critical conduction mode PFC

A

Adam. Seychell

Guest
I'm learning about PFC converters as a wish to make a 100W PFC front end
as part of a home made power supply I'm building. The web has given me a
some insight in building such a power stage. Now, I have found the
preferred topology for my 100W PFC converter is the critical conduction
mode. But what I don't understand is how do the authorities allow such
high frequency high peak currents on the AC line originating from a
critical conduction mode PFC converter ? Some analysis shows;

I_pk = 2*sqrt(2)*P / Vac_rms

where Vac_rms = RMS of line input voltage
P = average input power
I_pk = peak current through inductor (or line)

The current of the inductor (or line) is a high frequency triangle wave
with a peak that follows the line voltage AC cycle and the RMS of this
waveform is;

I_rms = I_pk / sqrt(6)
{ substituting for I_pk yields }
= 1.1547 * P / Vac_rms


So I_rms is 1.15 times higher that it would be for a resistive load of
the same power. Where is the PFC ?

Adam.
 
"Adam. Seychell" <invald@invalid.com> a écrit dans le message de
news:42719fd8$0$10307$afc38c87@news.optusnet.com.au...
I'm learning about PFC converters as a wish to make a 100W PFC front end
as part of a home made power supply I'm building. The web has given me a
some insight in building such a power stage. Now, I have found the
preferred topology for my 100W PFC converter is the critical conduction
mode. But what I don't understand is how do the authorities allow such
high frequency high peak currents on the AC line originating from a
critical conduction mode PFC converter ? Some analysis shows;

I_pk = 2*sqrt(2)*P / Vac_rms

where Vac_rms = RMS of line input voltage
P = average input power
I_pk = peak current through inductor (or line)

The current of the inductor (or line) is a high frequency triangle wave
with a peak that follows the line voltage AC cycle and the RMS of this
waveform is;

I_rms = I_pk / sqrt(6)
{ substituting for I_pk yields }
= 1.1547 * P / Vac_rms


So I_rms is 1.15 times higher that it would be for a resistive load of
the same power. Where is the PFC ?
Click to expand...
No it doesn't.
You have to have a LPF in front of your PFC that rejects the HF components
under the required levels.


--
Thanks,
Fred.
 
Fred Bartoli wrote:
"Adam. Seychell" <invald@invalid.com> a écrit dans le message de
news:42719fd8$0$10307$afc38c87@news.optusnet.com.au...

I'm learning about PFC converters as a wish to make a 100W PFC front end
as part of a home made power supply I'm building. The web has given me a
some insight in building such a power stage. Now, I have found the
preferred topology for my 100W PFC converter is the critical conduction
mode. But what I don't understand is how do the authorities allow such
high frequency high peak currents on the AC line originating from a
critical conduction mode PFC converter ? Some analysis shows;

I_pk = 2*sqrt(2)*P / Vac_rms

where Vac_rms = RMS of line input voltage
P = average input power
I_pk = peak current through inductor (or line)

The current of the inductor (or line) is a high frequency triangle wave
with a peak that follows the line voltage AC cycle and the RMS of this
waveform is;

I_rms = I_pk / sqrt(6)
{ substituting for I_pk yields }
= 1.1547 * P / Vac_rms


So I_rms is 1.15 times higher that it would be for a resistive load of
the same power. Where is the PFC ?



No it doesn't.
You have to have a LPF in front of your PFC that rejects the HF components
under the required levels.
Click to expand...
translation: filter the crap out of it. The advantage of a Discontinuous
Conduction Mode boost converter is that it can be made to automatically
give PFC behaviour (IIRC by keeping Ton constant). The disadvantage is
the current is a hairy-arsed mess. for a low power level like 100W, its
pretty easy to filter the DCM mess to the required level. At 1kW or so
it gets a lot trickier, and above that level (ish) it tends to be done
as Continuous Conduyction Mode, where the inductor current is sinusoidal
with small triangular ripple (IOW a straight-line approximation, rather
than a bunch of right-angled triangles whose average value is sinusoidal)

Cheers
Terry
 
Terry Given wrote:
Fred Bartoli wrote:

"Adam. Seychell" <invald@invalid.com> a écrit dans le message de
news:42719fd8$0$10307$afc38c87@news.optusnet.com.au...

