Plotter as digitizer

Plywood makes it hard to miss. :)

--
You can't have a sense of humor, if you have no sense.

Yeah, that was my thinking re: the 2x6's (assuming you mount the cabinets at top and bottom)

Plywood would make extra work pulling the Jboxes "forward" through the extra thickness that the plywood imposes.

Yeah, having a younger body and *extra* bodies always makes things a lot easier! In my case, just this ONE worn out body to do it all! A little over 1500 sq ft though I'll tackle it in small pieces -- hence the appeal of *buying* the lift (so I can take a day or three off when I discover

*just* how old this body has become! :-/

Biggest concern (aside from butt joints on ceiling) is dealing with all the holes that will need to be cut through the wallboard :-/

I build my cabinets, so it wouldn't be difficult to mount them over plywood. Plus, it eliminates any dead space between them & the wall.

--
You can't have a sense of humor, if you have no sense.

In the good old days of metal outlet boxes we would hold the sheet in place & lay a piece of plywood over the box and tap it a few times with a hammer to mark where to cut it. Now, a RotoZip tool will do the job. Screw the sheet in place, away from the box and zip out the hole, then install the rest of the drywall screws. What was fun was having to drill pilot holes to drywall 100+ year old hand hewn oak studs in my house in Ohio. They were so hard that the screws would snap, if you didn't.

--
You can't have a sense of humor, if you have no sense.

French cleats made out of 3/4" ply work like a champ.

Some metal (thinner) versions:

One aspect of schematic capture that I dislike most is coming up with aesthetically pleasing "layouts". I am loathe to needless crossings, "backwards" signal flow, etc. And, at the same time, dislike having to spread a functional block over more than one or two pages (too hard to keep track of what's going on -- hence the reason for the oversized sheets!)

Having come up with an acceptable "layout", its much easier to reproduce it exactly than it is to have to "move stuff around" because the new symbols don't *fit* (in the same space) or have their connections indifferent places.

I've heard Bad Things about Eagle.

Yup. Definitely not worth any potential offsetting gains! :-/

That's how I've done it in the past. E.g., I deliberately used mud rings to make this easier! But, the ceiling fixtures are a bit less robust (since they aren't designed to expect any mechanical stresses)

I've been told to try that approach. But, imagine I would have to remove the devices from their boxes? I.e., how do you avoid gouging a receptacle, etc.?

Yeah, I used 10 pounds of 2.5 - 4" screws to do all this framing. A bit more work (pilots, etc.) but has paid off as I have encountered things that I needed to change from the original plan. I am amazed at how *hardened* the philips tips ("bits") are! All the abuse that I've put them through (basically, a single bit -- though in a 10A *drill*, not "powered driver"!) has torn the hell out of lots of screw heads but the bits still retain their shape!

Yes, that's a problem I always wrestle with, too. I don't like large sheets because they're too hard to use in the lab, where they're needed. Depending on how individual circuits lay out, I use either letter or ledger sized sheets. More sheets aren't necessarily bad, but larger, cluttered, sheets are. OTOH, a schematic has to be more than a netlist (one component, or less, per page). It's always a tradeoff.

I'm not sure what you're getting at here, exactly but I make as many symbols "standard" as I can. All gates are the same size, all opamps, etc...

I *know* bad things about OrCrap. ;-)

What I meant is that one can create components that are an excellent match to the originals, where no existing components have the desired rotation, size, orientation, etc. Taken to extremes, one could completely ignore the existing library components and create new ones that would reproduce the routing in the original schematics perfectly, for all intents and purposes.

They make several very competent eCAD / layout packages at very economical price points. Versions of their software run natively on Wintel and Linux platforms without issues. Unlike many other past players in this market, Eagle appears to be a very well-run company that is unlikely to leave users in the lurch.

The freeware version is all I have ever needed for design of jigs and fixtures. It is plusgood.

I loved the full-boat Orcad years ago, but found their 'component creation' process to be nearly a vertical learning curve. In contrast one can create components in Eagle easily, even if it means one must 'go around' Eagle's documentation to get sensible answers! :) (Disclaimer: The very latest documentation and tutorials are probably much better than were the originals.)

