefore the shorter the time the rectifiers will be conducting. During this short time, the charging current into the caps must be high enough to recharge the caps for the next mains cycle (10mS with full-wave).
OK
Take a simple example - if the rectifiers are conducting for 20% of the time (2mS), the current pulses through the rectifiers will need to be 5X bigger than the DC output load current. Fitting a bigger cap will decrease the 2mS time and therefore increase the size of the current pulses.
OK
Now go on about the power factor part.
And why should we even care about power factor.
OK
The general principles are the same though - the larger the cap is, the lower the ripple voltage and therefore the shorter the time the rectifiers will be conducting. During this short time, the charging current into the caps must be high enough to recharge the caps for the next mains cycle (10mS with full-wave).
OK
Take a simple example - if the rectifiers are conducting for 20% of the time (2mS), the current pulses through the rectifiers will need to be 5X bigger than the DC output load current. Fitting a bigger cap will decrease the 2mS time and therefore increase the size of the current pulses.
OK
Now go on about the power factor part.
And why should we even care about power factor.
The power factor part is just another way of expressing that you've got much higher currents in your secondary circuit than you would have if there was only a resistive load, not a rectifier-capacitor load.
Why you should care about power factor is for the reasons already explained - your transformer's secondary current is specified into a resistive load, not a capacitive one. Hence the derating Andrew has already mentioned.
<edit> Having a poor power factor in this case means your mains waveform is going to get distorted (3rd/5th/etc harmonic distortion) - flattened tops. This might have implications for other appliances you have connected.
Why you should care about power factor is for the reasons already explained - your transformer's secondary current is specified into a resistive load, not a capacitive one. Hence the derating Andrew has already mentioned.
<edit> Having a poor power factor in this case means your mains waveform is going to get distorted (3rd/5th/etc harmonic distortion) - flattened tops. This might have implications for other appliances you have connected.
Last edited:
no. I cannot accept that.
no. That DF is I believe due to the internal heating effects of I^2R, not power factor.
no. Power Factor is the phase difference effect that power used and power charged for and inductive/capacitive loads have on the distribution system.
Abrax,
Schem has gone very quiet. Maybe he's still asleep.
Let him justify what he alleged.
Last edited:
Yes, but that's only a useful way to measure PF when both are sinewaves.
Right, they don't but they sure cause PF differences.
Hi,
Optimising power factor is part of optimising conversion efficiency.
But optimising conversion efficiency is not optimising power factor,
unless that is the only issue being considered, e.g. for a motor.
Here they are saying a "too big cap" reduces conversion efficiency.
(AFAICT) rgds, /sreten.
Last edited:
There are two kinds of power factor, PF and DPF (displacement power factor). Displacement refers to the fundamental only. DPF can be corrected with PFC inductors/capacitors; in general, PF cannot.
For linear loads, PF == DPF and you get simple sinusoidal power factor, phase angle, etc., all familiar from Power Systems. You can use L and C for power factor correction.
A rectifier is nonlinear. It produces strong harmonics in the current draw. The fundamental is nearly in phase, so DPF is very close to 1.00, even though PF can be less than 0.5. Distorted current waveforms cannot, in general, be corrected to any useful amount by adding capacitors or inductors to the line.
The problem arises from harmonics, which must be filtered using a low-pass filter. Such a filter is obviously going to be hard to construct at line frequency and impedance.
Better is to simply avoid harmonics altogether. Using a minimum of filter capacitance helps.
PFC circuits are designed for ideal behavior, while allowing any arbitrary amount of filter capacitance (as long as stability is not impaired). Notice that a PFC circuit always has sinusoidal, double-line-frequency ripple at the output. The DC voltage can be stabilized (generally something like 400V with good regulation), but the ripple is a necessary part of the circuit's nature.
The general definition of power factor is this:
S = Vrms * Irms
S is Apparent Power, measured in VA. In general, S is NOT watts (work, heat, etc.).
Real power can be measured with respect to time, in which case, it varies up and down with demand. Reactive components cause negative power, as they deliver stored energy back to the supply every cycle. The average is real power, in watts.
Instantaneous power: p(t) = v(t) * i(t)
Real power: P = p(t) averaged over any number of whole cycles
PF = P / S.
If you have 3A RMS current draw at 120V, with a real power consumption of 200W, the apparent power is 120*3 = 360VA and the power factor is 200/360 = 0.56.
