K7ITM wrote:
So, why do you want to use a choke-input filter in the first place?
AFAIK, they are most useful in giving you better output voltage
regulation under varying load than a capacitor input filter. They have
the added advantage that you can get more DC _power_ from a given
transformer by using a choke input filter, because although the output
voltage is lower, the RMS transformer winding current is lowered even
more.

BUT--the voltage regulation advantage is lost if you try this with a
half-wave rectifier circuit, because you cannot maintain constant
enough current in the choke. To get the voltage regulation, the
current in the choke must not drop to zero at any time in the cycle,
and that's not going to happen while maintaining reasonable output
voltage in a half-wave circuit. (There's some limited help if you put
a "catch diode" to keep the voltage across the choke from swinging too
far negative, but that's not enough to get the advantage of the
full-wave circuit.)

In addition, as John says, in the circuit as drawn, the choke is simply
in series with the transformer secondary, so you must reverse the
current in it between half-cycles to get conduction on both
half-cycles. It will not behave anything even close to the way that a
full-wave rectifier feeding a choke input filter will.

Suggest you try a simple Spice (e.g. the free LTSpice from the Linear
Techonolgy website) simulation of this and the normal full-wave
circuit, and look at the huge differences. Note especially what
happens when you vary the DC load on the output.

Cheers,
Tom

I think I understand what you are saying here, but even with
a full wave rectifier doesn't the current through the choke drop
to zero (though only for a brief instant) between the two phases
of rectification when the diodes switch roles? And since there
isn't a capacitor before the chokes the voltage at the input
to the filter would drop to zero, unlike with a capacitor input
filter.

Also, with either type of rectifier (FW or HW) is shouldn't matter
which leg the choke is placed in, as Kirkoff's law is satified
either way.

wa2mze(spamless) wrote:
John Popelish wrote:

Not at all well, because you have provided no path for the inductor
current when the voltage from the transformer tires to reverse bias the
diode. So the inductor will keep the diode conducting as the voltage
reverses. This is not at all the way a choke input filter acts with a
full wave rectifier. I am quite sure you have never seen a choke input
filter in a half wave supply, for this reason.

I guess I can't recall seeing a half wave rectifier circuit using
a choke input filter, but I thought that was because half wave
circuits are usually used in low voltage circuits where a choke
input filter would not have any advantage anyway.

The advantages of a choke input filter (lower RMS transformer current
per amp of DC output, much lower high frequencies in the ripple, less
voltage sag with increase in load current, etc.) are not directly
related to the output voltage. The concept of a choke input filter is
that the current is continuous through the inductor, and so, into the
capacitor. A capacitor input filter charges the cap during brief
pulses at the line peaks, instead, producing a higher RMS transformer
current and higher harmonic ripple components, but also higher peak
output voltage.

However in a full wave circuit how is there an alternate path?
The center tapped transformer simply provides two ac excitations
to two rectifiers 180 degress out of phase.

The inductor current switches from one rectifier to the other as the
transformer voltage goes through zero. At the moment the transformer
voltage is zero, the inductor draws current through both rectifiers,
producing an input voltage to the inductor 1 diode plus transformer
resistance drop below the center tap voltage.

This allows only
one rectifier to conduct at a time. True, there is a more or less
constant excitation to the choke, but there is NO reverse path as
the diodes still only allow conduction in one direction.

I suggest you follow those currents through the inductor for a full
cycle. Since the inductor is in series with the secondary, if the
secondary conducts in both directions (alternating which diode is
conducting) then the inductor must also conduct in both directions.

Either way the choke sees a DC current, not an AC one (minus the
ripple, which a half sinewave imposed on a dc current).

The diodes are turned opposite ways, so one conducts DC one way, and
one conducts DC the other way. Both those currents pass alternately
through the same inductor.

