Hi-Q inductors

We used them in some front end circuits in our telemetry products. A spiral on glass, with a set of leads attached. The only problem I ever had with them was that the manufacturer had scratched the negative, and an almost invisible line shorted out one turn. Of course they were on the bottom side, so you couldn't see it, after it was installed. When I found that, we checked the remaining stock, and ll were defective. All it too was a new Exacto blade and a few seconds each to remove the extra plating.

Sure, but it'll be like towing a car up to 30mph and then suddenly letting the clutch snap. Screech ... vroooom. It still takes a while until it has smoothed out.

If it wasn't so far away (and my wife didn't have so many honey-do projects on the list) I'd come over. Got to clean all the sawdust I made on the deck this morning, one more board to cut.

Sometimes the 330nH inductors are a bit higher Q, at least at 50MHz.

I really don't like plastic bobbins for this kind of stuff. Either air core or ceramic. But the ceramic has to be the good stuff.

Back in my ham radio days at high power it even mattered to keep the surface smooth. And that was below 30MHz. In our club we had an unspoken rule that one Saturday a month was "coil polishing day". I went through several tubes of Wenol paste over the years. Ran out a couple years ago and became concerned, thinking this might be unobtanium in America. Then I found two large tubes at a kitchen store. They became mine immediately.

Other option: Wind an air core inductor and let it ride 1/4" or so above board, then test your circuit again. Even off-the-shelf ones are listed with a minimum Q of 100:

A great opportunity to get your Business Person And Bean-Counter-In-Chief to let you have an HP 5373A pulse modulation domain analyzer. They show up on eBay for $3k or so all the time. I've been trying to justify buying one for years myself.

(There's one there for $1k OBO right now--probably just needs a backup battery replacmenet.)

Cheers

Phil Hobbs

Good luck differentiating a Web calculator.

Cheers

Phil Hobbs

Just do it numerically. That's often easier than actually doing the symbolic math. In my case, I don't even need the derivative; I just need to know how much L will change from, say, 20 to 50 C as the FR4 expands. Engineers are craven that way.

John

What I'm interested in here is using the triggered oscillator to count off time delays, and my criterion for goodness is RMS jitter versus time, where "time" is counted in 20 ns steps for my 50 MHz oscillator. That's best measured with a sampling scope.

I think I'll make a little board. It will have a good 10 or 20 MHz XO and a synchronous counter, and a bunch of candidate oscillators to test. I can use Q7 of the counter to enable the LC oscillator, and use other counter taps to trigger the scope, maybe with a flipflop in there somewhere. The point being that I can run long oscillation shots, trigger the scope at various times during the run, and measure the jitter of various edges. An 11801C has under 2 ps RMS jitter for short delays, but jitter piles up fast, so I need a better way to trigger the scope than just using the oscillator kickoff.

Or maybe I could use a digital magnitude comparator, and a resync flipflop, to let me walk the trigger anywhere within the full count cycle. Something like that.

Hmmm, even a one-shot and a trimpot, followed by a clock resync flipflop, would do. That's simpler. Or I could use one of those "digital delay generator" things.

OK, final offer: Very good XO drives a DDG and clocks two flipflops. The DDG can be programmed to drive the D input of each flop, so I can make an "oscillator run" gate and a scope trigger, all nicely movable, but with edges quantized to the low-jitter XO. That should let me measure the oscillator edge jitters to a few ps RMS out to many microseconds.

I wish people would make a family of LVDS logic.

John

Small FPGAs?

That's exactly what the HP 5373A is good for. It'll also display the chirp in the pulse, the pulse-to-pulse phase error, and so forth.

Cheers

Phil Hobbs

estate.

But I doubt it has the few-ps resolution I want. It looks like it timestamps edges to 200 ps resolution.

My little rig would let me measure the time from start and jitter on every edge, only in the time domain. Tediously, of course.

John

estate.

Picky, picky, picky. You can average to improve the resolution, but the HP modulation domain gizmos are really just glorified universal counters. You guys should build one with that nice timing interpolator you use in the P400.

Cheers

Phil Hobbs

voids

estate.

We've built TICs/TDCs in the past - still offer a couple - and had been musing on a time-interval counter with ballpark 1 ps resolution and jitter, sort of a rationalized Wavecrest. But for some reason we don't sell many time interval gadgets, maybe because oscilloscopes satisfy most of the market needs. Digital delay generators sell better, so that's what we do.

A couple of other people make ps resolution time-interval stuff, like Picoquant. We hate competition... who wants a price war in a tiny niche market?

John

Coilcraft has a great tool for finding the highest Q inductors at this link:

ke

of

Ahh math fun. I treated as a trangle and got 50 feet. (Assuming I didn't punch number wrong into my calcualtor.) (I also assumed a mile was 5,000 feet.)

George H.

t -

Show off.

George H.

(Grin)

bake

n't

est

ure

by

gs

s,

gh

at

nk of

to

al

^^^^^^^ oops triangle

xt -

I suppose, if showing off a basic, high school grasp of geometry and trig meant much.

The reason I posted, at all, was because it reminded me of all those hours grinding mirrors and my high school hours spent learning enough geometry and trig to compute curvature depth for the rougher grinding phases. I just enjoyed the recollection moment and once I enjoyed that, figured I might was well post a short response.

Jon

Yes, but a quick mental thought experiment says that there is significant error in treating it as a triangle. Imagine the two triangles you propose, where you 'know' the hypotenuse and one of the sides. That should yield too-large an estimate for the height at the center because there is a very broad curve in reality, as opposed to a stiff line and this broad curvature will allow the center to 'slump' a bit from where it would be if half the (mile+1') were split between two hypotenuses. And the difference would be significant given the huge span involved.

But you'd get into the right order of magnitude.

Phil may have chosen the problem itself to make a simple point. But he may have also supplied the precision he did for those who might imagine to try the hypotenuse approach to let them know they "aren't quite there."

...

My first thought was that it reminded me of all those hours grinding mirrors and testing them with tiny light bulbs and razor blades (Foucault testing) and then various masks and Ronchi gratings. And then my time learning geometry and trig to know better what in the heck to shoot for in terms of measurable depth in the curvature (or sagitta [I didn't learn the term versine until _MUCH_ later than those high school days]) for the rougher grinding phases.

So the idea of using triangles only momentarily came to mind. I tossed it, right away. That would have been a huge error in computing sagitta that couldn't have been accepted (were it that I ever needed to compute sagitta from some length of the curved surface itself) for rough grinding. So I didn't even bother to actually compute it that way, despite knowing that it would be in the very rough ballpark.

Jon

cal

Hi Jon, I hope I didn't get your dander up. I am impressed that you can recall math formula for cords and such from high school. (I'm not sure I ever knew them.) I remember being much relieved in tenth grade when I heard the statewide Regents exam for trigonometry had been stolen and that I would not have to take the exam. I found all those trig formula confusing. No one taught me Euler's equation so that I could derive them for myself. I loved Geometry on the other hand. Seven 'theorems' and the rest followed.

George H.