I'm learning about PFC converters as a wish to make a 100W PFC front end
as part of a home made power supply I'm building....


No it doesn't.
You have to have a LPF in front of your PFC that rejects the HF
components
under the required levels.



translation: filter the crap out of it. The advantage of a Discontinuous
Conduction Mode boost converter is that it can be made to automatically
give PFC behaviour (IIRC by keeping Ton constant). The disadvantage is
the current is a hairy-arsed mess. for a low power level like 100W, its
pretty easy to filter the DCM mess to the required level. At 1kW or so
it gets a lot trickier, and above that level (ish) it tends to be done
as Continuous Conduyction Mode, where the inductor current is sinusoidal
with small triangular ripple (IOW a straight-line approximation, rather
than a bunch of right-angled triangles whose average value is sinusoidal)
Click to expand...
Another thing I found is that DCM results in lots of AC flux in the
boost inductor, which means a great deal more core losses and proximity
effects compared to CCM. Is there a reason why a 100W PFC is not
normally built using the CCM approach ? The costs appear to be similar
between the two types.

How is such heavy line filtering implemented without compromising the PF
? I mean, if a large cap put after the rectifier then the that'll add
harmonics in the line current.
 
"Adam. Seychell" <invald@invalid.com> wrote in message
news:42733f31$0$4659$afc38c87@news.optusnet.com.au...
Terry Given wrote:
Fred Bartoli wrote:

"Adam. Seychell" <invald@invalid.com> a écrit dans le message de
news:42719fd8$0$10307$afc38c87@news.optusnet.com.au...

I'm learning about PFC converters as a wish to make a 100W PFC front
end
as part of a home made power supply I'm building....


No it doesn't.
You have to have a LPF in front of your PFC that rejects the HF
components
under the required levels.



translation: filter the crap out of it. The advantage of a Discontinuous
Conduction Mode boost converter is that it can be made to automatically
give PFC behaviour (IIRC by keeping Ton constant). The disadvantage is
the current is a hairy-arsed mess. for a low power level like 100W, its
pretty easy to filter the DCM mess to the required level. At 1kW or so it
gets a lot trickier, and above that level (ish) it tends to be done as
Continuous Conduyction Mode, where the inductor current is sinusoidal
with small triangular ripple (IOW a straight-line approximation, rather
than a bunch of right-angled triangles whose average value is sinusoidal)


Another thing I found is that DCM results in lots of AC flux in the boost
inductor, which means a great deal more core losses and proximity effects
compared to CCM. Is there a reason why a 100W PFC is not normally built
using the CCM approach ? The costs appear to be similar between the two
types.

How is such heavy line filtering implemented without compromising the PF ?
I mean, if a large cap put after the rectifier then the that'll add
harmonics in the line current.
Click to expand...
If done properly a PFC with CrCM (Critical Conduction Mode) can be easily
made to ZVS (Zero Voltage Switch) and yield low EMI. The only problem with
this topology is the large AC current in the boost inductor and the input
filter due to the higher peak currents but not as bad as DCM. As mentioned,
it is the best choice for PFC <150W due to it's nice clean waveforms an no
large current spikes due to diode recovery time. CCM is used for >150 watts
for reasons previously stated. For less than 50W you my consider DCM with
Ton constant.
The input filters for PFC have larger inductors and smaller capacitors than
non PFC or DC inputs. Luckily the PFC input can tolerate inductive filters
and not oscillate due to their non negative input impedance. These AC inputs
look resistive so they will D-Q the input filter without oscillation.
If I could design inductors with AC resistance < 2X DC resistance, it would
be the topology of choice up to maybe 500 Watts.
Regards,
Harry
 
