Building Your Own Flyback Transformer

[Joerg]

It does not seem worthwhile to spend time and energy to do experiments on
unknown circuit components, especially without the necessary equipment,
like a scope or an LCR meter. Either of these can be obtained very
inexpensively, especially if you are willing to get an older model and
perhaps do a little repair work (which can be educational as well).

It depends. Sometimes it can be quite sporty to pretend "What if I am on
a lonely island and only my laptop and these few parts?" You can get a
similar circuit going (not with the TIP31 though, and with a forward
concept) using just the sound card and a $2.99 Harborfreight multimeter.

Of course, on that island your laptop battery would croak after two
hours so you've got to be done by then ;-)

But seriously, I was in such situations more than once. Client had a
circuit issue that turned out to be a really nasty noise intrusion from
somewhere. But from where, and what's the path? Next to nothing in
suitable equipment, just my trusty laptop with its 18bit sound card.
However, we needed an iso xfmr that went down to 5Hz and mine quits at
15Hz or so. So we took a mains iso transformer and roached phono jacks
onto it. Of course this just had to be the day the regular OSHA
inspection happened. One engineer turned pale when he saw that. "Put
that away!" ... "Why? We need it right now." ... "Cause the OSHA guy is
next door" ... "Dee-deedle-dum, don't turn around, the commissar's in
town, dee-deedle-dum" ... "Stop that!" (I know the guys quite well, else
I would have pulled that Falco song)

 

[[email protected]]

Just don't expect much out of a TIP31. The 555 doesn't have enough
muscle to drive it properly. So wear goggles, seriously, because with
stuff like this parts can go, in chemistry speak, "exotherm".

Thanks for the goggle reminder. (I wear glasses, so, for years, I was
exempt from the goggle requirement in the undergrad chem labs. No
contact lenses in the chem labs... no no no... the solvents would
dissolve your contacts while on your eye...)

I think I'll try an IRF530 then, since my TIP31A won't do. (Or I
could try paralleling 4 2N2222s...) For a mosfet, that means I have
to increase the gate drive voltage to 10V or so, right? Scary. For
safety's sake I think I'll use a "safety resistor" in series with my
primary coil and the mosfet. A 12V lamp should do...

And you're right, it's more fun to adopt a "deserted islander
mentality" and do this on the cheap, which is partly why I opted to go
discrete for the multivibrator instead of using a 555. On the other
extreme, I could just BUY 6 lead acid batteries and connect them in
series to get my big voltages... but what's the fun in that...

Michael

 

[[email protected]]

I'd have to look it up; I'd probably put "flyback tutorial" or something
in google search.

Good Luck!
Rich

GREAT idea!
Flyback Converters for Dummies
http://www.dos4ever.com/flyback/flyback.html

I never would have guessed... just like I wouldn't ever expect
"Nonlinear Partial Differential Equations for Dummies" either...

Michael

 

[Eeyore]

Not at all, they can work quite well _if_ you keep any DC components out
of the flux.

Yabbut ! Many of those materials are made intentionally very lossy.

Graham

 

[Eeyore]

There is no industry wide standard for color code of toroids.

Would be damn handy if there was.

Your core could be solid ferrite (high permeability, low
saturation amp turns) or any one of a number of powdered
iron formulas. Based on the heavy wire originally on the
core, I suspect it is powdered iron with lots of air gap
distributed all around the core between the iron grains.

Especially if it's one colour on one side with another one the other. White and yellow are very
popular. Damn, which company is that ? Big American one.

Graham

 

[Joerg]

Yabbut ! Many of those materials are made intentionally very lossy.

Less than you might think. I've sent quite a bit of energy across
sub-penny class ferrite beads and none ever got hot.

 

[Joerg]

Would be damn handy if there was.



Especially if it's one colour on one side with another one the other. White and yellow are very
popular. Damn, which company is that ? Big American one.

Nope, it's Adidas:

http://cgi.ebay.com/ADIDAS-NEON-Yel...wItemQQimsxZ20080827?IMSfp=TL0808271238r12282

But seriously, yellow and white is usually a hydrogen reduced EMI
ferrite for low frequency stuff. MicroMetals uses the colors but so do
others.

 

[Joerg]

GREAT idea!
Flyback Converters for Dummies
http://www.dos4ever.com/flyback/flyback.html

Excellent tutorial. I'd save many miles on the freeway if everyone would
be familiar with the scope plot in figure six. When you see that shark
fin it's time for duck and cover.

