conventional power supply

[Ban]

I wanted to share some insights I gained by spicing up a normal supply, in
this case a 12V/4Adc, consisting of a transformer, bridge rectifier and
smoothing capacitor.
In another thread some of us gave their opinions about the choice of the
required components, and I tried to verify.

1. The transformer can be modeled with an ideal ac-source plus interiour
resistance, as we know from any DC-supply. Almost all losses are due to
resistance of the windings. The manufacturer specifies a certain obtainable
output power, consisting of the product Vrms * Irms = P(VA). The input power
is by some percentage higher and the difference is due to losses. It is the
difference between the open circuit voltage Vo and the rated voltage Vrms .
Ri = (Vo-Vrms) / Irms.

2. With the rectifier and cap the load current is not sinusoidal. Only when
the momentary voltage is higher than the remaining voltage on C plus the 2
diodes Vf, the current starts charging the cap. it reaches the max. before
the voltage peak and goes symetrically down to zero. The max. current is a
multiple of the average and the rms value is much higher as with a sine.
Both are mainly dependent on the ratio of the interiour resistance to the
load resistance, not much on the capacitor size. A higher Ri reduces the
peak current a lot, since Ipeak is proportional to 1/sqrt(Rload*Ri)

3. The required size of the transformer is mainly dependent on the Irms of
the current thru its windings, which dissipates in the Ri and heats up the
transformer. So a transformer with a very low Ri has not much advantage,
since the Ipeak and Irms go up accordingly. The losses in the diode bridge
raise too and can be estimated in the datasheet where there are curves for
Ploss (Ipeak/Iav). Since this is for a single diode you have to double the
ratio and multiply with 4. The rectifier losses plus the losses in the cap
Irms^2*ESR plus the output power have to be lower than the transformer
rating.

4. The ripple voltage is not so much lower with a low Ri, mainly dependent
on the cap size.

to be continued

 

[Ban]

to be continued

The output voltage looks like a sawtooth with a little nipple on top, its
main AC component is the 120Hz fundamental, the 2nd harmonic and the 4th,
the higher harmonics decrease continually.

5. Setting up the simulation:
___
.-|___|--o-->|---o----o----o----. o-.
| Ri | D | | | |
| | | .-. | |
/ \ .---)-->|---' | |ESR | |
( ~ ) | | D | | |- .-.
\_/ | | '-' / \ | |
|Vo | | |+ ( I ) | |
| | | === \_/ '-'
| | '--|<---. --- |+ |
| | D | |C | |
| | | | | |
'----o------|<---o----o----o----' o-'
D |
=== I-source Rload
GND
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)
Vo = open circuit voltage of transformer
Ri = interiour resistance of transformer
D = diode with 1/2 the average current of the bridge
The load can be a constant current, resistive or a combination. you can also
modulate the current or put a modulated voltage source in series with the
resistor to model the load conditions.

Now you can determine the rms_current through Ri, the ripple voltage of
Vout, ripple current of C, peak current through the diode... It is good to
measure only after the inrush conditions have settled down(give an initial
voltage on the cap) and to use a multiple of the mains period i.e. 50ms for
60Hz if you do a FFT.

6. Dimensioning of the components:
Do the simulation with worst case values, i.e. Vo +/-15%, max load current,
C-10%)+20% to see if all the required specs are met.
The Irms trough the tranny must be always smaller than the rated current.
The output voltage higher than the required minimum. Vout no_load smaller
than voltage rating of C etc.

7. Protection of the transformer:
We all know a fuse is required in the primary of the tranny. It should be
slow_blow to allow the inrush current of C. If we do not need the full power
of the transformer, we can use the actual required Power instead.
We calculate Irms= Prms/Vmains * (1+loss_factor) and choose the next higher
value. This will blow the fuse in case of catastrophic failure, but doesn't
help with overheating. Here we can choose a proper PTC-resistor in series
with the primary, best mounted in contact with the winding. But careful to
insulate it well, because the secondary is usually outside.

to be continued with filtering Vripple

 

[Ban]

to be continuedI hope this part finds your interest
9. Filtering the ripple:
I have set up 4 scenarios:

a) doubling the filter capacitor by paralleling another C. Advantage: the
output voltage stays the same, the ripple current can be double. The filter
action is obviously just 6dB better. With a 22mF/35V the cost is high.

b) the same as before, but inserting a small resistor (0R22) between the
caps. We loose 0.9V and gain 13dB suppression. Same ripple current rating,
but higher harmonics get reduced more.

