I am making a step up DC-DC converter using a full bridge configuration and bridge rectifier output, using high speed rectifier diodes.
I see that in many designs on the net, they use an additional inductor on the output side between the rectifying diodes and smoothing capacitor. I have tried this, but it causes horrific oscillations in the voltage at the rectifier and they very easily go over their voltage limit. A snubber circuit would make this more useable but I am then disipating a lot of power in it. What is the purpose of this inductor and is it really necessary?

The inductor is a core element in any switch-mode system that transfers energy from an input to an output circuit which operate at different DC voltages. It is used as a short term energy store that is 'charged' by the input circuit and then 'discharged' into the output circuit over each switching cycle - with very little energy lost.

Two types of oscillations are often encountered in switch-mode circuit development: a very brief 'ringing' (typically a rapidly decaying cycle or two at MHz) when the circuit switches from discharging to charging the inductor which can be limited by adding passive 'snubber' components and a much slower oscillation (typically in the 10-1000Hz region) which can be eliminated by tailoring the frequency response of the feedback loop that controls the output voltage.

 

Actually it's just feeding back more of the signal to the error amplifier.
A correctly designed error amplifier has 90 degrees phase advance on the input at that frequency. 180 degrees lag from the LC filter, 180 degrees because the error amplifier inverts. Total = 270 degrees. No oscillation.
It's like the dominant pole capacitor in a power amplifier. It feeds MORE signal back to the input of the voltage amplifier stage at higher frequencies, which makes it LESS likely to oscillate at higher frequencies.

A useful rule of thumb: ensure the slope of the feedback loop's open-loop frequency response is no steeper than 6dB/octave where it passes through unity gain - and allow a reserve for variations in the (unstabilised) open-loop gain. The open-loop gain can be measured during normal operation by peturbing the loop - switching rate ripple and general switch-mode noise make for a demanding exercise.

 

A useful rule of thumb: ensure the slope of the feedback loop's open-loop frequency response is no steeper than 6dB/octave where it passes through unity gain - and allow a reserve for variations in the (unstabilised) open-loop gain. The open-loop gain can be measured during normal operation by perturbing the loop - switching rate ripple and general switch-mode noise make for a demanding exercise.

a responose falling at 6dB/octave generally indicates a phase shift of 90 degrees - so we're in agreement.
And putting a 4-pole filter with 24dB/octave inside the loop isn't easily going to satisfy that criterion!

 

Two types of oscillations are often encountered in switch-mode circuit development: a very brief 'ringing' (typically a rapidly decaying cycle or two at MHz) when the circuit switches from discharging to charging the inductor which can be limited by adding passive 'snubber' components and a much slower oscillation (typically in the 10-1000Hz region) which can be eliminated by tailoring the frequency response of the feedback loop that controls the output voltage.

It's definitely not the latter, as it's at 1MHz, and it looks far more serious than the former, which is generally due to intrawinding capacitance or diode capacitance.

 

@LowQCab

Thanks for your reply.
Your post was so long it was hard to reply to every point so here is a short version.
Most people dont take that much interest in these kinds of things so it is nice to see someone that has your level of interest. Perhaps you should create an article presenting all your ideas.

Here are just a few of my replies to yours...

START_QUOTE
Quoting > MrAl
""DC current in the primary will eat up lots and lots of power.""
Mostly what it eats up is Core Inductance, Permeability, etc.,
END_QUOTE
The power loss occurs in other parts of the system, such as in the transistors as they
try to deliver a DC current to an otherwise AC application. The DC 'impedance' is
very very low as compared to the AC impedance and thus puts a strain on the drive
part of the circuit which then has to dissipate more energy to be able to supply the
increased current requirement all just to satisfy the DC current that should not even
be there in the first place.

START_QUOTE
As for Asymmetrical Drive, ....
anything you can do to speed-up any part of the turn-on, or turn-off, of the Gate,
without creating oscillations, is always a good thing.
END_QUOTE
That's not exactly the end of the story.
A controlled drive wave means we can control more such as lower amplitude ringing
and that means less danger to the transistors as well as less radiated energy.
With an incredibly fast rise and fall time, the ringing is maximum and then so is
the radiated energy. Also, the snubber has to be more robust to handle that.
Thus anything you can do to speed up part of the turn on/off is not always a
'good thing' it can actually be a bad thing.
Also, in a careful modern design gate power dissipation would be another factor to
consider.

