Trace inductance... Hobbs??

Does this help?

Throws up on my small dimensions :-(

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

HP-AppCad works, might be helpful if you do lots of RF design:

If you already have Z and also C per unit length, wouldn't it be the usual relationship Z=sqrt(L/C) ? (both C and L per unit length)

Of course I don't know how leaky your field oxide is.

Perhaps you cancalculate a scales analogon?

The usual answer for fast on-chip interconnect is, "not enough to matter"--they're almost always RC. As Joerg suggested, I'd probably try to work it out by the characteristic impedance if you know that. The actual inductance will depend on the oxide thickness and the substrate doping level--the image charges want to be about 1 Debye length into the substrate, and the Debye length depends on the carrier density. For a zero order answer, I'd probably find out the doping level, look in Gray & Meyer for a plot of the Debye length vs doping, build a 500,000x scale model and use a grid dip meter.

But then I'm a dinosaur. ;)

Cheers

Phil Hobbs

Wouldn't you rather use TDR?

ATLC isn't bad either.

John

I dunno--TDR is a pretty indirect way of measuring inductance, especially inductance that's as lossy as that one, isn't that so? Besides, I have this nice GDO with a 6CW4 Nuvistor or something like that in it, that I haven't had a chance to use yet...

Also, on further reflection, ;) due to the substrate resistivity the return current is liable to be spread out through a big enough volume to matter--the charge distribution won't be the same as the electrostatic case.

I haven't used ATLC, but it looks very useful, at least for perfectly-conducting metals...thanks for the steer. It could certainly calculate the wire mockup, but from a quick read of the docs I don't see how to specify a volume resistivity for the conductors.

Cheers

Phil Hobbs

For a "University of Science and Technology" imho a pretty poor presentation.

"h: height of the trace above ground"

Above ground or above medium?

"r: relative permeability of the medium"

Permeability of the ground or of (which) medium?

I am wary of people applying formulas to a problem like mothers would apply bottles with unknown contents to their babies, and this seems a good example of that illness :-).

Perhaps it wouldn't be half as bad had they at least given a complete example which the user could verify with his own data. For good measure, a range of applicabiltiy should have been given, too.

Regards, H.

I agree with Phil that the RC will likely dominate.

That said, a few years ago I found a formula that gave something on the order of 1pH/um for a line width on the order of 1.2um. As I recall it varied slowly, i.e. less than linearly, with width. Perhaps I can locate it tomorrow....

steve

To send email, remove the capitalized food item.

Yeah, it's a kludge and not meant as "scientific". But it has helped many folks at least over here to get a ballpark understanding of the topic and not completely screw up their layout.

I simulated it as a hundred lump R/C line, ignoring the inductance. Delay of 4.6ns, risetime of 8.57ns, still good enough for my application.

...Jim Thompson

Must be pretty slow stuff then ...

Let's see you design a 10-Bit ADC... absolutely monotonic (no trimmed components, 3% resistor ratios), 12us conversion time, < 700uA total power consumption ;-)

...Jim Thompson

Jim, I thought you were retired ?

Phil Hobbs wrote in news:htCdnYKRqIxkBf_XnZ2dnUVZ snipped-for-privacy@supernews.com:

As others have stated, it's probably negligible in terms of it's real impact. However, for future reference, the rule of thumb for almost any small conductor is 20 nH per inch. The size & shape of the conductor is largely irrelevant. Just getting current from Point A to Point B will produce roughly that amount of inductance. I was taught this about 25 years ago by the top RF guru at MIT Lincoln Laboratory, and I have yet to find a case where it was off enough to matter much.

Doug White

Interesting! That's actually close enough for government work to Goldstein's 1pH/um.

Who was the guru at Lincoln Labs? (I'm MIT, '62).

...Jim Thompson

He won't retire until you hear a loud thud from his office and then the trackball falls to the ground. We'll probably all get there, can't let go :-)

Yes, I can understand that. The way the system is going, I think a lot of us will be retiring with our tools in hand.

Please don't misconstrue that last part :)

Should we stop doing what makes us happy just because we make (good) money at it? I could have lived (not with all the toys, but lived) on my retirement and savings, but why? It's work, sure, and not everything is perfect, but I've never has as much fun as I'm having now.

No, I did nine months of that "retirement" stuff. It was fun rebuilding the house (being sick for the first three wasn't so much) and finishing all the projects I'd started. After that was done and SWMBO put a stop to the mess making, to put the house on the market, I got bored *fast*. Daytime TV really sucks. Fortunately, it only took a couple of weeks to find a job; contract, but at Jim's kind of money. ;-)