c) instead of the resistor a 1mH/0.22R coil is inserted. Good
suppression -20dB almost only the 120Hz fundamental left. I tried a 10mH
instead, but it needed a long time to stabilize and would "motorboat" at 10
to 20Hz.

d) insertion of a "Gyrator" made from one transistor:

.-----------|<------------.
2N3055 | MUR520 ___ |
o+---------------- --+-o o+-------------- --|___|-+-o
| \ ^ | | TIP35C \ ^ 68mR |
| --- | | --- |
| 1R_ 1R_ | | | ___ ___ | 4001 |
'-|___|-+-|___|---+ | '-|___|-+-|___|-+-->|->|--+
| | | 2.2R | 2.2R | 4001|
| || | | | || | |
'--||-----)---' '-||----)---------'
2.2m|| | 2.2mF || |2.2mF
--- ---
--- ---
2.2m| |
| |
=== ===
GND GND
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

This was reducing the output voltage by 1.5V, but also the ripple by 37dB
to a mere 15mVpp, as much as a10mH coil would have done. Disadvantage not
short circuit proof, needs small heatsink. The second cap from the O/P is an
idea of Bill Sloman.
The second circuit got all protection added and can probably make 20A with a
bit of cooling and modifying the Rs and Cs. I modulated also the O/P current
from 0 to 6A with a 50Hz sinewave, and the resulting output voltage
variation was just 200mV, hardly any 120Hz in there. Would make a good audio
supply.

During the simulations I tried various caps and snubbers across the
rectifier diodes and found another audiophool myth. These caps indeed reduce
the even harmonics in the diode current, but... the odd ones which are 30dB
higher stay identical. So what is the use?

finished, commentaries welcome.

 

[[email protected]]

I hope this part finds your interest
9. Filtering the ripple:
I have set up 4 scenarios:

a) doubling the filter capacitor by paralleling another C. Advantage: the
output voltage stays the same, the ripple current can be double. The filter
action is obviously just 6dB better. With a 22mF/35V the cost is high.

b) the same as before, but inserting a small resistor (0R22) between the
caps. We loose 0.9V and gain 13dB suppression. Same ripple current rating,
but higher harmonics get reduced more.

c) instead of the resistor a 1mH/0.22R coil is inserted. Good
suppression -20dB almost only the 120Hz fundamental left. I tried a 10mH
instead, but it needed a long time to stabilize and would "motorboat" at 10
to 20Hz.

d) insertion of a "Gyrator" made from one transistor:

.-----------|<------------.
2N3055 | MUR520 ___ |
o+---------------- --+-o o+-------------- --|___|-+-o
| \ ^ | | TIP35C \ ^ 68mR |
| --- | | --- |
| 1R_ 1R_ | | | ___ ___ | 4001 |
'-|___|-+-|___|---+ | '-|___|-+-|___|-+-->|->|--+
| | | 2.2R | 2.2R | 4001|
| || | | | || | |
'--||-----)---' '-||----)---------'
2.2m|| | 2.2mF || |2.2mF
--- ---
--- ---
2.2m| |
| |
=== ===
GND GND
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

This was reducing the output voltage by 1.5V, but also the ripple by 37dB
to a mere 15mVpp, as much as a10mH coil would have done. Disadvantage not
short circuit proof, needs small heatsink. The second cap from the O/P is an
idea of Bill Sloman.

Thanks for the credit, but while I can't remeber where I got the idea
from, it certainly isn't one that I invented

The second circuit got all protection added and can probably make 20A with a
bit of cooling and modifying the Rs and Cs. I modulated also the O/P current
from 0 to 6A with a 50Hz sinewave, and the resulting output voltage
variation was just 200mV, hardly any 120Hz in there. Would make a good audio
supply.

The protection on the second circuit is a bit iffy - two 1N4001 diodes
don't have a very predictable voltage drop. Philips used to do a diode
that did have a closely specified voltage drop (but I can't remember
the number) and I think the BC184 and BC214 low powered audio
transistor have reasonably closely specified base-emitter voltage
drops.

A "synthetic diode"

-+---- ----+-
| \ ^ |
| --- |
| | |
| v |
| ___ |
+---| |---+
---

might be even better - you have to set the potentiometer to get exactly
the voltage you want, but that lets you take out the tolerance on both
the base-biasing transistor and the on the power transistor that you
are biassing.

During the simulations I tried various caps and snubbers across the
rectifier diodes and found another audiophool myth. These caps indeed reduce
the even harmonics in the diode current, but... the odd ones which are 30dB
higher stay identical. So what is the use?

finished, commentaries welcome.