START_QUOTE
Quoting > MrAl
""The output inductor needs to be one that can handle significant DC current.""
Always go by the recommendations of the NEC (National Electrical Code (US)),
AS A MINIMUM wire Gauge size for a particular Amperage Circuit.
This is based on heat generation.
You can "fudge" on this by no more than one wire gauge number.
END_QUOTE
I am not talking about wire gauge, per se. To handle DC current an inductor has
to be designed such that a DC current does not take it up too high on the BH curve
and thus make it saturate with even a small amount of DC and AC current.


START_QUOTE
Quoting > MrAl
""A regular core with regular wire will probably saturate even with 100ma.""
This is a very uninformed statement, or generalization.
It is false.
First there is no such thing as a "regular core".
END_QUOTE
By a 'regular core' i mean one that was not designed in any special way and with
somewhat higher permeability that sometimes seems very desiriable so as to get the
inductance value up high with fewer turns of wire and maybe smaller size.
An inductor with even a level of permeability of 1000 could easily saturate without
a gap of some kind. A distributed gap would help here too though but that requires
more careful core material selection.


START_QUOTE
A physically larger, and/or, heavier Core can almost always handle
more power, WITHOUT SATURATION, than a smaller version. SIZE MATTERS
END_QUOTE
Perhaps, but you would not use a large core in a small core application not only
because of the physical size but because of the power loss associated with the core
material at the chosen frequency.

 

Getting switch-mode converters right is riddled (haha!) with pitfalls - to achieve high efficiency, so many things have to be right to avoid semiconductor failure, achieve high efficiency, maximise safety and curb emissions. Most professional electronic engineers have limited understanding of inductive components, could not select the one best suited to a particular switch-mode application from an available range; very few of them know how to design a 50/60Hz mains transformer let alone a switch-mode HF transformer or choke. The choice of rectifier type also has a major bearing on switching and other losses - and it goes on and on. A tough area for hobbyists.

 

ONe of the very few things that disappointed me about my university course was the lack of practical discussion on transformer design. We were lectured at a very stately pace by a Dr. Ferrari on electromagnetism, but there was very little on turning our knowledge of B-fields, H-fields, magnetomotive force, reluctance, permeance etc. into a practical design for a transformer.

 

Getting switch-mode converters right is riddled (haha!) with pitfalls - to achieve high efficiency, so many things have to be right to avoid semiconductor failure, achieve high efficiency, maximise safety and curb emissions. Most professional electronic engineers have limited understanding of inductive components, could not select the one best suited to a particular switch-mode application from an available range; very few of them know how to design a 50/60Hz mains transformer let alone a switch-mode HF transformer or choke. The choice of rectifier type also has a major bearing on switching and other losses - and it goes on and on. A tough area for hobbyists.

Yes that is a good way to put it.
If you look back at my post you will see several things i mentioned but that's not even the whole story. Perhaps we could start a community list that features all the issues and design points involved in all but the simplest converters.

I was 'lucky' enough to have worked in the industry for a number of years and some of the things i saw would never be thought of unless you had the experience to see those things take place and the problems they caused.
We did mostly high power (500 watts and above) synthesized sine converters that boasted 90 percent efficiency and 1 percent THD. It took a while to get to that point though and some of the problems that came up were interesting. Some of the designs required a huge output transformer you could not even pick up by yourself it was so heavy. One of the problems that came up with a fluctuating DC current in the primary, which would wax and wane, which repeatedly took the core up high on the BH curve then back down, then back high again, etc. This combined with the 50 or 60Hz sine frequency would cause a very loud audible noise which bothered the heck out of some customers that expected fairly quiet operation in the computer room. We had to design a secondary feedback circuit to compensate for this, as well as add expensive sound deadening material to the outer case shell which BTW was sometimes as big as a small house closet.

So anyway, a list of checkpoints and issues would help others who wanted to roll their own converter especially hobbyists.

 

ONe of the very few things that disappointed me about my university course

I have heard it said that you learn how to on the job and all college does is teach you how to learn.

 

I have heard it said that you learn how to on the job and all college does is teach you how to learn.

I completely agree. It would have been impossible to learn transformers without the background in electromagnetism, but without any practical examples at the time, nothing said to me "pay attention, this isn't just theory, this has practical uses, you'll need to know this"

 

I would like to thank everyone who has participated here for renewing
my interest in a subject that I haven't played with for over 30 years.

