small inverter driver chip

[Terry Given]

Half-bridge on the drive side? That's not a flyback.

John

diangonal half bridge, so a switch at either end of the winding.
250,000+ of them suggest they are flybacks.

the open-loop regulation comes from the flyback mode, when both primary
switches turn off. the pair of primary diodes then clamp the leakage
spike to the input rail, and if more energy is stored in the core than
the load needs, the excess energy is returned to the primary supply by
said diodes.

Cheers
Terry

 

[Spehro Pefhany]

diangonal half bridge, so a switch at either end of the winding.
250,000+ of them suggest they are flybacks.

the open-loop regulation comes from the flyback mode, when both primary
switches turn off. the pair of primary diodes then clamp the leakage
spike to the input rail, and if more energy is stored in the core than
the load needs, the excess energy is returned to the primary supply by
said diodes.

Cheers
Terry

Whether it's a flyback or forward converter depends on the way the
output diode is connected, no?



Best regards,
Spehro Pefhany

 

[John Larkin]

What could you do with these?
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7598144809

That's sort of a commodity part... lots of them in the Mouser and
Digikey catalogs. I used one in a 900 volt hv supply I posted some
time back...


================
+9-------///////ct///////--------to c-w multiplier chain
|
|
|
|
d
-----g 2N7002
s
|
|
|
|
gnd


The 2:1 stepup greatly improves the available flyback boost ratio.

The gate drive was tiny-logic schmitt oscillator. One version had
closed-loop hv regulation with one additional transistor.

John

 

[John Larkin]

Whether it's a flyback or forward converter depends on the way the
output diode is connected, no?

I guess it's a flyback if all the energy is stored in the inductor
before being transferred into the load.

John

 

[John Larkin]

but 30mH is a bit high to do much of anything with for a flyback, unless
the voltage is very high and the frequency is very low.

OK, let me try to restate my perspective one more time:

I AM NOT USING THESE TRANSFORMERS AS FLYBACKS. OTHER TOPOLOGIES DO
EXIST.

I'd use that puppy as a crude dc-dc, but slap a PCMC loop around it
(which is essentially free) to ensure startup is nicely controlled. Or
use naff all output C, such that dTj is acceptable for your switching
transistor(s).

The simple square-wave inverter needs no feedback loop and has no
startup problems. Since inductance is so high, these parts could be
run in the audio range, although 50-70 KHz seems about right. We often
deliberately slow down the fet gate drives to reduce noise, which
doesn't work as well in a flyback.

Custom magnetics don't make a lot of sense unless volume is huge, as
yours seems to be. The time and hassle spent in engineering is hard to
earn back over less than tens of thousands of parts, so we use a stock
transformer or inductor whenever we can.

John

 

[Terry Given]

dui4


I guess it's a flyback if all the energy is stored in the inductor
before being transferred into the load.

John

hen3 dui4

(I taught myself Mandarin recently)

Cheers
Terry

 

[Terry Given]

OK, let me try to restate my perspective one more time:

I AM NOT USING THESE TRANSFORMERS AS FLYBACKS. OTHER TOPOLOGIES DO
EXIST.

Dont beat about the bush John, tell us whether or not you intend using
these as flyback transformers. dont be shy

The simple square-wave inverter needs no feedback loop and has no
startup problems. Since inductance is so high, these parts could be
run in the audio range, although 50-70 KHz seems about right. We often
deliberately slow down the fet gate drives to reduce noise, which
doesn't work as well in a flyback.

wrt startup: depends on output capacitance, no? I guess there is a fair
amount of leakage, so it looks less like a short.

Custom magnetics don't make a lot of sense unless volume is huge, as
yours seems to be. The time and hassle spent in engineering is hard to
earn back over less than tens of thousands of parts, so we use a stock
transformer or inductor whenever we can.

John

Absolutely! If cheap is good, cheaper must be better

I often use planar cores, for that reason - standard cores, DIY custom
windings.

In this particular case its not all about volume (but I am using about 8
of these per design), but rather about what happens when things go
horribly wrong. an inductorless buck-derived converter isnt so happy
running into a dead short.

The last lot I did were exactly as you described, and they worked well.
I had a PLD, so it generated the oscillator, which I could control the
duty cycle (narrow pulses at startup), and no feedback. It worked very
well, BUT when I deliberately blew up an IGBT (10-20 desats in quick
succession) it snotted the gatedrive completely (look ma, no zetex
transistors) and buggered the supply too - tiny bjt didnt like switching
into what was essentially a dead short.

So I figured next time round, I'll specifically consider what happens
then. this new design is also cheaper

Cheers
Terry

 

[John Larkin]

hen3

Hey, chickens are off-topic in my thread.

John

 

[Terry Given]

Hey, chickens are off-topic in my thread.

John

not even chicken chow-mein?

Cheers
Terry

 

[John Larkin]

not even chicken chow-mein?

Empty calories. But we had some VietNamese BBQ chicken on Sunday that
was memorable.

John

 

[John Larkin]

Dont beat about the bush John, tell us whether or not you intend using
these as flyback transformers. dont be shy

Don't rush me. I need a chance to think it over.

wrt startup: depends on output capacitance, no? I guess there is a fair
amount of leakage, so it looks less like a short.

Who cares? Just let the 2N7002's grunt. But right, a rectified square
wave doesn't need a lot of filter caps like a flyback does. And none
of that silly right-half-plane nonsense for us, no siree.

Absolutely! If cheap is good, cheaper must be better

I often use planar cores, for that reason - standard cores, DIY custom
windings.