Nice one Ban.

Bill Sloman, Nijmegen

 

[Fred Bartoli]

Thanks for the credit, but while I can't remeber where I got the idea
from, it certainly isn't one that I invented

Looking it the other way, this gyrator is just a 2nd order S-K filter,
filtering its own supply.

 

[Winfield Hill]

Ban wrote...

During the simulations I tried various caps and snubbers across
the rectifier diodes and found another audiophool myth. These
caps indeed reduce the even harmonics in the diode current,
but... the odd ones which are 30dB higher stay identical. So
what is the use?

That's not what they are for. Your entire analysis has a huge
error, in that you ignore the transformer's leakage inductance.
As I showed in multiple postings of measurements on ac-line
transformers, voltage drops from leakage inductance typically
exceed those from copper resistance. The s.e.d. threads also
have extensive theory and formulas to calculate transformer
leakage inductance, and these reinforced my observations.

While the inductance does not contribute to power dissipation,
it has a *huge* effect on your rectifier and filter-capacitor
calculations. This is good, because it serves to reduce the
inrush current at the top of each half-cycle, spreading it out.

A complete and accurate analysis will have to include not only
the transformer's leakage inductance but the diode's reverse-
recovery time model. You have some more learning ahead of you.

Back to your snubber, this is to damp the inductive pulse and
high-frequency ringing that occurs from the rectifier diode's
fast snap-off after the reverse-recovery time is finished. In
this case it's the transformer's leakage inductance that was
charged during the reverse-recovery time and forms a resonant
circuit with the winding and diode capacitances. You add more
capacitance in parallel to lower the ringing frequency and take
control of the resonant energy, plus a series resistor to lower
the Q and damp the ringing to a single cycle. (We've written
up a nice story about this scene, complete with scope photos
and spice models for the 3rd edition of AoE.)

Anyway, instead of a very sharp high-voltage spike (too short
to see with a scope unless you trigger properly) followed by
RF ringing, you get a single small long smeared-out innocuous
pulse. If left unattended to, the magnetic radiation from the
inductive spikes can be picked up and create an audible buzzing
signal (harmonics of 120Hz) in low-level phono or mic inputs.

 

[Ban]

That's not what they are for. Your entire analysis has a huge
error, in that you ignore the transformer's leakage inductance.
As I showed in multiple postings of measurements on ac-line
transformers, voltage drops from leakage inductance typically
exceed those from copper resistance. The s.e.d. threads also
have extensive theory and formulas to calculate transformer
leakage inductance, and these reinforced my observations.

While the inductance does not contribute to power dissipation,
it has a *huge* effect on your rectifier and filter-capacitor
calculations. This is good, because it serves to reduce the
inrush current at the top of each half-cycle, spreading it out.

A complete and accurate analysis will have to include not only
the transformer's leakage inductance but the diode's reverse-
recovery time model. You have some more learning ahead of you.

Back to your snubber, this is to damp the inductive pulse and
high-frequency ringing that occurs from the rectifier diode's
fast snap-off after the reverse-recovery time is finished. In
this case it's the transformer's leakage inductance that was
charged during the reverse-recovery time and forms a resonant
circuit with the winding and diode capacitances. You add more
capacitance in parallel to lower the ringing frequency and take
control of the resonant energy, plus a series resistor to lower
the Q and damp the ringing to a single cycle. (We've written
up a nice story about this scene, complete with scope photos
and spice models for the 3rd edition of AoE.)

Anyway, instead of a very sharp high-voltage spike (too short
to see with a scope unless you trigger properly) followed by
RF ringing, you get a single small long smeared-out innocuous
pulse. If left unattended to, the magnetic radiation from the
inductive spikes can be picked up and create an audible buzzing
signal (harmonics of 120Hz) in low-level phono or mic inputs.

THX Win,

I have here the Wai-Kai Chen "Circuits and Filters" with the proper
description pg.338 figure 10.65. How do I simulate it with an ideal
transformer, and how big are these inductances in your opinion. I want to
simulate a 500VA toroid with 2X45V.

 

[Jonathan Kirwan]

<snip>
(We've written
up a nice story about this scene, complete with scope photos
and spice models for the 3rd edition of AoE.)
<snip>

Arghhh!