Filter Design has always been a sore spot,
especially with any frequencies above, let's say, ~30khz.
This prompted me to dig deeper, and try to get to the bottom line
on the seemingly un-Scientific, random, black art,
of the goings on inside a Coil of Wire.

I made 2 discoveries which I was totally unaware of which
I feel are of great significance.

1) The supposed "Capacitive Coupling" between windings,
while being seemingly "obvious",
falls apart as a model, under certain odd circumstances.

WHY,

2) That's because a coil of wire is, at all times,
acting as a "Helical Antenna",
usually wrapped back on its self,
instead of being used to tightly project an
Electro-Magnetic RF Waveform to another Antenna(s).
Which means that there are a crazy number of random Travelling Waves
going 'round and 'round, augmenting, and cancelling, each other.
So, if you should have enough "coincidental" aspects of your Coil of Wire,
and/or, the Core it's wound on,
and/or, the aspects of their relationship to each other,
should all come together in a certain way, at the same time,
you could have just built an RF Oscillator by accident.
( This is totally aside from the known interactions between a Coil and a Capacitor ).

This "Helical Antenna",
(a particular Antenna design that I am reasonably familiar with),
is not correctly tuned for any particular frequency range,
except maybe by accident,
and therefore WILL HAVE various dips and peaks in Frequency Response,
which can appear to be quite random.
But they are actually not random at all,
just extremely complex in most circumstances.

What this comes down to is that,
precision symmetry in a Coil will cause the various, unavoidable,
resonances in in Frequency Response in the Coil to be more pronounced.
And conversely, the more "randomized" the various parameters are in the
construction of the Coil,

( Turn Spacing,
Turn Diameter,
Wire Dimensions,
Core Cross Section Variations,
Core Material Variations throughout the Length of the Core,
and probably several other Factors that I'm probably missing ),

the less pronounced the resonances will be across the Full Frequency Spectrum.

There is a Voltage Component, and a Current Component, to deal with.
I'm starting to think that the real trick is to minimize the Voltage Component,
and Maximize the Current Component,
but don't ask me exactly how that can work,
even though, I'm fairly sure that it is a workable theory.
This is done, or vice/versa, and well known, in Antenna Design,
and a Coil IS an Antenna,
even if it's a really bad one,
and it's a disaster when it accidentally turns out to be a "good" Antenna.

The trick is,
to create an Electromagnet with a completely self-contained Magnetic Field,
that doesn't have a propensity to act like an Antenna that has been bent into a circle.

There are Conical Wound Air Core Chokes,
and Conical Wound Antennas as well,
both are designed to be "Broad-Band" devices ....... go figure.

Break-up, or randomly re-distribute,
anything, or any dimension, that can lead to Oscillations of any kind.

This should lead to having a High Frequency Inductor, or Transformer,
that has no "SRF" Self Resonating Frequency that is readily identifiable.

This may also mean that a larger Core, or more Turns,
are required for the same Inductance Value,
but possibly, it may not necessarily be so different, or terrible,
at the lower fundamental frequencies that are useful to Hobbyists.

Part of the solution to "SRF" could simply be a winding that goes from
very tight, to widely separated, and spaced apart,
as the windings are distributed around the Core.

Maximum Efficiency may turn out to be an undesirable target for a Hobbyist,
and the priority may swing towards making a highly damped,
albeit somewhat less efficient,
easy to build and use Coil,
with no bad manners.
( If that's even possible )
.
.

 

Some very interesting thoughts there.
Winding geometry is very important. If you think of a 2-layer "solenoid" winding. If, at the end of the first layer, the coils turn around and go back, it has a very different effect that if they start again at the other end. Even though the capacitance between layers is exactly the same, the voltage between the layers is completely different - full supply voltage at one end and next to nothing at the other.
There must be a point, as the frequency is increased, where it stops behaving as a single "bulk" capacitor and inductor and starts behaving as a whole series of L then C then L then C, as in a transmission line. For most switched-mode frequencies, I think it is the former.
The ferrite (or iron powder) cored inductor has an advantage over the air cored - it is smaller and has fewer turns, therefore less capacitance between turns for the same inductance, therefore higher self-resonant frequency. But fewer turns means more volts between adjacent turns.
The toroid is the ideal core shape for preventing stray field, but tricky to wind. A single layer winding would be about 60 turns in the centre but about 100 around the outside, so either an intrawinding capacitance problem in the centre or a poor coupling problem around the outside, if it's part of a multi-winding transformer.
My preferred solution is a bobbin with a single-layer winding. Often it seems like a lot of winding area is going to waste, but the lack of proximity effect makes up for it. The hard-to-get ERL cores are excellent (like the ER cores but longer).
I tried the "scramble" winding on some low-power half-bridge designs, with some success, but have yet to try it on anything bigger; but I wonder how much one scramble-wound transformer varies from the next; and whether a winding machine could do it repeatedly!