I read in sci.electronics.design that Harry Dellamano
<harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,
Click to expand...
Use tape instead of round wire?
--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Click to expand...
Not sure it will help. Usually use Litz, does good job on Skin Effect but
not Proximity Effect. Normally need about 300uH and 2.5A peak at 100KHz. I
usually use a lot of flux density to keep the turns down and AC effects on
windings. Takes about 40 turns with a gap. Does not foil (tape) still have
Proximity effects?
Regards,
Harry
 
I read in sci.electronics.design that Harry Dellamano
<harryd@tdsystems.org> wrote (in <QeRce.9394$r81.2176@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:
"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect but
not Proximity Effect. Normally need about 300uH and 2.5A peak at 100KHz. I
usually use a lot of flux density to keep the turns down and AC effects on
windings. Takes about 40 turns with a gap. Does not foil (tape) still have
Proximity effects?
Click to expand...
I don't know that it's better than Litz, but the geometry of any
proximity effect would be quite different. You could either search for a
paper on the subject, or try it and see.

Since the proximity effect is due to the magnetic field, it might not be
a bad idea to try reducing it. Presumably there is an optimum balance
between ordinary resistance (goes up with the number of turns) and
proximity effect (goes down with the number of turns because the
induction goes down).

--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
On Sat, 30 Apr 2005 21:12:19 +0100, John Woodgate
<jmw@jmwa.demon.contraspam.yuk> wrote:

I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <QeRce.9394$r81.2176@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect but
not Proximity Effect. Normally need about 300uH and 2.5A peak at 100KHz. I
usually use a lot of flux density to keep the turns down and AC effects on
windings. Takes about 40 turns with a gap. Does not foil (tape) still have
Proximity effects?

I don't know that it's better than Litz, but the geometry of any
proximity effect would be quite different. You could either search for a
paper on the subject, or try it and see.

Since the proximity effect is due to the magnetic field, it might not be
a bad idea to try reducing it. Presumably there is an optimum balance
between ordinary resistance (goes up with the number of turns) and
proximity effect (goes down with the number of turns because the
induction goes down).
Click to expand...
The problem with tape winding is the large number of effective layers
formed by the turns count. Usually only practical at higher
frequencies when turns count is low - approaching unity - or the
current has mostly LF or DC components.

Critical conduction currents aren't right-angle triangles, nor is the
current discontinuous. Critical conduction mode is, by neccessity,
frequency-varying, which can make EMI control a little tricky. Lowest
conversion frequencies correspond to maximum power.

RL
 
I read in sci.electronics.design that legg <legg@nospam.magma.ca> wrote
(in <2l9a719qfv83r7td45ai9p4po8072ee06u@4ax.com>) about 'critical
conduction mode PFC', on Sun, 1 May 2005:

The problem with tape winding is the large number of effective layers
formed by the turns count.
Click to expand...
How is the number of layers a problem?
--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk
 
On Sun, 1 May 2005 22:08:20 +0100, John Woodgate
<jmw@jmwa.demon.contraspam.yuk> wrote:

I read in sci.electronics.design that legg <legg@nospam.magma.ca> wrote
(in <2l9a719qfv83r7td45ai9p4po8072ee06u@4ax.com>) about 'critical
conduction mode PFC', on Sun, 1 May 2005:

The problem with tape winding is the large number of effective layers
formed by the turns count.

How is the number of layers a problem?
Click to expand...
The near-by current sheets in the conductors have the same immediate
direction, so adjacent and buried layers will have current crowding
towards the sheet edge (high Rac/Rdc ratio), resulting in both higher
leakage fields and power loss, for AC. This is one reason the shape is
not used often in power coupling transformers, if turns count strays
far from unity.

