another board layout

On 7/28/2018 9:30 AM, John Larkin wrote: (Snip good stuff)

Which version of LT Spice, John?

JohnS

The new one, XVII. It only does that one time in ten or so, so it's no big deal. I always keep my left forefinger over the ESC key. In PADS, too.

Yes, I think that version has several problems that you might discover along the way.

I never had a problem with IV and I still prefer it. Tastes vary.

I think the old one isn't being supported. If I run the old version, it wants to update but the update fails. Does IV still update? Does it include the Analog Devices parts?

XVII is a bit faster and seems to work fine. Of course, the simulate/edit simulation thing is horribly broken in XVII.

Yup. Plus I hate the special little editors for params and stuff. (The simulation command syntax is complicated enough that the special editor is helpful, but not as good as IV's.)

I've taken to putting all the params in a block with a comment as the first line, and the same with the simulation command.

That way, clicking on the comment brings up the normal text editor box, so it's easy to move the semicolon from the .TRAN line to the .NOISE line, say.

Cheers

Phil Hobbs

No, it doesn't seem to update. I'm not surprised, though. Well, the parts are the ones I have on hand. The parts have probably not been updated since the change.

Yeah, that's one of the reason I'm staying with IV.

Switching between transient and frequency response is annoying in XVII, but I don't do much frequency sweeping so it's OK.

I do like having (passable) models of the Analog Devices parts.

'scuse my ignorance here, but these IC models you get with LT and rest of 'em, do they go to the trouble of building a full & complete sub-circuit in software of the original device, or is it just a black box job that mimics the responses to input signals that the original device has? (IOW just a couple of formulas bolted together IYGMM). Taking a jelly bean op- amp as an example, say. I've often wondered...

My understanding is the more complex devices are functional models that run faster. I doubt they construct transistor level models of most designs, in no small part because they don't want the competition getting proprietary info.

Rick C.

I think they are behavioral models, because full transistor-level models would run too slow. Probably hybrids. Some things are obviously short-cut, like internal current sources that keep going when there is no power applied to the chip.

Of course quirks don't get modeled. Even a full-transistor model can miss things like sneaky substrate currents.

I simulated one rrio opamp whose negative output saturation was

I'm guessing that the Analog Devices models were thrown together fast.

The MC33078_MC model does that too.

Cheers

Phil Hobbs

It's placed!

almost. I'll let my layout guy place the other three channels. And a few parts, series terminators and one bypass, aren't necessary so we'll ECO them out. I always put too many bypasses on the schematic, that don't look necessary on the layout.

It's hard to place parts on a low density board like this, without leaving big open areas that look silly.

Those four signal transformers are gigantic, but we have them and they fit.