Instead scopes

I can't say I've seen that. I do have an advanced degree, but I can invent things and, and I've hung out with people with rather more patents than I can claim, some of them with equally advanced degrees.

Again, not something I've seen. Some of the people who think they have invented something useful get attached to some rather bad ideas, and resent informed criticism.

I worked at EMI Central Research where you were encouraged to patent stuff. The engineers were much more relaxed about it than others I'd worked with in other places.

My father got most of his 25-odd patents while I growing up, so I got to hear about the interactions in his work-place. It all seemed to run pretty smoothly.

There's book that covers the period -"The Pulp" ISBN 978-0-9870915-5-0 - which records some resentful observations from some people well down the pecking order, but the guy that wrote the book didn't join the firm until after my father had moved on, and didn't know all that much about the period.

It is a subject that I do know quite a bit about. You've got your name on just one patent that was taken out by a group you worked with. You may be less well-informed than you like to think.

I’ve done a lot of lowish-power RF stuff, and mostly agree with you about the practicality of using S parameters in hand calculations.

However, I cordially disagree with your sentiments regarding Smith charts.

For one thing, they’re super useful for designing optical coatings, but that’s a minority interest on SED.

In RF work one runs into a lot of matching jobs involving modulated sine waves.

One typical example from my work is coupling sine modulation into a diode laser, for modulation-generated carrier interferometry. (*) A Smith chart makes it super easy to try out different schemes, such as series/shunt stubs, lumped elements, or any combination thereof.

Useless for bandwidths of an octave or more, and so apt to be undervalued by crass time-domain types. ;)

Cheers

Phil Hobbs

(*)MGC is dear to my heart—it’s about the only thing I’ve ever invented that was published by someone else first. (There are an amazing number of things to work on.)

You put sinusoidal FM on the laser, and adjust the amplitude to the first carrier null (M=2.405 radians) at the interferometer output. At that point the first (Q phase) and second (I phase) signals are the same size, and account for about 85% of the total signal power.

That lets you measure the complex signal just by looking at the amplitudes of the first two harmonics. Great for fiber sensors and other situations where there’s a lot of low frequency instability.

The interesting parts of the world are wideband and nonlinear. So are we.

Sine waves are BORING.

One thermistor is on the heater board, on the bottom of the big block. Three are on the platform that mounts the e/o modulator. We really don't need three up there, but we wanted to error check and snoop for gradients and optionally do some averaging if we had noise.

The EOM platform is spaced off the bottom of the big block, which makes us a 2nd order thermal system. The main block has a 75 minute time constant, and the platform inside is 17 minutes. Our control algorithm uses the difference as, essentially, a derivative term.

Coolers have lots of problems, including condensation. Heating to 30C worked fine. This is in maybe the world's biggest single clean room and the local air is always 20C.

I noticed. ;)

You’ve been hanging out with the wrong crowd, obviously. Tsk tsk.

Cheers

Phil Hobbs

snip

:-)

There is a story about Feynman. Somebody bumped into him in a hallway and suggested using multilayer mirrors. He came back two days later with the complete theory of multilayer optical filters.

ASML uses zillion-layer mirrors in their EUV systems, at 13 nm.

We did characterise the thermal time constants of our system pretty carefully, as is spelled out in the paper, which does speculate on whether a better control algorithm would have let us achieve finer temperature control.

Our machine was designed to work on biological specimens, and had to stabilise the specimen temperatures at industry-defined arbitrary levels, most of them a little above normal room temperature. The customers were warned that they'd have to purge the optical path with dry air (usually dry nitrogen) if they set the target temperature below the local dew point.

The Peltier junction was unavoidable.

But they can capture useful parts of reality, if you know what you are doing.

John Larkin's simulated inductors tend not to have any parallel capacitance.

It does happen. You do have to know what to look out for.

Ten year-old don't usually know all that much, and quite a lot of perfectly correct ideas look wrong to them, as they do to you, for much the same reasons.

That's not any kind of organic light-emitting diode. Graphite is elemental carbon, not any kind of organic compound. A ten-ear-old might not know that. You still don't seem to be aware of it.

It's not "sold". It's published as model that fits pretty much all the observations we've collected.

Explosion isn't quite the right concept. The universe is pictured as starting off very small, very dense, and expanding rapidly, but it created the space it expanded into as it expanded.

If shrank back into what it started off as it would clearly be a black hole, but if it were it wouldn't have expanded. There weren't any fragments - the early inhomogeneities are visible as small variations in the cosmic microwave background, and they have been measured.