 

If you go by the elementary explanations of how a transformer works,
the only thing that matters is, how many times does the conductor pass through the middle of the Core.
In that case, you could have a very compact winding that is virtually the same dimensions as the Core,
and it would work just as well as an evenly "distributed" winding.
And maybe it actually does .......
A compact winding produces the greatest inductance for a given wire length / number of Turns ........
But that would be more expensive to produce, or too big, (or time consuming).

The Wire Length will "predict" the extreme, unwanted, frequencies that will self resonate.
And, the magic number Isss ........... One Quarter of a Wavelength, ........ just like with tuning an Antenna !!!!
9 feet of wire, and yepp, there it is, 27mhz peak ............. It's a stinkin' Antenna, all wound-up tight !!!!

Now, the $64,000 question is ........ how do you prevent a Coil from acting like an Antenna ????
How about small Ferrite Beads on the wire before winding ??
Or maybe, forget about the "Skin Effect" and use a massively oversized wire gauge ???,
or maybe even small 1/4 or 1/8" Copper Tubing ???
4ga. solid Copper Wire anyone ???
Or, how about, after the coil is wound on the Core,
the whole assembly gets coated with Epoxy containing Ferrite or Iron Dust particles,
or has anybody tried a thick layer of Tape on the Core before winding ??
A parasitic, "Shorted Turn" on the Core,
the dimensions of which match the estimated Self Resonance Frequency ???

Realistically, I think the answer may lie in making the windings as compact as possible,
and making sure that they are not exactly 180 degrees opposed from one another.
Multiple Windings would be placed at odd intervals around the Core, NOT evenly spaced and pretty.
Cores could even be made with molded-on plastic "Bobbins" around the circumference.
Or better yet, make the Core larger in diameter, with grooves molded in around the outside,
so that the Coils would have the same compact dimensions on the inside and outside of the Core.
.
.

 

Any Comments on this Idea ??

 

Reminds me of the shape of core used in fluorescent light chokes. Probably much the same reasoning applies, albeit at 50Hz instead of 50kHz.

I modelled each pair of your oval cores as a pair of E-cores (I think it works, the result is a figure of 8 shape with two holes in it).
I then invented a core which is a stack of 10 of these pairs, based on an EF25. The results are interesting:

My original EF25 transformer was for a 50W half-bridge at 100kHz, using single layer windings. The calculated loss was 0.5W, and it ran at 100mT
For the stack of ten cores, the optimum value of B reduced to 25mT, and the loss reduced to 0.15W. The number of turns reduced from 92 to 36, but each turn is 5.5 times a long, so the total wire length increased, and presumably, so does the intrawinding capacitance.
Using the original value of B, obviously resulted in a tenth of the number of turns, so the wire length reduces by a half, but the core loss increases, giving a total loss of 2.6W.
Choosing a value of B that gave the same amount of loss as the original transformer, also gave the same wire length.

Now, if my original design had been for maximum bobbin utilisation not single layer windings, things would have probably worked out differently.

 

Core losses are directly related to Frequency, Eddy-Currents, and how close you are getting to Saturation.
The Idea is to provide excessive Core Mass,
in the form of numerous small, cheap, readily available Ferrite or Iron Powder Cores,
this, in combination with ~95% of the surface of the Windings, all running exactly parallel,
and being in a tightly packed bundle, with no where to go,
will insure a very high level of Mutual Inductance, and Inter-Winding Capacitance,
even before the Core(s) is/are taken into consideration.

The Cores basically have the job of concentrating the Magnetic Field generated by the Primary Winding,
and directly applying that Magnetic Field,
through the shortest path possible,
to the Secondary Windings.
But not necessarily having to "transfer" the Magnetic Field
from one side of a circle, over to the other side of the circle, and then into another Winding.

