Help! uWave xistor bias oscillates!

IIRC, it did take a few cans, didn't it? ( maybe that's why I don't IIRC!)

Thanks for the tip Jeroen. AXV makes 'em too, as I discovered on the next page of their .PDF. Table 2 lists an ESL of 610pH for their 0612 package, and the text projects 530pH for a 0306.

Your parts are better: Syfer specs 500pH for their

Thanks again. James Arthur

Hello James,

For GHz circuits I always try to create enough of an island that the capacitance between this island and the ground plane kicks in. These ultra high frequency caps can be pricey.

Regards, Joerg

I read in sci.electronics.design that snipped-for-privacy@yahoo.com wrote (in ) about 'Help! uWave xistor bias oscillates!', on Sat, 30 Apr 2005:

The inductance is just a function of physical size. A difference of 500 pH to 610 pH for same-size components is difficult to explain.

"John Woodgate" wrote in message news: snipped-for-privacy@jmwa.demon.co.uk...

page of their .PDF. Table 2 lists an ESL of 610pH

Difficult, yes. For several reasons, it is easy to believe.

The inductance contributed by the interleaved portion electrodes would be affected by the number of leafs and their dimensions. Capacitors with the same outside dimension can vary in both respects.

The results might be obtained by trying to measure the inductance from different portions of the endcaps.

There is also an interesting issue in the definition of a two terminal inductance -- one which I would like to see resolved if that is possible. Presumably, when an inductor is measured, (especially a small valued one), inductance of the rest of the circuit (the test equipment and connecting cables and fixture) is supposed to be eliminated from the stated result. This is commonly done by placing some object other than the DUT into the unknown position and "nulling", where the object is assumed to have known (or 0) inductance. However, unless that object and the DUT have the same magnetic field interaction with the unchanging portions of the measurement circuit, the result will be skewed as it is dependent on the test circuit geometry.

This is obviously a bit academic for normal inductors, wound to increase internal inductance. But for bits of metal that resemble a straight conductor, I do not see a good way of glossing over the issue if one cares to distinguish 500 pH from 610 pH and claim either to be right or wrong.

"Fred Bloggs" wrote >>

Simulation is not going to have any bearing on how inductance varies with plate count and dimensions.

use, these various factors are considered

Of course, they also take into account their own production processes and constraints. For low voltage ceramic caps, uniformity of the dielectric imposes limits on thickness versus yield. So it would be naive to think that all ceramic caps of the same type, size and value are the same just because cap makers hire competent engineers.

of the measurement.

Assertion without evidence or argument. I see no reason to believe that some measurements are not made from endcap ends, and others made from endcap sides. Nor do I see any reason to believe that the effects of such differences cannot explain the 110 pH difference reported earlier.

As for "frequencies make this ...", that looks like irrelevant gibberish to me.

plane extension effects has been characterized

It is clear to me that you have missed the point. The fact that any instrument purports to provide some kind of correction for fixturing does not eliminate the definitional issue. The simple fact is that there is no defined way to partition mutual inductance between a component and connections to that component such that a lumped inductance can be assigned and claimed to belong to the component, except by fiat.

calibrations as you suggest.

Amazing that you take my "known (or 0) inductance" to be perfect anything other than inductance. You are so desparate to detect ignorance that you invent it.

That makes clear the depth of your delusion. I have used a variety of impedance analyzers, as well as a few network analyzers for measuring impedance.

I read in sci.electronics.design that Larry Brasfield wrote (in ) about 'Help! uWave xistor bias oscillates!', on Sun, 1 May 2005:

For a particular meaning of 'believe'. I think you imply that the difference is due to different methods of measurement, and thus the 'lower' inductance part may not in fact be better. If so, I entirely agree.

With the same dielectric, the options are few. We don't know if the dielectric are the same.

Indeed.

This is why test fixtures with precise dimensions and mechanical stability are used for measurements of small inductors and capacitors.

I wouldn't gloss over it at all. I'd ask the manufacturers if their measurements are traceable to national standards, and if so, how.

There is such a thing as simulation, and of course, the capacitor manufacturer employs competent engineers for whom these things are not as mysterious as they are for you. When they embark on a new product line specifically for low ESL high frequency use, these various factors are considered beforehand and not after the fact.

Not quite- the probing and frequencies make this the least significant error of the measurement.

As usual, you are fretting like the ignorant wisp of an intelligent lifeform you are- every single one of the test fixture, fixture probing, and reference plane extension effects has been characterized accurately and the impedance analyzer provides for their correction. It is not necessary to present simple-minded perfect opens and shorts for calibrations as you suggest.

You should keep the list short and just tell us what you do understand. As usual, your post is more space-taking, mindless, garbage. It is clear you have never used an impedance analyzer in your life.

It most certainly does, idiot- the simulation drives the choice of plate count and dimensions.

Are you talking to yourself or some other internal demon? That statement does not seem to be in response to anything I wrote.

You just don't worry yourself about it- let Agilent and the T&E community work this out. They seem to have gotten along just fine without your assistance thus far.

plane extension effects has been characterized

No- that is bs- they have 4T connections that eliminate fixture mutual inductance effects- there is no "purporting" to it.

calibrations as you suggest.

So says you....

And what drug regimen were you on at the time? Some with hallucination side effects to be sure....

Not a bad idea when you've got the real estate. I measure a loose piece of 304 x 80 x 1.5mm (0.062") glass/epoxy material (presumably FR-4) from my heap at 660pF, or 2.7pF/cm^2. Using the thinner 0.031" stuff, and situating your island between solid planes above and below, one could get quadruple that to about 11pF per cm^2, and cheaply too...,

if space-constrained, even 0603 MLCCs used near series resonance work pretty well...,

and if you've got budget, see AVX's Low Inductance Chip Array:

By controlling the current flow in the capacitor's plates (Fig. 4) they cancel quite a bit of the plates' mutual inductance - they claim ESL under 30pH (pg. 45).

I'd suppose that such subtleties of plate design might account for the difference in ESL between the Syfer and AVX 0612-sized caps. In particular, plate sizes, overlaps, plate terminations, and number of plates could all vary slightly between vendors.

Cheers, James Arthur

Hello James,

In your case you could extend the small strips on the DC side of the inductors to under the caps, under the resistors, in between etc.

Inductors can also be done on FR4. At a few GHz you don't need much and this way you have the performance under your design control, plus they cost next to nothing.

What I used to like were embedded beads that follow a stripline pattern. But they are sometimes frowned upon by PCB mfgs because of the large drill sizes. Plus you had to place a drop of glue onto them or they'd rattle. But at a few GHz ferrites aren't so hot, they are just to muffle any lower frequency noise that could enter.

My designs usually are.... so, can't use the fancy stuff.

Regards, Joerg