DC transformer

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qonr1t$12n5$1@gioia.aioe.org...

Tim Williams wrote:

https://www.seventransistorlabs.com/Images/PlanarInductor4t.png

This layout is a beauty. Is a single VIA sufficient?

Thanks! Yeah, only a few amperes. Design spec 10W total output. The chip
inductor is 6x6mm if scale isn't apparent. The vias there are pretty big
(1mm?).

The traces and vias under the MOSFET are tiny, 0.15 and 0.3mm respectively.
But it's short so it's not going to burn out.

Why "very much like"? Those are 1mm thick or more, so the 2oz does not
even come close. Or do you mean the entire stackup?

Not like the ones you use, at least? I've seen them made with nearly
paper thin sheet, not sure how they manage to bend the stuff without tearing
or scrunching it, but they do.

And yeah the stackup, hard to follow the colors, but it's about 0.9 turn per
layer, then a via to the next layer and so on. Trace width probably doesn't
need to be so wide -- and Q may well be higher with less, despite the
reduced cross section, because of eddy currents -- I just figured on that
width as a starting point for the overall geometry.

I started by calculating a wire helix of 100nH, then drew a planar version
of similar dimensions -- guessing that the mean current path is not quite
the ID nor the mean diameter, but a bit inbetween, due to current crowding
and skin effect. Ended up with about 90nH, not bad!

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[John Larkin]

On Tue, 22 Oct 2019 22:52:27 +0200, Piotr Wyderski
<peter.pan@neverland.mil> wrote:

John Larkin wrote:

PCBs are like piston engines, very old but still the best way so far.

They are most cost-effective and very repeatable, but "best" in purely
technical terms -- no, I don't think so. The limitations imposed by this
technology are severe.

Kind of the best description of socialism I know of: a system bravely
fighting problems unknown in other systems. Exactly the same with the PCBs.

Best regards, Piotr

PCBs have been replaced by welded cordwood modules, hybrids, flip
chips, SLT modules, multiwire, all sorts of now-gone stuff.

Piston engines (pistons, rings, crank, cam, poppet valves, spark
plugs, clutches, gears, 100 year old tech) have been replaced by
turbines, many sorts of rotary engines, steam, all kinds of now-gone
stuff.

--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Bill Sloman]

On Wednesday, October 23, 2019 at 2:34:54 AM UTC+11, jla...@highlandsniptechnology.com wrote:

On Mon, 21 Oct 2019 19:24:37 -0500, "Tim Williams"
tiwill@seventransistorlabs.com> wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message
news:ea02381f-7c58-420d-88f7-04e712b4922e@googlegroups.com...
Perhaps. I spent half an hour at a trade show some years ago talking to a
printed circuit guy who specialised in the business, and a couple of years
ago we had thread here about a similar sort of firm. They could get the
copper layer thicker than the substrate.

It's certainly going to worth talking to somebody like that.


Wonder if it's more of a potting operation than lamination -- or if the
result simply has voids; give or take what spec you order?

You can of course get more copper density within a given layer, by using
wide traces, but you can't fill in those gaps with regular prepreg of
inferior thickness.

Typical dimensions might be, like...

2oz/layer: ~2.8 mil copper, 7-10 mil prepreg. Fills in fine. Trace
width/space minimum about 7 mils (some 8 or 10, some down to 5 or 4).
Meager winding factor (~15%).

Heavy copper, say 12oz/layer: ~20 mil copper. 5-10 mil prepreg is just
going to sit on top of it. Can't deposit copper that heavy on a flimsy
core, either (so at least one dielectric layer has to be thicker).
Apparently they need closer to 20 mil (as finished?) prepreg to fill in the
gaps (and it's probably high-resin% prepreg at that?). Width/space about 30
mils, so you can't get many turns in a given area (should you happen to need
them). Result, still a meager winding factor (~20%?).

But yeah, if someone can do thin prepreg + fill, and someone else can do
vertical sidewalls (Lintek comes to mind?), the fill rate could get pretty
amazing.

Actually that's even a bit concerning, simply because all the facing area
(edges between adjacent turns), and the sharp corners, are going to make for
wicked eddy currents / skin / proximity effect (whatever, all basically the
same jumble of up-close AC effects).

I know! Bring back that laminated-wire-PCB technology and make printed litz
windings! :^)

Tim

As switcher frequency goes up, skin depth starts to make extra-thick
copper not worthwhile. Very roughly, 2 oz copper is good enough around
1 or 2 MHz sinewave. For switcher waveforms, the tradeoff further
favors not wasting copper.

