Instead scopes

Based on your understanding of the data sheet, which was obviously inadequate. It's revealing that you don't post links to the data sheets or specify the number that you relied on when you assumed that they ought to have worked.

Solder is metal, and has a higher conductivity than your gap-pad material. You can over-fill the joint, which would help.

No. It's merely an obvious possibility.

What makes you think that? The fact that the part is sintered doesn't mean that you won't get close to solid metal electrical conductivity.

And even you must concede that that wouldn't have been lossy.

Heat pipes don't need that. A closed system doesn't need a water-tank, and lots of top-end computer coolers do rely on circulating water.

Imagined massive losses.

It's hand-wound, so it looks cheaper than it is.

Or come up with a more sensible solution?

If the FR4 losses matter, the printed circuit board under the coil would darken, which the customers wouldn't like. The Cambridge Instruments

It's a sawtooth so it has quite a lot of higher harmonic components with even thinner skin depths. Baxandall's preference for sine waves has incidental advantages.

The turns are wide and flat, which reduces the effect of skin dept.

That 25A rms isn't going to include any allowance for skin effect.

They don't look as if there would be much cross-talk from one to the next. Making space for more parts might have been a better approach.

With +/-20% tolerance on inductance, putting them in parallel wouldn't have been a good idea.

Particularly when you don't think about what you doing.

So I can end up as ill-informed as you are?

Somebody who believes what he reads in Russia Today has already been brainwashed, and you are complaining that I haven't been suckered by the mis-information that you have swallowed, hook, line and sinker.

Though not even they are silly enough to have gone for the Le Sage theory of gravity.

Or choose a 12mm OD mandrel, and adjust elsewhere. The advantage of

.

Actually, the requirement is a certain inductance while handling a

The dimensions et al are the construction details needed for Highland to be able to replicate the part without your help.

But useful. Or essential in some cases.

Joe Gwinn

Lo these forty year gone, I had this RF gig that involved making a lot of VHF LC oscillatior and filter protos.

We had a hand-cranked coil winder that had a good selection of cylindrical steel mandrels with helical grooves to guide the wire, plus three or four sheets with tables of measured values for single-layer coils of various lengths. With a couple of training runs, one learned how hard to pull on the wire so that it would just spring free from the mandrel.

That made it easy to make nice looking, high-Q coils for the inductance range of interest. Good Medicine.

Cheers

Phil Hobbs

It wouldn't help much to conduct heat into a PCB pad. FR4 is a terrible heat conductor.

Ha.

Skin depth is about 30 microns here, and we need a smooth, homogenous, annealed surface. Ask a chemist.

Impractical, and cast copper is probably a worse conductor than annealed.

My gadget is cheap and easy and works.

Certainly imagined. Please make a 3D fabbed inductor and measure its Q and report back to us.

This Pockels Cell driver is maybe 1/20 the volume of competitors' and uses a few per cent of the power. The inductor is a detail.

Most drivers dissipate

F * C * V^2

in the driver itself, but it should take zero energy to charge and discharge a capacitor.

More sensible than winding an inductor from magnet wire?

That's another function of the gap-pad material. And the flattish part of the circular solenoid windings make a big contact area into the pads, for more heat transfer.

It would be cool is the bottom of the windings were actually flat, a square or elliptical winding, for more heat transfer area.

Don't they have the same tolerance in series?

In parallel, each would get 1/3 the current. But each would need to be

My coil opens itself up for a lot of air cooling, and bare copper can run pretty hot.

But it works. A big laser company buys them.

Why don't you design a 1200 volt, 4 MHz pulse generator and we can discuss it here.

That's past my pay grade. I wound coils until it worked. The cooling issues complicate things.

How about Litz wire wound on a spiral-grooved aluminum nitride tube, with air blown inside maybe?

We are lucky to have resistors and capacitors and inductors, linear and derivative and integral. There is no thermal equivalent of an inductor, which is why thermal systems are dynamically sloppy.

All good. Needs to be in the drawing.

.

This is what's easily available. If you make your own mandrels, they can be any shape. Make sure that at least one end has a index slot or radial hole, to make winding easier by allowing the clock angle of the mandrel easily fixed.

Joe Gwinn

I'd compute it. The fundamental may not be the whole story.

That could certainly work, but would be far more expensive that the plain coil currently used. I would conformal coat the assembly for thermal contact between wire and AlN tube. Best thermal story would be silicon-rubber conformal coating.

For one use case, using a toroid with both supply leads (ort and back) threaded through the aperture allows one to measure the common current and the differential current (reverse one wire) with essentially infinite CMRR (needs shields), over a very wide dynamic range.