Example:
If current is drawn in spikes, p(t) is also in spikes. A spike of 1A at a peak of ~340V (i.e., 240VAC RMS) is 340W peak (instantaneous) power. If that spike is 2ms long, triangular shape (a typical approximation for a cap-input filter), and the half cycle is 10ms (50Hz line frequency, full wave rectified), the average is 34W. The RMS current is 0.258A (Irms = Ipk * sqrt(t_on / (3*T)) for a triangular waveform of period T with dead time T - t_on), so S = 62VA. Then, PF = 34/62 = 0.549, a very reasonable value.
Tim
For linear loads, PF == DPF and you get simple sinusoidal power factor, phase angle, etc., all familiar from Power Systems. You can use L and C for power factor correction.
A rectifier is nonlinear. It produces strong harmonics in the current draw. The fundamental is nearly in phase, so DPF is very close to 1.00, even though PF can be less than 0.5. Distorted current waveforms cannot, in general, be corrected to any useful amount by adding capacitors or inductors to the line.
The problem arises from harmonics, which must be filtered using a low-pass filter. Such a filter is obviously going to be hard to construct at line frequency and impedance.
Better is to simply avoid harmonics altogether. Using a minimum of filter capacitance helps.
PFC circuits are designed for ideal behavior, while allowing any arbitrary amount of filter capacitance (as long as stability is not impaired). Notice that a PFC circuit always has sinusoidal, double-line-frequency ripple at the output. The DC voltage can be stabilized (generally something like 400V with good regulation), but the ripple is a necessary part of the circuit's nature.
The general definition of power factor is this:
S = Vrms * Irms
S is Apparent Power, measured in VA. In general, S is NOT watts (work, heat, etc.).
Real power can be measured with respect to time, in which case, it varies up and down with demand. Reactive components cause negative power, as they deliver stored energy back to the supply every cycle. The average is real power, in watts.
Instantaneous power: p(t) = v(t) * i(t)
Real power: P = p(t) averaged over any number of whole cycles
PF = P / S.
If you have 3A RMS current draw at 120V, with a real power consumption of 200W, the apparent power is 120*3 = 360VA and the power factor is 200/360 = 0.56.
Example:
If current is drawn in spikes, p(t) is also in spikes. A spike of 1A at a peak of ~340V (i.e., 240VAC RMS) is 340W peak (instantaneous) power. If that spike is 2ms long, triangular shape (a typical approximation for a cap-input filter), and the half cycle is 10ms (50Hz line frequency, full wave rectified), the average is 34W. The RMS current is 0.258A (Irms = Ipk * sqrt(t_on / (3*T)) for a triangular waveform of period T with dead time T - t_on), so S = 62VA. Then, PF = 34/62 = 0.549, a very reasonable value.
Tim
In those cases the power consumed from the line is low enough that the problems discussed in this thread (rectifier peak current, small conduction angle) are insignificant.
Use enough capacitance to be sure the ripple voltage does not fall below the regulator's dropout voltage. I presume all these low-power applications are using regulators - they're inexpensive enough and the benefits are great enough that it makes no sense not to use them.
Actually, they are similar enough - they are different changes from the ideal.
Unity power factor is when a sine wave voltage goes into a resistive load. The current into such a load is a sine wave AND it is in phase with the voltage. If either (or both) of these two things are not true, then it is NOT unity power factor. Consequences are that the current (not just peak but also RMS, the important measure in this case) through the supply conductors is greater, causing greater IR drop (through power lines, house wiring and power cords) before the voltage gets to the device being powered, and it reduces how much REAL power the circuit can consume before a (current-activated) fuse or circuit breaker blows due to excess current.
thanks for taking the time to explain all that, at length.
PF and DPF. In my ignorance I did not know that DPF existed nor by implication that PF was any different from DPF.
I will need to go and do some homework.
Hello Rick,
There are many reasons why:
a) Big capacitors used are generally electrolytic capacitors. These kinds of components have many defaults (coloratura). Increasing or adding capacitors adds defaults too!
b) Big capacitors have very low ESR and charge during very short time. This charging time is critical because of excessive current flowing across power supply. Charging time occur 100 or 120x by second and timing charge can vary from 20% to 1% of time. Example for a device fed by 40mA DC current.
