inductor problem

I have about half a square inch of PCB surface to fit this inductor into. I could tolerate maybe half an inch of height.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

I have some Micro-metals powdered iron toroids coming, type 6 and 7 material. I do fear I may burn the paint off from core loss... I've done that before in switching regulators.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Someone clearly hadn't processed what 6A means.

An RM6 core fits into a 14.6mm square. Half an inch is 12.7mm. At least the core is only 12.5mm high. For three turns of nine (2.906mm OD) or ten gauge (2.588mm OD)copper wire you wouldn't need a former.

Sadly, 6A peak through three turns on an RM6 core is 600 ampere turns per metre.

The TDK/EPCOS B65807+0040A001 core does have an 0.8mm gap but that probably isn't enough. More cores or a bigger core might work.

The board does seem to need to be re-worked - as if that hadn't become obvious already.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Easy, just extract mu'' from the datasheet. They even give a differentiable expression!

formatting link

Might not actually be too terrible at 30MHz, but I suspect that's more in the sense that it'll be more inductive than resistive, but not enough to be worthwhile...

Joerg has exploded #2's (very similar material) at much lower frequencies and modestly higher power levels. Probably >10x more power, but you're looking at a >10x smaller core to fit the space, I think, so that doesn't bode well.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

This will be better than 2 material, but baluns and tubes offer a better geometry if copper loss is the concern. (two tubes ~= balun) A 1728 will give L values in your range with a single turn. Distribute losses as before in series/parallel.

You say 6Arms with a pulsed waveform - are you sure you're calculating rms accurately, with the high crest factor? What sort of voltsecond and rep rate?

Small efficient parts can still get hot - is this really a problem in non-isolated magnetics in a non-power-conversion lab application?

RL

Reply to
legg

They are one of the best companies aroud, also WRT to component data.

Yes, but it seems that John is up against the skin effect and there is not much that can be done about that other than better coating and more copper surface. Or live with the losses and sink the heat away to somewhere, as long as no lacquers or anything develops a scent.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

A spiral PCB trace might be interesting. If I put it on the bottom of the board, with no solder mask, I could seriously heat sink it to my baseplate. The magnetics get too complex for me to simulate, but it wouldn't be hard to prototype.

PCB traces are pretty on the outside but grungy (black copper treatment for adhesion) so there would be a lot of skin loss on one side of each trace.

The nickel layer in an ENIG trace might get interesting too. So maybe I should stay with the solder mask.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Am 08.04.2018 um 20:17 schrieb John Larkin:

We have tried the Piconics chokes. Using them makes it easy to build a bias tee that is absolutely flat to 20 GHz. But their prices were ridiculous as long as their patent lasted. For production, we had to do without them. IIRC Mini Circuits now has sth. similar, also. But not suitable for your power level.

regards, Gerhard

Reply to
Gerhard Hoffmann

Maybe. Seems likely it'll be much worse, because you just can't get as much metal in play. Even with a blob of thermally conductive epoxy to an AlN insulator, you might have overheating problems.

Current flows on the edges of traces, to a depth determined by the sheet resistance of the conductor. At least until frequencies are so high that regular skin depth is below the foil thickness, and the middle doesn't carry any current anymore.

Note you need the top and bottom free of metal, a hole in the ground plane about >= 1.5 times the spiral O.D. On the upside, you can put a spiral on every layer, and swap them around, in parallel and series, to equalize path lengths -- printed Litz as it were. It might not be as good as real Litz, but it's what you can do.

It will take up many times more footprint than you have available right now.

Solder mask isn't a problem until GHz RF. On the other hand, you could paste it, which gets rid of the gold (probably not the nickel) and adds more meat on the surface.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Am 08.04.2018 um 21:09 schrieb Tim Williams:

Also for uwave boards, Au over Ni is not an option.

Cheers, Gerhard

Reply to
Gerhard Hoffmann

Coilcraft makes them too. They also have a cool higher-current part,

formatting link

Bias tees don't mind increasing skin resistance loss, as long as the RF power level is low.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

I did an FFT on the simulated current, and it turns out that most of the current will be below 16 MHz, so I may be able to do it with a few of the 1010VS parts, well heat sunk.

formatting link

Coilcraft's data and models exclude any ground plane coupling, so I'll have to experiment with that. Thermals, too.