Jon

 

[Fred Bartoli]

THX Win,

I have here the Wai-Kai Chen "Circuits and Filters" with the proper
description pg.338 figure 10.65. How do I simulate it with an ideal
transformer, and how big are these inductances in your opinion. I want to
simulate a 500VA toroid with 2X45V.
--

From a 630VA, 230V-2x(2x65V) I've measured some months ago:

Lp=12.7H
N=3.34
Lf, seen from the secondary side, measured @10kHz :
Lf1a=Lf1b=16uH
Lf2a=24uH, Lf2b=22uH

 

[Ken Smith]

pulse. If left unattended to, the magnetic radiation from the
inductive spikes can be picked up and create an audible buzzing
signal (harmonics of 120Hz) in low-level phono or mic inputs.

and ...

There are harmonics up to, and beyond, the AM band in the turn off of the
diodes. As a result, you want the snubber to be near the rectifier to
reduce the area enclosed by the wiring.

 

[Ken Smith]

d) insertion of a "Gyrator" made from one transistor:

.-----------|<------------.
2N3055 | MUR520 ___ |
o+---------------- --+-o o+-------------- --|___|-+-o
| \ ^ | | TIP35C \ ^ 68mR |
| --- | | --- |
| 1R_ 1R_ | | | ___ ___ | 4001 |
'-|___|-+-|___|---+ | '-|___|-+-|___|-+-->|->|--+
| | | 2.2R | 2.2R | 4001|
| || | | | || | |
'--||-----)---' '-||----)---------'
2.2m|| | 2.2mF || |2.2mF
--- ---
--- ---
2.2m| |
| |
=== ===
GND GND

On the second circuit:

Using a second transistor in place of the two 1N4001's produces a sharper
knee in the current limit.

The two 2.2 Ohm resistors will have a huge current flowing in them if the
output gets shorted. You could look at using a PTC device in series with
one of them, or in place of it.

 

[Winfield Hill]

Ban wrote...

THX Win,

I have here the Wai-Kai Chen "Circuits and Filters" with the
proper description pg.338 figure 10.65. How do I simulate it
with an ideal transformer, and how big are these inductances
in your opinion. I want to simulate a 500VA toroid with 2X45V.

I don't have that book, but their ideal transformer model isn't
very useful if it fails to include Leakage Inductance. You can
read one of the posts I was referring to by following this info:

From: [email protected] (Winfield Hill)
Subject: Re: Leakage Inductance, measurements, etc.
Date: 1998/01/03
Message-ID: <[email protected]>

http://groups.google.com/group/sci.electronics.design/msg/30d3845f697a9eec

If you examine my table of transformer measurements, covering
2.5 to 100-watt Signal Transformer "241" split-bobbin types,
http://www.belfuse.com/Data/DBObject/page_22.pdf (223kB long)
http://www.belfuse.com/Data/DBObject/signalcatalog.pdf (6MB),
you'll see the copper resistance dominates for the small ones
(where the manufacturer thinks he can get away with using a
small wire size without overly heating the transformer), and
is on the same order as the leakage-inductance reactance for
the larger ones. So, lacking better information, you could
choose your modeling leakage-inductance values that way.

Google says the thread, "Leakage Inductance, Please Explain?"
started by Chris Carlen on 31Dec1997, has 131 posts. But there
were also offshoot threads at the same time. Anyway, you'd do
well to read the entire 131-post thread, starting here:
http://groups.google.com/group/sci.electronics.design/msg/4873c3c4428471d

A toroid transformer may have lower leakage inductance than
the split-bobbin types I measured, if the secondary is wound
directly over the primary. But if that's the case, it'll also
have much poorer long-term high-voltage-spike insulation, an
issue that most users have learned is very important. Anyway,
we've learned that Leakage Inductance is a good thing, helping
to reduce rectifier-to-capacitor in-rush charging currents.

 

[Winfield Hill]

Fred Bartoli wrote...

From a 630VA, 230V-2x(2x65V) I've measured some months ago:

Lp=12.7H
N=3.34
Lf, seen from the secondary side, measured @10kHz :
Lf1a=Lf1b=16uH
Lf2a=24uH, Lf2b=22uH

And the DCR? Was the N=3.34 an actual measurement?

 

[Winfield Hill]

Jonathan Kirwan wrote...

Arghhh!

Arghhh?

 

[Jonathan Kirwan]

Jonathan Kirwan wrote...

Arghhh?

You are quoting some text I'd like to read, from a book I'd like to
have.

Jon

 

[Tony Williams]

While the inductance does not contribute to power dissipation,
it has a *huge* effect on your rectifier and filter-capacitor
calculations. This is good, because it serves to reduce the
inrush current at the top of each half-cycle, spreading it out.

[snip]

W.M Flanagan's book on transformer design has a chapter
where he presents design curves similar to O.H Schade's,
but which include the effects of leakage inductance.
Very useful for transformers with high leakage inductance,
such as the modern divided bobbin transformers.