Yes.

Joe Gwinn

I still design LC oscillators!

Coilcraft makes a bunch of bare-naked RF inductors.

We like this encapsulated part:

Sounds awfully high.

To about 1% accuracy (over most useful aspect ratios), the inductance of a single-layer coil is

L(uH) = a**2 n**2 /(9a + 10b),

Where a is the mean radius and b is the overall length, both in inches.

If the thermal expansion is unconstrained, a and b vary together, so the TC of inductance of such a coil is the same as the CTE of copper, about 17 ppm/K.

A solid plastic form is strong enough to stretch the copper and increase the TCL to the CTE of the plastic.

If the length only is constant, the TCL is increased, and for a short fat coil it’s nearly doubled.

In the case of something like B&W Miniductor, which has fairly fine-pitched turns held by a few small axial stringers, the plastic stretches the length of the coil but leaves the radial expansion free.

Because both a and b enter in the denominator, it’s possible to choose dimensions that make the TCL of a Miniductor or similar coil very nearly zero.

A normal single layer coil has a TCL around +30 ppm. I’ve never measured one as bad as +120.

Cheers

Phil Hobbs

That's where your gap-ad dumps the heat. It may be a terrible conductor (if a lot better than air) but it was clearly good enough for you.

So not obvious to you.

Ask a guy with a pulsed laser, who can melt the top 30 microns of the part for enough milliseconds for it flow into a smooth surface. I've always been puzzled why people haven't resumed silver plating RF parts when you can now fuse the surface electro-plated crystals into a continuous film of silver

Why?

and cast copper is probably a worse conductor than annealed.

So anneal it.

But looks cheap. And a hand-wound coil is never easy.

If you'll pay for it, and the time I'd have to spend to dig out somebody to do it. Back when I was working I had access to people who would know somebody who would do it, but I moved back to Australia, and those people are less accessible (and some of them have died).

But a crucial one.

Sadly. capacitors can be lossy, and the current involved has flow through conductors which are resistive - more so when the skin effect kicks in.

Why do you think that the Coilcraft inductors were wound with flat wire?

The skin effect means that most of your magnet wire is unused.

With flat wire, you also have more surface area to couple the wire to the air you are relying on to cool it.

From a practical point of view, changing the layout to accomodate enough of the Coilcraft parts to dissipate the heat would give you a much tidier-looking product, if marginally bulkier.

If what you've got has 5% of the volume of its competition, you've got room to do that.

When they are in series, the same current passes through each one.

When they are in parallel the lower inductance parts would carry up to

And melt the solder holding it in place?

Why would I bother? I wouldn't get paid for it. More to the point, I'd need a lot more information than 1200V, 4 MHz and 25A peak to peak, and your customer wouldn't be happy if you broadcast that much detail.

You may put them together, but it sounds as if you evolve them rather than design them. And you'd have your own coil-winding gear if you did much of it. As Phil did.

At George Kent in Luton (1973-76) I got to wind my own small-signal transformers. At Cambridge Instruments (1982-1991) I had to ask the coil-winders on the shop floor to do it for me.

I posted a followup to my tc comment, but sometimes usenet posts disappear.

Right, the L of a loop goes about as d-squared, so the +17 PPM TCE of copper should make L increase at maybe 35 PPM.

We have measured much higher tempcos of small inductors used in our triggered LC oscillators. That may be a secondary effect of encapsulation or something. I was pleased that the Coilcraft Midi parts were about +40 and apparently very stable.

We have used some surface-mount coils that not only had awful tc's, but their L would creep over months as the potting stuff apparently flowed. Annealing helped.

Inductors are a pain.

Design, simulate, build, test, evolve. That's how engineering usually works. At the bleeding edge of performance, unpredictable higher-order effects happen. Sometimes whacking the competition depends on understanding and taming those effects. That's more fun to me than pushing a bunch of equations around.

My Sharpie is a nice red marker when it's not winding coils.

I used to have toroid winding machine. That's not actually a rational thing to do.

Serious engineering goes more like design, simulate, redesign,simulate again, build, test, modify-evolve, retest, design, simulate, build, test, ship.

As if you were ever there.

But if you don't push the equations around you don't get to understand the effects properly, no matter how proud you are of your "intuition".

But it is still a ppor excuse for a coilwinder.

It's entirely rational if you need special purpose toroids - perhaps non-progressively wound (so they really do have zero external field).

It's a mechanically complicated device - not that I've ever seen one - and presumably a pest to keep working.