No, with sufficient load current, the choke current never goes to zero.
See the formula for the minimum choke inductance versus supply voltage
and current...the usual formula also assumes 60Hz---120Hz ripple.
Remember: V=L*di/dt. The output DC voltage is the average of the
input voltage to the choke, less any I*R drop in the choke. It MUST
be, or the current in the choke would change until it was. So the
output voltage is (nominally) 0.9* the RMS input voltage, or
0.9/sqrt(2) times the peak voltage, assuming the choke input voltage
tracks the full-wave rectified sine (in other words, the absolute value
of the sine). When the input voltage to the choke is the output
voltage, the choke current increases. When it is less, the choke
current decreases. The average choke current must be the average load
current, or electrons would accumulate somewhere. If the inductance is
large enough, and the cycles come fast enough, then the change in
current is less than enough to make the current go to zero.

It's a bit of calculating to do it for a sine wave. Just think of this
example: a square wave that sits at zero for one second, then at 2
volts for one second, then repeats. Feed it to a choke, say 2 henries.
Output of the choke to a mongo capacitor, so the output voltage
doesn't change significantly. Put a 1 amp load current on it. The
output voltage must be one volt, since that's the average of the input.
So half the time the choke has one volt across it in one direction,
and half the time it has one volt in the other direction. One volt
divided by two henries is half an amp per second. Since the average
current is one amp, the current must swing between 0.75 amps and 1.25
amps. It never goes to zero, or even close. But drop the load current
to 0.25 amps, and the choke current goes just to zero when the input is
at the end of the low period. A bit lower load current, and the choke
current would go negative (or to zero if there's a diode keeping it
going one direction only). That's why a choke input filter looses its
good regulation if the load current gets too small.

In the full wave rectifier, the choke current (if it doesn't drop to
zero) will force the diode output voltage to be one diode drop below
ground, PLUS the I*R drop in the transformer winding, when the voltage
across the outside of the secondary is zero. At that point, both
diodes will be conducting equally, assuming they are matched, and half
the choke current will come from each half of the transformer
secondary.

Hope that helps!

Cheers,
Tom
(off to Holden for a week...no internet there, so you're in John's
capable hands on this one!)

To illustrate this wi

In article , "wa2mze(spamless)"
wrote:

I think I understand what you are saying here, but even with
a full wave rectifier doesn't the current through the choke drop
to zero (though only for a brief instant) between the two phases
of rectification when the diodes switch roles? And since there
isn't a capacitor before the chokes the voltage at the input
to the filter would drop to zero, unlike with a capacitor input
filter.

Also, with either type of rectifier (FW or HW) is shouldn't matter
which leg the choke is placed in, as Kirkoff's law is satified
either way.

Spamless-

Yes, it wouldn't matter what leg the choke is in as long as it is on the
output side of the rectifier (assuming full wave).

For Half Wave, you must also consider the voltage across the choke.
Voltage is L times di/dt where di/dt is very high at the moment the diode
stops conducting. This is why a diode is often placed across a solenoid
or relay coil, to prevent a high voltage pulse across the switching
device. As a side-effect, relay drop-out is slow since current continues
flowing as the magnetic field is discharged.

With the relay analogy in mind, perhaps there could be some advantage if a
diode were placed across the choke of a choke input filter fed with a half
wave rectifier. It would be connected with cathode towards the cathode
end of the rectifier, and would allow choke current to continue flowing
during the off-portion of the rectifier's conduction cycle.

73, Fred, K4DII

Some real brain fodder here. John P. Your Spice model with the coupled
inductors seems to take a divergent turn and I am not sure about this
"coupled" part. Did you try a single inductor.

On to my original thoughts. This really takes me back and requires serious
thought. One thing to keep in mind. An inductor (by virtue of the
magnetic field cutting its own turns) tries to keep whatever current is
flowing, flowing. An inductor will make the voltage across it "do whatever
it takes" to keep that current flowing-- and allow this current to decay
(some say discharge) in what can be considered a normal manner. The diode
on the relay coil is a good example. The voltage can rise very high without
the diode, but using this model, you can figure out what the inductor
voltage does when the normally conducting device turns off.