Exotic PCB material probably doesn't help much. The main cooling is
air at the surface. [1]

I'm guessing that adding more loops on more internal layers doesn't
help. There's basically no cooling, and proximity effect will be
serious. It might help if you need a lot of inductance.

Copper is a pretty good thermal conductor. As John says, the main thermal barrier is the air at the surface, and there's not going to be steep thermal gradient inside the windings.

how about a rising-sun copper pattern on some inner layers, to
conduct heat out?

Printed circuit board substrate material aren't as thermally conductive as copper, but they are still roughly a thousand times better than air.

John Larkin isn't great at quantitative thinking.

--
Bill Sloman, Sydney

 

[Piotr Wyderski]

jlarkin@highlandsniptechnology.com wrote:

It's a soft sheet material,

3GSHIELDING TW-T600-2MM

Thank you, John! Very useful info.

Best regards, Piotr

 

[Piotr Wyderski]

Tim Williams wrote:

> Thanks!

I would happily use something like that, but now this is my integrated
magnetics LLC prototype trafo (~100A):

https://i.postimg.cc/8kdy4Zxq/llc-planar-trafo.png

The traces and vias under the MOSFET are tiny, 0.15 and 0.3mm
respectively. But it's short so it's not going to burn out.

Technically correct, but I don't feel comfortable with that.
I think I am just prejudiced and should work on that.

> Not like the ones you use, at least?

I use what is available and have never seen anything thinner.
In the transition zone between the hair-thick PCBs (2oz) and
that 1mm slabs is some room for laser/water-cut copper plates.
Press-punched in quantity. Downside: you can make just two turns that
way. More require welding, riveting or another origami, neither very
attractive.

But yes, these flat wires are fun. Just like the microwave
waveguides, they have E(asy) and H(ard) modes. At least if it comes to
bending. ;-) Here is another E-mode prototype for 100A high efficiency
bidirectional buck-boost:

https://i.postimg.cc/wvgNjQFm/inductor.png

This is a wide swing one (22uH@0A) to allow deep CCM (<=5% deltaI_L) in
two load points (10% and 100%) that are important to me and still have
decent DCR (1.2 mOhm -- nice but not stellar, OTOH there is a lot of
room for improvement, literally) and compact size (by my standards).

I have so much inductance to spare that I am considering coarse
frequency adjustment in order to minimize all kinds of dynamic losses.
That is, still fixed frequency hard switching SMPS, but fixed for a
given load range.

I've seen them made with nearly paper thin sheet, not sure how they manage to bend the stuff
without tearing or scrunching it, but they do.

Are you sure it wasn't a plate varnished after cutting?

Best regerads, Piotr

 

[Piotr Wyderski]

Tim Williams wrote:

Which by the way, is only about 70% here, rather embarassing for a
resonant converter. :^)

Hm, at 10W why do you even bother? It's a flyback realm or push-pull if
needs to be quiet. LT3999, LT3439 and their inferior brethren. Knowing
your skills there needs to be a good reason for that.

I am happy with my 150kHz DC, have fun boldly going where no man has
gone before!

Best regards, Piotr

 

[Piotr Wyderski]

Tim Williams wrote:

It's too bad there's no high frequency (>1MHz) resonant controllers out
there yet, all that area (the discretes) would be easy to save.

No totem-pole PFC controllers either, fortunately I have already had an
FPGA there.

Best regards, Piotr

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qooq6h$1971$1@gioia.aioe.org...

I would happily use something like that, but now this is my integrated
magnetics LLC prototype trafo (~100A):

https://i.postimg.cc/8kdy4Zxq/llc-planar-trafo.png

Nice. I take it the mirrored winding is bent into a 'C' (side view) around
the litz(?) primary?

Unfortunate that it doesn't use the outer surface to carry current (only the
inner face), but eh, that's how it goes. More interleaving would reduce
necessary leakage (and then you might end up with something like I have,
with additional external inductance).

I did something like that for a recent induction heater build,
https://www.seventransistorlabs.com/Images/Induction1503.jpg
the primary is 10AWG equivalent litz, secondary connects between two wide
copper plates. Three parallel sections of 2t, each positioned over each
section of the primary (which is tapped for tuning purposes).

I cut each winding out of flat copper, which works because of the angle
between turns. Was a bit of a mess to put together, certainly not a
preferred production method, but there it is.

The bigger surprise was managing to solder the poor bastards with just my
hot air machine (turned up to 320°C) and a little help from my 70W iron.
Some fiberglass blankets helped with that...