I never would have guessed... just like I wouldn't ever expect
"Nonlinear Partial Differential Equations for Dummies" either...

A few days ago I saw a book "The Bible for Dummies". At first it shocked
me a bit but hey, whatever it takes.

Anyhow your IRF530 is a lot better than the TIP31 but keep in mind that
whatever is driving it needs to wallop the 700pF gate capacitance around
and it needs to push it up to 10V. And fast.

 

[Eeyore]

Less than you might think. I've sent quite a bit of energy across
sub-penny class ferrite beads and none ever got hot.

The prices you design to never fail to amaze me.

Graham

 

[Eeyore]

one.


Nope, it's Adidas:

http://cgi.ebay.com/ADIDAS-NEON-Yel...wItemQQimsxZ20080827?IMSfp=TL0808271238r12282

But seriously, yellow and white is usually a hydrogen reduced EMI
ferrite for low frequency stuff. MicroMetals uses the colors but so do
others.

Micrometals were indeed the ones I had in mind.

Graham

 

[Joerg]

The prices you design to never fail to amaze me.

Not always. What I am designing right now is more the "whatever it
takes" kind. But even there one must keep later mass production in mind,
it would not help anyone if I came up with a Princess on the Pea solution.

As for hand-wound stuff the world changes as well. Labor rates in Asia
are picking up.

 

[Paul E. Schoen]

It depends. Sometimes it can be quite sporty to pretend "What if I am on
a lonely island and only my laptop and these few parts?" You can get a
similar circuit going (not with the TIP31 though, and with a forward
concept) using just the sound card and a $2.99 Harborfreight multimeter.

Of course, on that island your laptop battery would croak after two hours
so you've got to be done by then ;-)

But seriously, I was in such situations more than once. Client had a
circuit issue that turned out to be a really nasty noise intrusion from
somewhere. But from where, and what's the path? Next to nothing in
suitable equipment, just my trusty laptop with its 18bit sound card.
However, we needed an iso xfmr that went down to 5Hz and mine quits at
15Hz or so. So we took a mains iso transformer and roached phono jacks
onto it. Of course this just had to be the day the regular OSHA
inspection happened. One engineer turned pale when he saw that. "Put that
away!" ... "Why? We need it right now." ... "Cause the OSHA guy is next
door" ... "Dee-deedle-dum, don't turn around, the commissar's in town,
dee-deedle-dum" ... "Stop that!" (I know the guys quite well, else I
would have pulled that Falco song)

It is useful to pull a MacGuiver once in a while. A few years ago I was on
a troubleshooting job at NSA and I found that a glitch in a logic circuit
was causing it to malfunction, and it was preventing the use of a $50,000
piece of equipment used for critical circuit breaker calibration. I figured
that a small capacitor would fix it, but we didn't bring an assortment of
parts, and their cal lab didn't have that sort of thing either. It would
have been very difficult to go back out through security and go somewhere
to get some parts and then come back in, and we really needed to get this
test set up and running. Luckily, someone had a broken multimeter that we
were able to cannibalize and get a capacitor that fixed the problem. So
they were able to get their testing done, and national security (or at
least my job security) was once again assured.

There should be a techie version of "Survivor" that require the
participants to use electronic junk to cobble together some sort of circuit
that allows them to get out of situations.

Paul

 

[[email protected]]

It is useful to pull a MacGuiver once in a while. A few years ago I was on
a troubleshooting job at NSA and I found that a glitch in a logic circuit
was causing it to malfunction, and it was preventing the use of a $50,000
piece of equipment used for critical circuit breaker calibration. I figured
that a small capacitor would fix it, but we didn't bring an assortment of
parts, and their cal lab didn't have that sort of thing either. It would
have been very difficult to go back out through security and go somewhere
to get some parts and then come back in, and we really needed to get this
test set up and running. Luckily, someone had a broken multimeter that we
were able to cannibalize and get a capacitor that fixed the problem. So
they were able to get their testing done, and national security (or at
least my job security) was once again assured.

There should be a techie version of "Survivor" that require the
participants to use electronic junk to cobble together some sort of circuit
that allows them to get out of situations.

Paul

Wow, that would make a great show!

Michael

 

[[email protected]]

On Aug 29, 4:05 pm, Joerg <[email protected]>
wrote:
....

Excellent tutorial. I'd save many miles on the freeway if everyone would
be familiar with the scope plot in figure six. When you see that shark
fin it's time for duck and cover.