Placing the sheets at right angles to the bobbin axis, as is formed in
a flat spring coil, doesn;t correct for this effect, though helically
formed parts are marketed as being superior in some way - their turns
count is 'fortunately' limited by mechanical fabrication constraints.

Adjacent litz twisted conductor bundles have adjacent turn currents
flowing effectively at right angles to each other, though the averaged
direction is still parallel in the group.

Practical measurements were attempted in conjunction with research
into the use of capacitor stock foil winding methods for the
fabrication of inductors -

"Analysis of ac Resistance in Aluminium Foil, Air Cored Inductors"
N.Williams abd C.Pollock
Proc. of 8th European Conference on Power Electronics and
Applications, 7-9 September 1999
http://manuales.elo.utfsm.cl/conferences/seminarios/EPFL/pc/papers/114.pdf

Another article deals with the same structure for superconducting
material -

"Air-core transformer for high frequency power conversion"
Cheng, Chan abd Sutanto.
Same conference
http://manuales.elo.utfsm.cl/conferences/seminarios/EPFL/pc/papers/338.pdf

I expect there are other references, but just the general discussion
of transformer layering and sectioning will show the principles
involved (admittedly buried within a host of others), even in
uncoupled structures.

Past my bed-time.

RL
 
"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:QeRce.9394$r81.2176@trnddc02...
"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect but
not Proximity Effect. Normally need about 300uH and 2.5A peak at 100KHz. I
usually use a lot of flux density to keep the turns down and AC effects on
windings. Takes about 40 turns with a gap. Does not foil (tape) still have
Proximity effects?
Regards,
Harry
Click to expand...
I would not claim for certain that this is true but.

If you take that program what I wrote and select the round wire diameter
with light insulation closest to the thickness of foil you are using and
make the number of layers equal to your number of turns then you will get
perhaps a wild guess as to what the RAC/RDC ratio is.

Then you have to get the calculator out to do the other sums.

DNA
 
"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dNtde.126$c4.0@newsfe3-win.ntli.net...
"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:QeRce.9394$r81.2176@trnddc02...

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect
but
not Proximity Effect. Normally need about 300uH and 2.5A peak at 100KHz.
I
usually use a lot of flux density to keep the turns down and AC effects
on
windings. Takes about 40 turns with a gap. Does not foil (tape) still
have
Proximity effects?
Regards,
Harry



I would not claim for certain that this is true but.

If you take that program what I wrote and select the round wire diameter
with light insulation closest to the thickness of foil you are using and
make the number of layers equal to your number of turns then you will get
perhaps a wild guess as to what the RAC/RDC ratio is.

Then you have to get the calculator out to do the other sums.

DNA
Click to expand...
Hi DNA,
Using your program with #44 wire (0.002") and 25 layers at 100KHz gives 633
increase in Rac to Rdc.
Sounds large but in the right direction. Less than one or negitive numbers
tend to PMO.
Cheers,
Harry

>
 
"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:l8vde.9531$eC.5548@trnddc07...
"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dNtde.126$c4.0@newsfe3-win.ntli.net...

"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:QeRce.9394$r81.2176@trnddc02...

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>) about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect
but
not Proximity Effect. Normally need about 300uH and 2.5A peak at
100KHz.
I
usually use a lot of flux density to keep the turns down and AC effects
on
windings. Takes about 40 turns with a gap. Does not foil (tape) still
have
Proximity effects?
Regards,
Harry



I would not claim for certain that this is true but.

If you take that program what I wrote and select the round wire diameter
with light insulation closest to the thickness of foil you are using and
make the number of layers equal to your number of turns then you will
get
perhaps a wild guess as to what the RAC/RDC ratio is.

Then you have to get the calculator out to do the other sums.

DNA

Hi DNA,
Using your program with #44 wire (0.002") and 25 layers at 100KHz gives
633
increase in Rac to Rdc.
Sounds large but in the right direction. Less than one or negitive
numbers
tend to PMO.
Cheers,
Harry
Click to expand...
Yoiks! I get Rac/Rdc Ratio = 1.0814?

