capacitance between pads

NSC an-1205 says 0.1pF, IIUC. But I think (hope) it's probably a little less. What if there's a 10-mil ground trace between the pads?

The gross (big pads, no fringing) capacitance between sides of an

It depends what's underneath and once that is known I'd use one of the online cheat sheets:

If that is based on the UltraCad stuff, I wouldn't entirely trust it.

The Saturn PCB calculator looks really nice:

One of my guys did check its asymmetric dielectric diff pair against an em simulator, and it was close.

So far I have lucked out. At least down to the 100psec pulse widths everything panned out. You guys probably go a lot lower than that. Problem is, I have to fly blind part of the way and have to wait for the results because I don't have one of those 20GHz sampling scopes. Meaning the circuit tests itself bug then, of course, it's quite late in the game. Nail biters like with the New England Patriots.

You can always drop in here and use one of our scopes. We go to 7 GHz realtime and 50 GHz sampling. You would have to buy the beer.

Nothing is underneath, bare FR4, no copper. It's a simple case of side-by-side pads.

Want me to measure one for you? My Boonton is on my bench.

I thought you'd never ask.

This is a plain double-sided board.

C56 = 0.3 pf. That includes the traces and vias.

C34 = 0.06 pf, just the pads with traces cut away.

Measured in a Boonton 72B, probably not super accurate.

I also have instruments that can measure to 0.01pF, but setting the zero value is a bear. Your 0.06pF measurement is encouraging, better than the 0.1pF numbers I've seen. Well, sheesh, I need 0.02 or 0.03 pF. I'm hoping that judicious ground traces can make up the difference. Alternately, I can entirely eliminate some pads on the pcb (and cut device leads). Maybe that's a better approach.

I bet a ground plane would help a lot. I'll try it again on a 4-layer board with L2 ground plane, but it's going to be hard to resolve on the Boonton. It's an analog meter with 1 pF full scale.

I think there is a way to measure 0.02 pF with a scope and a signal generator.

A ground trace between the pins would help, too. But the SO8 package itself will have C between the pins, and inside.

How about some unplated drilled slots between the pads?

Er of FR4 is ~4 vs. 1 for air.

--sp

Am 22.02.2017 um 02:31 schrieb John Larkin:

Maybe he can bring some of his home brewed beer? ;-)

Jorgen

I think not. He likes that dark stuff, and I don't like to drink anything that I can't see through.

We'll have to go to Zeitgeist.

Or grounded plated slots. More capacitance to ground, less pad-to-pad.

Or switch to a single op amp so it doesn't matter as much. ;)

Cheers

Phil Hobbs

A tight row of grounded vias, maybe zigzag, would have about the same effect.

And make the SO8 pads smaller!

Thanks. That would also be a good excuse for a visit.

Since I am married I can only brew the occasional dark stuff and that is almost pitch-black. Then there is the Belgian Tripel which looks light-amber but has almost 8 volts. Most other beers I brew are golden-colored and between 4.5% an 6%. Pale Ale, IPA, Cream Ale, Koelsch. Slightly darker but still to your taste would be Irish Red, Autumn Amber and Copper Ale (like a miner's beer).

Looks like the Marco building only a few more steps walking distance to Zeitgeist. Otherwise I couldn't imagine you guys even considering a move.

That, and if cost is not too critical consider board material with a lower dielectric constant.

In the extreme maybe porous silicon dioxide.