Figure 16 has pretty much what I was trying to do, but better -
brilliant use of the 555 feedback pin! But... No protection diode for
the mosfet, though? And no capacitor to block the DC?

Shark fin... well, at least it guarantees income, right?

A few days ago I saw a book "The Bible for Dummies". At first it shocked
me a bit but hey, whatever it takes.

Anyhow your IRF530 is a lot better than the TIP31 but keep in mind that
whatever is driving it needs to wallop the 700pF gate capacitance around
and it needs to push it up to 10V. And fast.

Can my 555's output pin 3 handle it? Should I use a resistor in
series with the gate, or don't bother...?


Thanks,

Michael

 

[Eeyore]

Wow, that would make a great show!

Michael

It sounds promising. An electronic 'Scrapyard Challenge' kind of.

Graham

 

[Jasen Betts]

Wow, that would make a great show!

sounds like junkyard wars, but with added politics.

Bye.
Jasen

 

[Joerg]

On Aug 29, 4:05 pm, Joerg <[email protected]>
wrote:
...


Figure 16 has pretty much what I was trying to do, but better -
brilliant use of the 555 feedback pin! But... No protection diode for
the mosfet, though? And no capacitor to block the DC?

What diode, and why block DC? Big MOSFETs carry a staunch parasitic body
diode that can often handle around the same current the MOSFET itself
can take. DC flux: That's why you must use a gapped core here. A
non-gapped core will run away within tens of usec ... kablouie.

Shark fin... well, at least it guarantees income, right?

Yeah, except some of the proto boards I get fill the lab with such a
burn stench that I have to keep the windows open even in winter.

Can my 555's output pin 3 handle it? Should I use a resistor in
series with the gate, or don't bother...?

I rarely use gate resistors. But if the trace/wire length is more than
1/2" total including FET pin length you might want to slip a ferrite
bead over it or hang 20-30 ohms in series (right at the FET).

 

[Joerg]

It is useful to pull a MacGuiver once in a while. A few years ago I was on
a troubleshooting job at NSA and I found that a glitch in a logic circuit
was causing it to malfunction, and it was preventing the use of a $50,000
piece of equipment used for critical circuit breaker calibration. I figured
that a small capacitor would fix it, but we didn't bring an assortment of
parts, and their cal lab didn't have that sort of thing either. It would
have been very difficult to go back out through security and go somewhere
to get some parts and then come back in, and we really needed to get this
test set up and running. Luckily, someone had a broken multimeter that we
were able to cannibalize and get a capacitor that fixed the problem. So
they were able to get their testing done, and national security (or at
least my job security) was once again assured.

There should be a techie version of "Survivor" that require the
participants to use electronic junk to cobble together some sort of circuit
that allows them to get out of situations.

I once went around and asked whether someone had an older portable
transistor radio. "What for?" ... "Well, I need the little audio
transformer in there and I'll pay $20" ... "Twenty bucks? Deal! Here!"

BTW, next time that happens ask if there is a RadioShack nearby. One
that hasn't mutated into a cell phone shop like ours did. They have
little Smorgasbord-kits for a few Dollars, containing a nice mix or
ceramic through-hole capacitors. I have been known to do Viking-style
raids on RadioShacks near clients. Once we had a pile of a dozen
clamp-on ferrites and it still wasn't enough. "Oh, I'll just hop over to
RadioShack" ... "Ahm, that's where these came from, I cleaned out all
four stores around here yesterday evening".

 

[[email protected]]

On Aug 31, 10:50 am, Joerg <[email protected]>
wrote:

....

If I build a Boost converter instead of a transformer, does my
inductor still need an air gap?

(figure 3):
http://www.dos4ever.com/flyback/flyback.html

Thanks,

Michael

 

[Joerg]

Yes. It makes no difference whether the boost (flyback) converter has
an isolated output winding or gets the output off the primary winding.
Both versions alternately store input energy in a magnetic field, then
dump that energy to the output. Solid ferrite does not store very much
energy before saturating, but the air gap in the flux path stores lots
more energy per volume than the ferrite core material does.

I rarely use gapped cores, for two reasons. They are almost never
catalog parts and they can cause EMI headaches. In a boost you can be
fine with a non-gapped core as long as you keep the peak current from
saturating the core. This is one of the reasons why my designs are
almost exclusively current-mode. Reduces the kablouie factor quite a bit ;-)