I never liked RF, or talking to people, so I was never a ham. I still don't like RF.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Here's an FFT of the simulated current waveform. I'm making fast voltage pulses, but the inductor current is mostly harmonics of the pulse rate, which may save the situation.

formatting link

This is, optimistically, not a lab application.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

That doesn't look so bad but the skin effect will still be dominant.

Coupling isn't so much a cause for heat in the coil but in the ground plane or in whatever the coil couples to. It also spoils the RF performance of an inductor. Best to carve out a void in the planes below. And never, ever run any sensitive traces through that "demilitarized zone" or close to it.

I grew up with RF and like it way more than anything digital. Some stuff is cool and really impresses people. Like when a few bucks worth of RF parts make a TDR that has really fine resolution. They look at the schematic, "You mean, that's ALL?".

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

I am talking to walls probably but induction heater coils do not get hot. They use large coils with only a few turns. And thick wire.

formatting link
A coil with a diameter of 40mm, 3 turns, and 30mm long gives 0.337uH
formatting link

Reply to
LM

Is there a dc component?

RL

Reply to
legg

The FFT doesn't express phase relationships, so the current waveform can't be reconstructed with an accurate peaked form, even with a dc shift added.

Could you just illustrate the waveforms?

A 400V 20nSec voltage pulse would get a flux peak deviation of almost

1/2T in the fingernail-sized balun. Hollow rod pairs of similar accumulated dimensions would be more conservative.

If you did that at 4MHz, the 300mW loss (book values in 6matl or lower permeability dust) would raise the temperature of the part. Series arrangements could spread this out to reduce surface temperatures.

A solid .06in dia wire should manage the current and fit the holes provided.

RL

Reply to
legg

The coil he's got, with two layers, close wound, probably has a Q around at

1MHz. The inner layer serves more as _shielding_ against the outer layer. The eddy currents are disgusting!

Even a single layer has a lot of eddy currents (self induction, proximity effect, however you like to call it). Which is why, when you look at a plot of current versus wire cross section, across a solenoid like this, it's strongly concentrated towards the inside face of the turns, sometimes to the point where it's flowing in the opposite direction on the outside face.

Example at 1MHz:

formatting link
That was a chunk of #26 powdered iron, which while it's quite a lossy material (mostly a resistor at this frequency), it actually didn't heat all that well, until it got red hot and started to sinter.

The coil is 12 AWG enameled, and the middle turns got quite hot just on account of being there -- not simply because there was a red-hot lump beside them.

And this was only about 100W.

Hamwaves calculator, by the way -- the best inductor calculator I know of, period! Includes all effects, analytical and empirical, with better accuracy than you'll ever measure a coil at.

Looks like the best John could do might be... D = 14mm N = 7 l = 18mm d = 2mm f = 30MHz

Leff = 0.33uH Qeff = 395

Apparently he's doing over 300VAR -- which would've been an easy number to work with, if he had put it in the OP, rather than being characteristically cagey about it -- which suggests a Q under 300 will dissipate over 1W, which seems borderline inconvenient at this size.

A smaller inductor (say that fits in a 10mm cube) will only dissipate more, for a double-whammy on temperature. The best, then, might be to wind a spring coil on a square former, and glue that to an AlN plate, for heatsinking without impairing the field around the coil. That might stand 5 or 10W, so it could be small enough to have a Q of only 50. But that would be a /lot/ of effort for something so easily solved with a board spin.

It's hard to estimate the effect of powdered iron or ferrite at this frequency; no one gives curves at frequencies and power levels this high. Some give Q factors, but they aren't very competitive against air-core coils, except in terms of stray fields (an inductor might physically fit within a 10mm cube, but its fields need a 20mm or larger cube).

Speaking of which, one can put a solenoid in a ferrite box, about doubling its inductance while shunting external fields nicely. Basically a pot core with no center leg. That's a good way to get a smaller outline, without costing too much on losses (maximum flux density in the ferrite shouldn't be too high, whereas a rod core might be lossier).

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

The complex FFT is invertible, and does indeed preserve phase.

Cheers

Phil Hobbs

Reply to
pcdhobbs

Yes, but John showed only the magnitude.

Jeroen Belleman

Reply to
Jeroen Belleman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.