Each turn connects to the top plate on the outside (you can see two lap
joints in the picture, and the third is just hiding but you can see the
solder fillet). The first turn goes to a flat section hiding under the
core, just above the top plate. The other turn terminates at the inside, in
a hole in the bottom plate. Tabs on the bottom plate make lap joints with
the turns, completing the circuit.

So the top plate looks something like a spanner, and the bottom plate like a
breadboard with a hole in it.

(I've ran 1kW through this so far, no problem; I still need to fix the PFC
module to get up to 3kW, which is all I have available in my current
location.)

I use what is available and have never seen anything thinner.
In the transition zone between the hair-thick PCBs (2oz) and
that 1mm slabs is some room for laser/water-cut copper plates.
Press-punched in quantity. Downside: you can make just two turns that
way. More require welding, riveting or another origami, neither very
attractive.

Huh. Well they're definitely able to bend them the 'H'-mode. Example:
https://www.coilcraft.com/pdfs/ver2923.pdf
Appears to be 0.8mm material across the whole family. In fact, same
resistance, same turns, just varying air gap it looks like... Lazy. :^)

Wurth, Bourns, everyone makes these things. I've got a few of these in my
box:
https://www.digikey.com/product-detail/en/wurth-electronics-inc/7443551221/732-2183-1-ND/2175698
Peeling up the terminal, I find it measures 7.7 mils (~0.2mm) thick (as
tinned). Looks to be exactly the same, uniform material throughout (except
where tinned at the ends, obviously).

But yes, these flat wires are fun. Just like the microwave
waveguides, they have E(asy) and H(ard) modes. At least if it comes to
bending. ;-) Here is another E-mode prototype for 100A high efficiency
bidirectional buck-boost:

https://i.postimg.cc/wvgNjQFm/inductor.png

Nice. Brown-blue? Not sure what core material that is.

I once did these,
https://www.seventransistorlabs.com/Images/Highamp_Chokes.jpg
T300-26D powder cores. So, ~75mm across. Terrific overkill, 25mm wide
copper (cut-up water pipe in fact). Would be good for 200A (give or take
saturation of the core, which is pretty severe on that material at 1000
At+), but the supply was only designed for 50 or 100A, somewhere around
there. (Low voltage, better for charging single cells than welding. A few
in series would make a skookum welder though.)

Was also the first multi-board project I made, the first with PFC, first
with synchronous rectification, first with digital control... It was my
college project, just throw everything in, right?

Are you sure it wasn't a plate varnished after cutting?

Doesn't look like it, but that would be a lot easier on the finish. If not,
they must order it with some pretty stretchy varnish, which really, probably
isn't anything special when you think about how much strain a wire goes
through in a tight bend (kink) or when stretched to breaking. The enamel
doesn't give up, it's very good stuff these days.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qop1dd$ac1$1@gioia.aioe.org...

Hm, at 10W why do you even bother? It's a flyback realm or push-pull if
needs to be quiet. LT3999, LT3439 and their inferior brethren. Knowing
your skills there needs to be a good reason for that.

I am happy with my 150kHz DC, have fun boldly going where no man has gone
before!

Yep, ordinary approach would be flyback. Switching performance would be
quite good with the low leakage (31nH primary referred). It just kinda
stinks for 30V output with so few turns. Controller/regulator again would
have to be pretty fast to succeed on such a small core.

There are regulators this fast, or this powerful, but I don't recall if
there are any that offer both at the same time, i.e., 10W ~2MHz. Not so
many fast controllers either, but I'm also not sure as I was looking into
resonant or half-bridge controllers when I looked into this.

Obvious solution would be, move the transformer to a riser board fabbed with
8 layers (to get enough turns), or just use the next size bigger core (to
get enough V/t).

Kind of walked myself into the trap, but it's just a proto, no big deal.
Good experiment to see if the too-small thing works, and indeed it does.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Piotr Wyderski]

Tim Williams wrote:

Nice. I take it the mirrored winding is bent into a 'C' (side view)
around the litz(?) primary?

The primary is 7x0.3mm triple-insulated litz wire (TEX-ELZ) potted in
transparent epoxy resin for mechanical stability.

The secondary blank should be bent 180 degrees in the middle of the
narrowest strip connecting the two halves. They will be separated by
a layer of kapton foil and varnished.

The fins, when bent by 90 degrees, allow direct soldering of the SR
drains. The horns are for mechanical support (rigidity) and serve no
electric function.