That said... I had never studied choke input filters to such a degree...
However, its action must allow the filter cap to charge for a longer time,
thus keeping the average diode current lower...
Does the current through the inductor drop to zero in the normal choke input
filter?
If not, a close look at the current path in the full-wave circuit will show
where the current goes at the cross-over points. Brain full - can't figure
out now. I'd have to model it in Spice and watch things

Very interesting thingh. If the current does drop to zero, then it seems
the single choke would work.

Then, reading some of the latter posts, I too, wonder why the desire for
choke input.

73, Steve, K,9;D.C'I



John Popelish" wrote in message
...
Fred McKenzie wrote:
In article , Ken Scharf
wrote:


I was looking at some power supply circuits for
tube linears and was thinking about the full wave
voltage doubler. This is basicly two half wave
rectifiers in series. Now I could build this
circuit with a choke input filter for each half
wave rectifier of the voltage doubler, and I could
put the chokes in the lead without the rectifier.
In this case I could use one choke for both halfs
of the voltage doubler.


Ken-

This doesn't make sense to me. My recollection of the choke-input
filter,
is that it can only be used following a full-wave rectifier. You are
suggesting they be used prior to the rectifier, which is not where a
"filter" is normally placed. Instead, the choke would act as a series
impedance to the AC source.

It seems to me that you can't separate the capacitors from the
rectifiers,
or you wouldn't have doubler action. Therefore, capacitor-input is the
only filtering that makes sense for this circuit. Of course you might
use
the choke in a Pi configuration between the output and another filter
capacitor.

If you have any success with this approach, it will be from extra
voltage
generated by the choke's collapsing magnetic field. This is similar to
how switching regulators work, but without any active regulation.

73, Fred, K4DII

I played around with choke input filtering for this circuit with Spice
and got "continuous inductor current" if I used two highly coupled
inductors, one after each rectifier, and another pair of diodes from
the input side of the chokes to the capacitor common point. However,
this "continuous current" switches back and forth between the two
coupled inductors on alternating half cycles so each end of the
capacitor pair sees current as a half cycle approximately square wave
pulse. So each capacitor charges and discharges with a quite
triangular voltage ripple. But the sum of the two capacitor voltages
is a very pure DC, compared to the no choke version, since the ripples
cancel quite well. However, this reduces the output voltage to only
half of the no choke version, so you might as well have made a full
wave supply, instead of a doubler configuration.

Steve Nosko wrote:
Some real brain fodder here. John P. Your Spice model with the coupled
inductors seems to take a divergent turn and I am not sure about this
"coupled" part. Did you try a single inductor.

Yes. First I tried a single inductor in the common leg of the
transformer. I used a 1 Hy inductor and two 220 uF capacitors and a
100 ohm load resistor across the doubler, with a +-10 volt peak sine
source as a transformer secondary. The inductor current settled to a
+24 mA to -24 mA sine wave. The output voltage across the load
resistor was about .825 volts with a 120 Hz ripple that swung from
about .77 to .88 volts. Not much of a doubler. Almost all the
secondary voltage is dropped across the inductor.

I also tried several different configurations with two separate
inductors, one between each diode and capacitor. Only the coupled
inductor did anything like a choke input filter. And such coupled
chokes exist. See type 2-2690 and 2-2691:
http://www.stancor.com/pdfs/pg56.pdf

On to my original thoughts. This really takes me back and requires serious
thought. One thing to keep in mind. An inductor (by virtue of the
magnetic field cutting its own turns) tries to keep whatever current is
flowing, flowing. An inductor will make the voltage across it "do whatever
it takes" to keep that current flowing-- and allow this current to decay
(some say discharge) in what can be considered a normal manner.

The relation between voltage and current is V=L*(dI/dt). The only way
the current can change is if the inductor has voltage across it.

The diode
on the relay coil is a good example. The voltage can rise very high without
the diode, but using this model, you can figure out what the inductor
voltage does when the normally conducting device turns off.

Yes. The current ramps down as determined by the drop across the
inductance. In this case, that is a diode drop added to the resistive
drop of the coil.

That said... I had never studied choke input filters to such a degree...
However, its action must allow the filter cap to charge for a longer time,
thus keeping the average diode current lower...