DNA
 
"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dxIde.3270$c4.2079@newsfe3-win.ntli.net...
"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:l8vde.9531$eC.5548@trnddc07...

"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dNtde.126$c4.0@newsfe3-win.ntli.net...

"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:QeRce.9394$r81.2176@trnddc02...

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>)
about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin Effect
but
not Proximity Effect. Normally need about 300uH and 2.5A peak at
100KHz.
I
usually use a lot of flux density to keep the turns down and AC
effects
on
windings. Takes about 40 turns with a gap. Does not foil (tape) still
have
Proximity effects?
Regards,
Harry



I would not claim for certain that this is true but.

If you take that program what I wrote and select the round wire
diameter
with light insulation closest to the thickness of foil you are using
and
make the number of layers equal to your number of turns then you will
get
perhaps a wild guess as to what the RAC/RDC ratio is.

Then you have to get the calculator out to do the other sums.

DNA

Hi DNA,
Using your program with #44 wire (0.002") and 25 layers at 100KHz gives
633 increase in Rac to Rdc. Sounds large but in the right direction. Less
than one or negitive
numbers tend to PMO.
Cheers,
Harry





Yoiks! I get Rac/Rdc Ratio = 1.0814?

DNA

Something is wrong, That isn't possible, even with just two layers.
Click to expand...
Cheers,
Harry
 
"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:KvMde.11$Ri.1@trnddc08...
"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dxIde.3270$c4.2079@newsfe3-win.ntli.net...

"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:l8vde.9531$eC.5548@trnddc07...

"Genome" <ilike_spam@yahoo.co.uk> wrote in message
news:dNtde.126$c4.0@newsfe3-win.ntli.net...

"Harry Dellamano" <harryd@tdsystems.org> wrote in message
news:QeRce.9394$r81.2176@trnddc02...

"John Woodgate" <jmw@jmwa.demon.contraspam.yuk> wrote in message
news:qHpu5afai8cCFwWZ@jmwa.demon.co.uk...
I read in sci.electronics.design that Harry Dellamano
harryd@tdsystems.org> wrote (in <6ePce.8518$r81.8312@trnddc02>)
about
'critical conduction mode PFC', on Sat, 30 Apr 2005:

If I could design inductors with AC resistance < 2X DC
resistance,

Use tape instead of round wire?
--
Hi John,
Not sure it will help. Usually use Litz, does good job on Skin
Effect
but
not Proximity Effect. Normally need about 300uH and 2.5A peak at
100KHz.
I
usually use a lot of flux density to keep the turns down and AC
effects
on
windings. Takes about 40 turns with a gap. Does not foil (tape)
still
have
Proximity effects?
Regards,
Harry



I would not claim for certain that this is true but.

If you take that program what I wrote and select the round wire
diameter
with light insulation closest to the thickness of foil you are using
and
make the number of layers equal to your number of turns then you will
get
perhaps a wild guess as to what the RAC/RDC ratio is.

Then you have to get the calculator out to do the other sums.

DNA

Hi DNA,
Using your program with #44 wire (0.002") and 25 layers at 100KHz
gives
633 increase in Rac to Rdc. Sounds large but in the right direction.
Less
than one or negitive
numbers tend to PMO.
Cheers,
Harry





Yoiks! I get Rac/Rdc Ratio = 1.0814?

DNA

Something is wrong, That isn't possible, even with just two layers.
Cheers,
Harry
Click to expand...
Well, without reading further, harder, about such things I'm not in a
position to comment further. Yes I am a lazy bastard. However,

dpen = 75/SQRT(f) in mm

So at 100KHz thats about 10 thousandths of an inch. Your foil is 2
thousandths so that's five times less. I know that doesn't include proximity
effect but things might not be so unreasonable.

DNA
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top