The secondaries will be put on a separate half-core. The yellow
polyester tape provides reasonable level of fine tuning of the gap
(and hence the M factor); it will be replaced by a solid kapton sheet
later. So, no "C" shape, just two semi-independent halves. The secondary
needs to be pretty far away in order to get the required M (8..10).

The bigger surprise was managing to solder the poor bastards with just
my hot air machine (turned up to 320°C) and a little help from my 70W
iron. Some fiberglass blankets helped with that...

Gas-powered soldering irons can do miracles. Soldering a 16mm^2 wire
with Dremel Versatip was a pleasant experience, the previous attempt
with a hotair station was a disaster -- it can be farting all day long
and you will be never sure if the connection is robust enough.

> Nice.  Brown-blue?  Not sure what core material that is.

Khaki. The blue tint is just some unknown reflection. The material is
Hi-Flux type 125. Type 90 would have been better, but they don't produce it.

> The enamel doesn't give up, it's very good stuff these days.

Once upon a time I wanted to remove the enamel from the 7.1x2.5mm wire
to get a meter of a decent busbar. It was just next to impossible, a
true horror story.

Best regards, Piotr

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qoqkcn$1o03$1@gioia.aioe.org...

The secondaries will be put on a separate half-core. The yellow
polyester tape provides reasonable level of fine tuning of the gap
(and hence the M factor); it will be replaced by a solid kapton sheet
later. So, no "C" shape, just two semi-independent halves. The secondary
needs to be pretty far away in order to get the required M (8..10).

Ah. Which stinks even more for skin effect... Is it push-pull (so only one
turn active at a time)? That would save a little eddy current, but you
still have a circulating current across the inner turn when the outer turn
is active.

The bigger surprise was managing to solder the poor bastards with just
my hot air machine (turned up to 320°C) and a little help from my 70W
iron. Some fiberglass blankets helped with that...

Gas-powered soldering irons can do miracles. Soldering a 16mm^2 wire with
Dremel Versatip was a pleasant experience, the previous attempt
with a hotair station was a disaster -- it can be farting all day long
and you will be never sure if the connection is robust enough.

I should probably pick up a plumbing torch. A few hundred watts in a
confined space is awfully useful. At least the metal was completely clean:
the solder wicked perfectly, even though it was but a few degrees above the
melting point.

Khaki. The blue tint is just some unknown reflection. The material is
Hi-Flux type 125. Type 90 would have been better, but they don't produce
it.

Ah, and that'll explain the swing, the higher mu will saturate early and
often.

The enamel doesn't give up, it's very good stuff these days.

Once upon a time I wanted to remove the enamel from the 7.1x2.5mm wire to
get a meter of a decent busbar. It was just next to impossible, a true
horror story.

https://www.youtube.com/watch?v=cDfqwHck2Qg

Which... isn't even as bad an option as it sounds, if you don't mind sanding
off the scale later, or using acid flux.

Occasionally I'll make the mistake of constructing litz cable from old
(alkyd?) literally-enameled wire. Unsolderable, it just turns to glassy
carbon. The only reasonable course of action: take a dirty old tablespoon,
scoop up a little lye (granular drain cleaner), melt it over the stove, and
dip the wire. That shit eats up EVERYTHING. Wash away residues with water
(wear gloves preferably; also, it goes without saying, wear eye protection!)
and there's a nice pink copper surface ready for tinning.

Chemically, the base actually forms a salt with copper, solubilizing it.
Deep blue color, you'll see it in the melt and transiently when washing up.
(Lye doesn't work as a flux, I think it actively corrodes tin.) Different
from ammonia, which forms a complex with copper (which is much darker).

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Piotr Wyderski]

Tim Williams wrote:

Ah.  Which stinks even more for skin effect...  Is it push-pull (so only
one turn active at a time)?

Yes, an LLC with a two-MOSFET synchronous rectifier. The narrow part is
the positive rail, the MOSFETs' sources provide the negative one.

  That would save a little eddy current, but
you still have a circulating current across the inner turn when the
outer turn is active.

I'd say it is an inherent property of a planar winding. What can you do?

Ah, and that'll explain the swing, the higher mu will saturate early and
often.

Yes, but the saturation is gradual and predictable. It follows the
output current prety well, keeping the differential inductance
reasonably close to an optimal value for a given current (for a given
deltaI_L and loop dynamics).

> Which... isn't even as bad an option as it sounds

Would work, but cancer included. ;-)
I resorted to abrasive methods.