I'm thinking that this arrangement will generate the highest possible level of Induction,
for a given Winding Length,
and the highest level of inter-winding Capacitance,
in the smallest possible physical space,
for the least expense,
and made with the easiest possible construction method.

And that,
it will have the least propensity to act like a poorly designed
"Helical Antenna" that has been twisted into a circle.
In other words, it should not create random oscillations, and/or, standing waves that are unwanted,
because of the geometrical relationships of the various parts.

I wish I still had a proper shop, (and a job), so that I could test this theory out.

When you mentioned "Core Losses", I'm assuming that you were actually referring to
the short length of the windings.
If there is enough Inductance generated by the interaction of the wire and the Core structure,
and no Secondary Winding to reduce that same Inductance,
very little Current will flow in the winding, even if it's only 10" of #4 AWG wire.
Of course this is for a limited period of time, as in "high frequencies".

Back in the day, when everybody had a TV Antenna on their roof,
it was very common for the Antenna, and the TV set Antenna Terminals, to have a "300 Ohm" Impedance,
but a "75 Ohm" Coaxial Cable was used to connect the two together, this necessitated the installation
of a "Balun" or "Matching" Transformer, on each end of the Coaxial Cable to match the Impedances.
The matching transformers had a tiny "Binocular" style Ferrite Core with 2 holes,
and there was 1 single "Turn" of a ~32 gauge wire,
into, and then back out of, the same end of the Core,
for the "75 Ohm" side of the transformer,
and 4 passes on the "300 Ohm" Secondary side.
so the transformers first exchanged Voltage for Current,
and at the other end of the Coaxial Cable,
the transformer changed the Current back into Voltage,
all with extremely low loss of Signal Strength.

The trick was that there was a tremendous level of Induction,
in relation to the amount of POWER (Watts) being transferred,
and the frequencies were very high.
.
.

 

I am making a step up DC-DC converter using a full bridge configuration and bridge rectifier output, using high speed rectifier diodes.

How's it going? Any progress?

 

When you mentioned "Core Losses", I'm assuming that you were actually referring to
the short length of the windings.

Core loss refers to the energy lost in the core (i.e. turned into heat) by the magnetisation of the core being reversed every cycle.
It's generally approximated as being proportional to the square of the peak-to-peak flux excursion.

I remember the 300:75 ohm balun mainly from FM radios. If you peeled back the braid of a piece of 75 ohm coax, and folded the braid one way and the core the other to make a letter T, the length of the top parts of the T being equal to quarter of a wavelength at 100MHz (750mm), then it made a 75 ohm antenna. Being only about 10 miles from the transmitter meant that the usual Yagi-Uda array wasn't required.

If the two ends of the T were joined together with a piece of wire (to make a flattened loop) then it was a 300 ohm antenna.
Then depending on whether the radio had 75 ohms or 300 ohm inputs, you may need the balun. I thought it was was 2:1 turns as the impedance ratio is proportional to the square of the turns ratio.

 

Sorry about that,
I was under the impression that Core Losses were related to Core Material vs Frequency, plus
getting near Saturation, the Core Losses go up exponentially.
I've never heard of a problem caused by excessive Core Mass, ( which my idea definitely has ),
other than being able to fit a workable number of Turns, and Length of Wire, on to the Core.
But excessive Core Mass will not "fix" having an inappropriate Material selection.

Core Mass, and Core Material Properties, set the Saturation Limits for an anticipated Power Transfer Level,
which will be reduced by,
Waveform Harmonics, ( which are out of the Frequency Range of the Core Material ),
and Core Temperature.

I didn't specify a Core Material because I simply assumed that most
common Core sizes are available in a variety of suitable Core Materials.

If there is some aspect that I am not taking into consideration, please let me know.

BTW, on the Balun, I never actually counted the Secondary Turns,
I was only intrigued by the fact that the Primary was basically 3/4 of a Turn.
Power Transformers are almost a simple Multiplication or Division problem.
Antenna and Transmission Line Transformers may use a different Formula, for different reasons,
I don't know, I only casually read about them for interest.
But I am familiar with matching multiple Antennas by using specific lengths of Coax between connections,
and using specific lengths of Coax between the Receiver/Transmitter and the Antenna(s).
It works just fine, and without any Matching Transformers, or odd Impedance Coax sections needed.
.
.