In this particular case its not all about volume (but I am using about 8
of these per design), but rather about what happens when things go
horribly wrong. an inductorless buck-derived converter isnt so happy
running into a dead short.

We do worry about a shorted channel on the 16-channel boards. One fix
is a small source resistor shared between the fets, with appropriately
scaled gate drive (~~4 volts works) so the fets will current-limit if
the load is shorted.

The last lot I did were exactly as you described, and they worked well.
I had a PLD, so it generated the oscillator, which I could control the
duty cycle (narrow pulses at startup), and no feedback. It worked very
well, BUT when I deliberately blew up an IGBT (10-20 desats in quick
succession) it snotted the gatedrive completely (look ma, no zetex
transistors) and buggered the supply too - tiny bjt didnt like switching
into what was essentially a dead short.

So I figured next time round, I'll specifically consider what happens
then. this new design is also cheaper

Power electronics is a nuisance when things go wrong.

John

 

[Terry Given]

Empty calories. But we had some VietNamese BBQ chicken on Sunday that
was memorable.

John

Gamun

Cheers
Terry

 

[Terry Given]

Don't rush me. I need a chance to think it over.






Who cares? Just let the 2N7002's grunt. But right, a rectified square
wave doesn't need a lot of filter caps like a flyback does. And none
of that silly right-half-plane nonsense for us, no siree.

fair enough, too. Something I learned a long time ago: Aint no laws
against slavery in electronics. Yay for m*cp

We do worry about a shorted channel on the 16-channel boards. One fix
is a small source resistor shared between the fets, with appropriately
scaled gate drive (~~4 volts works) so the fets will current-limit if
the load is shorted.

good thinking. Nowadays I throw in one more transistor, and run peak
current mode control. I also monitor average gate voltage, so I can shut
the whole thing off if it sits at low duty cycles. The PCMC protects the
junction for a long enough time that its easy to notice and do something.

Power electronics is a nuisance when things go wrong.

John

Yes. "Component no longer there" is a common failure mode.

Its made me rather careful. I once sat beside a flywheel that
self-destructed with 1MJ of stored energy. came close to shitting my
pants, but luckily the containment all worked well (the kind of "luck"
that involves careful worst-case-scenario design). It certainly got the
attention of everyone in the factory

and whats the deadliest near-miss I've seen? A guy nearly lose his head
in a wicked storm, when a sheet of roofing iron flew over his shoulder -
he was running, and the iron went by a LOT faster. 2' to the right and
he'd have been bisected



Cheers
Terry

 

[Spehro Pefhany]

hen3 dui4

(I taught myself Mandarin recently)

Cheers
Terry

Très bon... err.. hao3 ji2 le!


Best regards,
Spehro Pefhany

 

[Terry Given]

Très bon... err.. hao3 ji2 le!


Best regards,
Spehro Pefhany

You are far, far too kind. I have several friends who are fluent in
mandarin, and I recently spent 3 weeks there on a consulting job. Before
I went there, I could say "hello", "thankyou" and "goodbye", albeit so
badly I'm sure it was only thru context that anyone understood me (sigh
chair?!)

My good friend Paul spent my first weekend there teaching me, and making
me do all the speaking. After the first week, I could get around - eat,
sleep, travel, shop, swear at beggars (isnt PK a great insult? OK its
cantonese), all the fun stuff. Dice games taught me the numbers

Now I have a dictionary, and a Berlitz phrase book & CD, which I've
ripped to the MP3 player I bought for this purpose (long flights are
boring). I listen to the CD every time I'm driving.

The mandarin phrase book is terrible, as it doesnt use pinyin, but
rather an american-phoneticised mutation. This has actually been very
helpful, as I am slowly translating that crap into pinyin.

There are lots of chinese here in NZ (12.5% or so), and I quite often
get to conduct entire transactions in Putonghua. It surprises the hell
out of most chinese, that a white guy would bother to learn such a thing.

It will be a while before I can have a real conversation, but I'll get
there. the writing isnt exactly easy either (yi, ar, san are OK, the
rest is harder), but the language structure is fun.

I also write emails in pinyin, after first apologising (qing wen wo de
putonghua bu hao), as I need the practice.


Zai4Jian4
Terry

 

[Spehro Pefhany]

On Tue, 13 Jun 2006 15:32:29 +1200, the renowned Terry Given

I also write emails in pinyin, after first apologising (qing wen wo de
putonghua bu hao), as I need the practice.

Should be: dui4 bu qi3 wo3 de ... ;-)

Zai4Jian4
Terry

Ditto. I'm working more on the characters these days. A bit of a hard
slog, but it's coming.


Best regards,
Spehro Pefhany

 

[Terry Given]

On Tue, 13 Jun 2006 15:32:29 +1200, the renowned Terry Given




Should be: dui4 bu qi3 wo3 de ... ;-)

Ah, I see, Xiexie. I have no shame, and am happy to experiment and be
corrected. Qing Wen is essentially "please ask" so should preface a
question.

Ditto. I'm working more on the characters these days. A bit of a hard
slog, but it's coming.


Best regards,
Spehro Pefhany

One of my mates can read 10,000 characters, which is no mean feat.
Another can read and write chinese emails. Me, Im just happy I can
recognize ge

I cant believe I waited so long to learn another language. Still, there
is no time like the present.

Back to writing 8051 machine code....hilarious, I'm using a hex editor,
and hand-compiling. I've been arguing with a programmer about some of
his crappy code, and so far am winning, by dint of hacking the bin file
directly. I sorted out the inability to generate a 76kHz square wave
problem (moral: never let computer programmers design anything; get an
engineer), now I am figuring out why the LEDs flash at power-on.

Cheers