Best regards, Piotr

 

[Mikko OH2HVJ]

"Tim Williams" <tiwill@seventransistorlabs.com> writes:

Occasionally I'll make the mistake of constructing litz cable from old
(alkyd?) literally-enameled wire. Unsolderable, it just turns to
glassy carbon. The only reasonable course of action: take a dirty old
tablespoon, scoop up a little lye (granular drain cleaner), melt it
over the stove, and dip the wire. That shit eats up EVERYTHING. Wash
....

Yep, the burnt spoon with white residue is well described in the parents
guide to eletronics paraphernalia

--
mikko

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qosqa4$1mcd$1@gioia.aioe.org...

That would save a little eddy current, but you still have a circulating
current across the inner turn when the outer turn is active.

I'd say it is an inherent property of a planar winding. What can you do?

Well, you could get away with a single turn, using a full bridge rect. Also
a PITA though.

Hm, leakage between the two halves of the winding shouldn't be a big deal,
right? Your build is quite low leakage to begin with, which is nice (though
its impedance is not fractional-ohms low, so it will still manifest as
leakage inductance), but in a resonant topology, the current can be small or
even crossing zero during commutation?

Yes, but the saturation is gradual and predictable. It follows the output
current prety well, keeping the differential inductance reasonably close
to an optimal value for a given current (for a given deltaI_L and loop
dynamics).

Yup, powder cores saturate nice and gradually.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Piotr Wyderski]

Tim Williams wrote:

Well, you could get away with a single turn, using a full bridge rect.
Also a PITA though.

I considered that, but at this current level a milliohm is ~10W of waste
heat. Eventually I would like to use the PSMNR70-40SSH for the SR, but
that would add 620uOhms to the critical path. Then are the dynamic
losses. And it is so much simpler to cool a massive copper slab than an
8x8mm MOSFET... So the simpler solution has won.

Hm, leakage between the two halves of the winding shouldn't be a big
deal, right?

I think so, but I don't know how to measure it with reasonable accuracy.
The magnetizing inductance of the primary is 420uH and the leakage
inductance
is 41H (with shorted secondaries), which gives a nice factor of 10.2
and no external resonant inductor should be necessary. But
secondary-to-secondary inductance -- no idea.

Your build is quite low leakage to begin with, which is
nice (though its impedance is not fractional-ohms low, so it will still
manifest as leakage inductance)

The DCR of the primary is ~160mOhm and fr=131kHz.

but in a resonant topology, the current
can be small or even crossing zero during commutation?

Yes, ZCS is my goal.

> Yup, powder cores saturate nice and gradually.

But losses are absurd. Hence so deep CCM -- it is basically a DC-only
choke with some AC noise. ;-)

Best regards, Piotr

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qoug11$14i8$1@gioia.aioe.org...

I think so, but I don't know how to measure it with reasonable accuracy.
The magnetizing inductance of the primary is 420uH and the leakage
inductance
is 41H (with shorted secondaries), which gives a nice factor of 10.2
and no external resonant inductor should be necessary. But
secondary-to-secondary inductance -- no idea.

Can estimate it, at least -- the parallel plate formula applies. So, just
the dimensions of the space between plates. Which is what, 10mm wide, 100mm
long, 0.1mm space, so... a few nH? The lead lengths will dominate (where
there's no overlap).

Yup, powder cores saturate nice and gradually.

But losses are absurd. Hence so deep CCM -- it is basically a DC-only
choke with some AC noise. ;-)

Is that cap-input or choke-input? Seems like I see cap-input most often.
Oh, that probably wouldn't work well with sync rect, would it?

(Cap input probably still needs an inductor for clean output, so I'm not
ruling that out, but it wouldn't need so much inductance either..)

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[Piotr Wyderski]

Tim Williams wrote:

Is that cap-input or choke-input?  Seems like I see cap-input most
often. Oh, that probably wouldn't work well with sync rect, would it?

What's capacitor/inductor input? Input of what? I'm afraid I don't know
this terminology.

The primary would be driven by an H bridge and 2x18nF "divider",
so this would be the input.

Best regards, Piotr

 

[Tim Williams]

"Piotr Wyderski" <peter.pan@neverland.mil> wrote in message
news:qp00r5$52e$1@gioia.aioe.org...

Tim Williams wrote:

Is that cap-input or choke-input? Seems like I see cap-input most often.
Oh, that probably wouldn't work well with sync rect, would it?

What's capacitor/inductor input? Input of what? I'm afraid I don't know
this terminology.

Sorry, cap-input rectifier filter. So, secondary, to diodes/transistors, to
cap/choke, etc.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/