Handbook of Transformer Design and Applications.
McGraw Hill, ISBN 0-07-021291-0. Pages 18.1 to 18.15.

 

[Phil Allison]

"Ban"

How do I simulate it with an ideal transformer, and how big are these
inductances in your opinion. I want to simulate a 500VA toroid with 2X45V.

** The leakage inductance of a toroidal transformer is *entirely
negligible* as far as the behaviour of a mains supply frequency rectifier
and capacitor input filter system is concerned.

Same goes for a mains frequency transformer where the secondary is wound
over the primary - whether or not there are two such combinations on the
same core ( ie as with "double C- ore " or R-core types).

Only with " split bobbin " construction does leakage inductance become a
significant cause of voltage drop - but in most examples winding
resistance is still the dominant cause.




........ Phil

 

[Ban]

I don't have that book, but their ideal transformer model isn't
very useful if it fails to include Leakage Inductance. You can
read one of the posts I was referring to by following this info:

From: [email protected] (Winfield Hill)
Subject: Re: Leakage Inductance, measurements, etc.
Date: 1998/01/03
Message-ID: <[email protected]>

http://groups.google.com/group/sci.electronics.design/msg/30d3845f697a9eec

If you examine my table of transformer measurements, covering
2.5 to 100-watt Signal Transformer "241" split-bobbin types,
http://www.belfuse.com/Data/DBObject/page_22.pdf (223kB long)
http://www.belfuse.com/Data/DBObject/signalcatalog.pdf (6MB),
you'll see the copper resistance dominates for the small ones
(where the manufacturer thinks he can get away with using a
small wire size without overly heating the transformer), and
is on the same order as the leakage-inductance reactance for
the larger ones. So, lacking better information, you could
choose your modeling leakage-inductance values that way.

Google says the thread, "Leakage Inductance, Please Explain?"
started by Chris Carlen on 31Dec1997, has 131 posts. But there
were also offshoot threads at the same time. Anyway, you'd do
well to read the entire 131-post thread, starting here:
http://groups.google.com/group/sci.electronics.design/msg/4873c3c4428471d

A toroid transformer may have lower leakage inductance than
the split-bobbin types I measured, if the secondary is wound
directly over the primary. But if that's the case, it'll also
have much poorer long-term high-voltage-spike insulation, an
issue that most users have learned is very important. Anyway,
we've learned that Leakage Inductance is a good thing, helping
to reduce rectifier-to-capacitor in-rush charging currents.

Win, many thanks for the references, I have saved them. The book I'm
referring to is giving this model:

leakage inductance leakage inductance
___ ___ ___ ___
o-|___|--UUU--+----. .--UUU--|___|-o
| '-. ,--'
C| )|(
C| )|(
C| .-' '----------------o
| |
o-------------+----'

magnetizing inductance

And the actually non-linear core losses can be approximated with a resistor
in parallel with the magnetizing inductance.
BTW Gordon E. Carlson who wrote the chapter about transformers gives AoE in
his References. He must like your book very much, because I didn't find much
about transformers and leakage inductance there.

That book is one that requires a lot of study, very theoretical. And
2991pages, so I have a lot to study, you are right. Thanks again for your
constructive comment. It gave me the energy to start going also through the
magnetics, something I never was interested, like electrical machines and
motors etc.
Another thing is to really check how the Spice models are fitting. I use
Spice since 1984 and that was after my time at university. It works fabulous
for the opamp circuits I usually do, but I know the shortcomings in this
area.

 

[Phil Allison]

"Winfield Hill"

A toroid transformer may have lower leakage inductance than
the split-bobbin types I measured,

** You bet your fat arse it is lower.

if the secondary is wound directly over the primary.

** Err - there is *another* method to wind a toroidal ???

But if that's the case, it'll also
have much poorer long-term high-voltage-spike insulation, an
issue that most users have learned is very important.

** What a load of posturing drivel !!!

Toroidal transformers using impregnated cloth or plastic tape insulation
have proved to be VERY reliable over many decades of use - long
outlasting the equipment they power.

Anyway,
we've learned that Leakage Inductance is a good thing, helping
to reduce rectifier-to-capacitor in-rush charging currents.

** ROTFL.

Better chuck out all those overwound and toroidal transformers now !!

Win the Expert has no time for them !!!




........ Phil

 

[Winfield Hill]

Jonathan Kirwan wrote...

You are quoting some text I'd like to read, from a book
I'd like to have.

Arghhh! ;(