From a ten-year-old's point of view.

One that doesn't work.

The discovery of electricity preceded the discovery of electrons

- J J Thompson didn't discover them until 1897.

Neither CERN nor ITER is designed to go anywhere - they are static installations. CERN has managed to find the Higg's Boson, so it has made some progress. ITER is still being put together

The trick is to know when it matters. ESR and core loss are usually more important.

I designed this surface-mount inductor for my Pockels Cell driver, after several tries using commercial parts. They all smoked.

It's wound on a specially marked Sharpie pen that we have carefully reserved.

You've got that backwards. Black holes are entirely finite, because they contain enough mass to close space back in on itself.

With black holes it's the curvature of space-time.

Georg Ohm published his law in 1827, 70 years before J J Thompson discovered the electron. It works just as well for other charge carriers.

By which you mean that you can't understand it.

Ask any ten year-old.

Neither CERN nor ITER have much to do within Einstein's interests.

CERN is about nuclear physics, which has moved on quite a way from Einstein's insights, and ITER is about getting nuclear fusion to work, which is even less dependent on Einstein's insights - it does depend on E=MC^2, but once that had been used to explain the mass defect in elements heavier than hydrogen, Einstein didn't have any real involvement.

The popular press made a lot of fuss about Einstein, but Bohr, Dirac, Heisenberg, Schroedinger, Lorentz Planck and Pauli were all in much the same class.

And you don't simulate them either. Simulation is - in part - about letting the math throw up unexpected effects that appear when you hook up a bunch of components. Knowing when it matter relies on the simulation inside your head.

So you didn't read the data sheets carefully enough. It's not a part that would usually be described as "surface mount". If you'd scraped the enamel off the bottom of the coil and soldered each turn down onto an isolated copper pad on the board, it probably would qualify as surface mount, and would have had better thermal contact with the board.

You might have had to make it as sintered metal 3-D printed structure to get this to work - the wound coil looks a bit irregular.

Lost wax casting could have worked too.

That defines it diameter. Measuring that with a vernier caliper would give you a number you could document.

If you'd wound it with copper tube you could have pumped water through the tube, or made it a heat pipe.

A 3-D printed structure would have offered more options.

It better have a regular calibration schedule, or your semiconductor customers may give you the raised eyebrow.

Cheers

Phil Hobbs

Hmm. To be overly serious: With traceability to NIST (US) or NPL (UK) or the like.

The trend in standards is to eliminate standards tied tp a physical object.

I have a Sharpie in hand. The barrel that is not covered by the cap is a truncated cone, being 11.0 mm at the blunt end and 12.32 mm near the cap, 73 mm away.

Actually, all that's needed is to specify an ideal geometric shape, with tolerances, in the formal documentation.

Joe Gwinn

Mine is pretty cylindrical for the length of the coil. I expect that the operator's (ie, my) applied tension affects the radius too.

That inductor sees 25 amps p-p, roughly a sawtooth, at 4 MHz. The Coilcraft parts that I tried all smoked, I guess from skin effect and proximity effect.

I'll have someone start on a SolidWorks model.

Only when they matter.

Simulation is - in part - about

Sure I did. They should have worked, based on the data sheets.

It's not a part

I did that on the ends. I think the gap-pad works better thermally than soldering every turn to the board.

Have you ever used a surface-mount coil that soldered every turn to the board? Got a link?

Losses would be crazy.

And supply a water tank and a pump and water connectors?

Again, massive losses.

My inductor is cheap and simple and works.

If I get a gigantic order, I'll have a coil winding company make them and retire the Sharpie.

Most likely.

I bet you need the standoff, so the lossy FR4 material isn't too close. That should be in the requirements as well.

I'd specify the coil dimensions, not the mandrel dimensions, which may be provided as a helpful suggestion only.

Joe Gwinn

The turns squish down into the gap-pad gunk, which is an OK heat conductor. The PCB under the pad is a big copper pour, top and bottom, with a zillion thermal vias. There's more gap-pad on the underside of the board to dump heat into the baseplate.

At 4 MHz, skin depth is 32 microns, so most of the copper is wasted. That's why it gets so hot.

I tried three of the Coilcraft 1010VS parts in series, but they smoked, probably skin+proximity effect. Maybe parallel would have been better.

I could have a mandrel machined or 3D printed, to more accurately wind the inductor. The improvement would be mostly cosmetic.

Inductors are a pain.