- If charging is performed during 20% of time (who is a very very long time), current for charging cap is 160mA.
- If charging capacitors is performed during 10% of time, current is 320mA.
- If charging capacitors is performed during 5% of time, current is 640mA.
- If charging capacitors is performed during 1% of time, current is 3.2A.
You can imagine effects on the global design and disasters in the ground path...
What about design?
I my personal power supply, I only use 300uF lytic with 10uF MKP before regulators. When you study fine manufactured realizations, you can see that Audio Research only use 100µF with one MKP capacitor and Conrad Johnson use 20uF or 40uF MKP (no lytic at all)!
In those configurations, you need two cascaded regulators to silent undulation.
Quality of filtering capacitors is very important (perhaps as important as capacity itself).
What append when increasing capacitors?
Each time I try to increase capacitors, I obtain worse listening result: high-frequency became very metallic, lost details and poor sound, bass play always play the 'same note' and lost extreme bass.
I design preamplifiers and DAC since many years. In my experience, increasing the size of filtering capacitors never improve anything (subjective)!
There are many reasons why:
a) Big capacitors used are generally electrolytic capacitors. These kinds of components have many defaults (coloratura). Increasing or adding capacitors adds defaults too!
b) Big capacitors have very low ESR and charge during very short time. This charging time is critical because of excessive current flowing across power supply. Charging time occur 100 or 120x by second and timing charge can vary from 20% to 1% of time. Example for a device fed by 40mA DC current.
- If charging is performed during 20% of time (who is a very very long time), current for charging cap is 160mA.
- If charging capacitors is performed during 10% of time, current is 320mA.
- If charging capacitors is performed during 5% of time, current is 640mA.
- If charging capacitors is performed during 1% of time, current is 3.2A.
You can imagine effects on the global design and disasters in the ground path...
What about design?
I my personal power supply, I only use 300uF lytic with 10uF MKP before regulators. When you study fine manufactured realizations, you can see that Audio Research only use 100µF with one MKP capacitor and Conrad Johnson use 20uF or 40uF MKP (no lytic at all)!
In those configurations, you need two cascaded regulators to silent undulation.
Quality of filtering capacitors is very important (perhaps as important as capacity itself).
What append when increasing capacitors?
Each time I try to increase capacitors, I obtain worse listening result: high-frequency became very metallic, lost details and poor sound, bass play always play the 'same note' and lost extreme bass.
I too think the first cap in a rCRC or rCLC smoothing/filtering bank should be smaller than half the total capacitance.
But that first cap sees a pulsing charge and a nearly constant load.
This requires tolerance to ripple.
The first cap of the filter MUST have high Ripple Capacity.
A fairly cheap way to achieve this is, multiple parallel low value caps. 5off 1mF will have better ripple capacity than an expensive, high ripple 4700uF cap.
But that first cap sees a pulsing charge and a nearly constant load.
This requires tolerance to ripple.
The first cap of the filter MUST have high Ripple Capacity.
A fairly cheap way to achieve this is, multiple parallel low value caps. 5off 1mF will have better ripple capacity than an expensive, high ripple 4700uF cap.
cascading power filters ( a lot of guessing and misinformation floating around)
remember folks you want to keep the rectified high current pulses from the far reaches of sensitive circuitry. In other words> Keep the charging loop area small!!
The 1st cap s/b the biggest with a small series element. It's job is to provide the most DC at the peaks of the XFMR for max conversion efficiency. Then use a smaller cap/s with larger series elements to distribute the supply voltage to various other circuits. The more sensitive the circuit the larger the series drop. If you favor larger caps at the end of a power filter string it becomes less effective in RFI/EMI.
remember folks you want to keep the rectified high current pulses from the far reaches of sensitive circuitry. In other words> Keep the charging loop area small!!
The 1st cap s/b the biggest with a small series element. It's job is to provide the most DC at the peaks of the XFMR for max conversion efficiency. Then use a smaller cap/s with larger series elements to distribute the supply voltage to various other circuits. The more sensitive the circuit the larger the series drop. If you favor larger caps at the end of a power filter string it becomes less effective in RFI/EMI.
Star,
could you start a new PSU thread and show us how?
Explain the operation of the circuit and from there take us through the design exercise to select components and values.
It can be links to other designs, not necessarily original.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.