Make that "the peak diode current lower". The average diode current
has to be equal to the average DC output current, regardless of the
filter.

Does the current through the inductor drop to zero in the normal choke input
filter?

If the inductance is below the critical value, it certainly does. But
most choke input filters are designed to produce continuous (but
varying current) throughout the cycle at minimum current load. But
all choke input filters will go into interrupted current operation at
some minimum load current.

If not, a close look at the current path in the full-wave circuit will show
where the current goes at the cross-over points. Brain full - can't figure
out now. I'd have to model it in Spice and watch things

Very interesting thingh. If the current does drop to zero, then it seems
the single choke would work.

I guess that depends on what you mean by "works". It cannot ever work
as a normal (continuous current) choke input filter.

Then, reading some of the latter posts, I too, wonder why the desire for
choke input.

It has advantages for lower transformer heating and low line harmonic
currents and improved DC voltage regulation (compared to a capacitor
input filter) with changing load currents (as long as the minimum is
above that which maintains continuous current) and low output line
harmonics above the second. If any or all of those are important to
you, it may justify the high weight and cost of an inductor.

If the inductor in your design is replaced by a piece of wire, the diodes
will directly charge the two capacitors and the connected equipment will
draw charge from the capacitor pair.

If you then replace the wire with an inductor, ALL that will change is that
the capacitors will charge less effectively and you'll have a less
efficient, bulkier and more expensive power supply.

As simple as that? ;-}

John Popelish wrote:
wa2mze(spamless) wrote:

John Popelish wrote:


Not at all well, because you have provided no path for the inductor
current when the voltage from the transformer tires to reverse bias the
diode. So the inductor will keep the diode conducting as the voltage
reverses. This is not at all the way a choke input filter acts with a
full wave rectifier. I am quite sure you have never seen a choke input
filter in a half wave supply, for this reason.


I guess I can't recall seeing a half wave rectifier circuit using
a choke input filter, but I thought that was because half wave
circuits are usually used in low voltage circuits where a choke
input filter would not have any advantage anyway.


The advantages of a choke input filter (lower RMS transformer current
per amp of DC output, much lower high frequencies in the ripple, less
voltage sag with increase in load current, etc.) are not directly
related to the output voltage. The concept of a choke input filter is
that the current is continuous through the inductor, and so, into the
capacitor. A capacitor input filter charges the cap during brief pulses
at the line peaks, instead, producing a higher RMS transformer current
and higher harmonic ripple components, but also higher peak output voltage.

However in a full wave circuit how is there an alternate path?
The center tapped transformer simply provides two ac excitations
to two rectifiers 180 degress out of phase.


The inductor current switches from one rectifier to the other as the
transformer voltage goes through zero. At the moment the transformer
voltage is zero, the inductor draws current through both rectifiers,
producing an input voltage to the inductor 1 diode plus transformer
resistance drop below the center tap voltage.

This allows only
one rectifier to conduct at a time. True, there is a more or less
constant excitation to the choke, but there is NO reverse path as
the diodes still only allow conduction in one direction.


I suggest you follow those currents through the inductor for a full
cycle. Since the inductor is in series with the secondary, if the
secondary conducts in both directions (alternating which diode is
conducting) then the inductor must also conduct in both directions.

Either way the choke sees a DC current, not an AC one (minus the
ripple, which a half sinewave imposed on a dc current).


The diodes are turned opposite ways, so one conducts DC one way, and one
conducts DC the other way. Both those currents pass alternately through
the same inductor.
I tried a mental exercise, I redrew the voltage doubler adding another
winding to the power transformer to provide output 180 degress out of
phase and added two more diodes so I now had each capacitor feed by
both rectified phases. The result, is of course, a full wave bridge
rectifier, but with a center tap of the transformer coupled to the
junction of the two filter capacitors. This is similar to the
dual voltage power supplies so often seen in the ARRL handbooks from
the 60's and 70's for tube transmitters. I suppose a choke could
be placed in the lead from the anodes of one pair of diodes to ground,
so it would be commond to both outputs and the lead from the transformer
centertap isn't needed.