They have yet to come up with a graceful and fair way to address copyrighted circuit configurations, but that does not affect either you or me and is unlikely to, any time soon.

(Aesthetics, functionality)

Indeed. One has to appreciate one's pain threshold. :)

--Winston

I'm not usually worried about that. Its a lot easier to find a big tabletop than trying to "glue" a bunch of little dinky sheets together (in your mind as well as on that tabletop) so you can understand where a signal is going and the role it plays in the circuit/design.

Nowadays, that's my preference. Everything B sized. Unfortunately, even B size isn't big enough for some of the SoC's and highly integrated components, nowadays. I despise arbitrarily cutting a single chip in half just so you can display all of the connections on those *two* arbitrary blocks because the *one* wouldn't fit on a page. Likewise, I dislike putting outputs on the top or left side of a symbol "just because that's the only spot with room available". Similarly for inputs.

I've worked with designs where the schematic was several B-size binders. You end up with lots of bogus signal names devoid of any meaning -- just to allow a signal to cross a page boundary. Chasing down signal paths takes up most of your debugging effort. (Heaven help you if the page you seek is missing or out of order!!)

I mean the arrangement of symbols on the page so that it conveys meaning in and of itself. So you can *see* how signals and components relate instead of just putting components wherever you happen to have some whitespace!

The schematic is the designer's *product* (deliverable). It should reflect the care that was put into the design. If, OTOH, you just start drawing on Day 1 and put things willy-nilly as they come to mind, it doesn't speak well of your development process. E.g., the cliche: Think Ahea

In contemplating how to automate schematic conversion from, say PSpice, to Virtuoso, the pin-out grid is all you care about when it comes to matching symbol-to-symbol. Sooner or later I _will_ crack that problem ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    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 http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

I've never had a "big table top" in a lab. Not one that wasn't full of test equipment, anyway. 'C' sized prints just pissed me off. A cow-orker designed on them, and then sat around making origami when he was debugging. As long as the schematics are broken down into logical blocks, it's no problem understanding where a signal goes or what it does. Of course I draw block diagrams (front sheet) and always use off-sheet references.

Ledger fits nicely into a 3-ring notebook. I keep one for each project. As far as splitting a block, there is no reason to do it arbitrarily. I'll often separate address busses from data busses from power and JTAG. Sometimes I'll separate I/O banks on FPGAs. Whatever makes the most sense. It makes it far more readable than a huge block with everything descending on it.

Someone had no imagination if they had to make up bogus signal names. It was a very poor choice of organizing the schematic. I've worked on mainframes, where the logics were in 30-40 (or more) C-sized binders in cart(s). I don't remember any bogus names there, either. Some got pretty complicated, though. I have seen some pretty bogus names in the VHDL inside processors, though. ;-)

Oh, absolutely. A cow-orker used huge pages, then put 20 little circuits on them, wherever they fit, connected by reference only. It was impossible to follow.

d ;-)

Well, the working widget has always been my deliverable, but your point is well taken. The schematic is a *lot* more than a netlist.

Yes, but often there are constraints that pop up when you start trying to address all of the peculiarities of a particular symbol representation. E.g., minimum font sizes, letter spacings, support for overbars, etc.

I've just been bit in the past trying to move drawings from one package to another -- only to discover that I can't draw a particular component the same way *in* the same relative size.

*Nothing* beats DASH when it comes to crappy symbol editing! OTOH, I found STRIDES to be one of the most *efficient* (in terms of user interface) programs I've ever used. It's as if the developers actually had *real* experience drawing schematics (instead of just "conceptual knowledge" of the process).

Of course, the features that are present in packages

30+ years later make comparisons impractical (OTOH, I suspect STRIDES running on a 3GHz processor would have the schematic *drawn* and *plotted* before you had placed the first component! :-/ [I wonder if I have any ISA machines lying around that I could try that on...]

C size isn't worth the hassle. If you end up on C size, you can *probably* find a way to make things fit on B size.

When you need to move to *larger* sizes, OTOH, you've no chance of getting things to fit comfortably on B.

I've tried drawing on D and reducing to B -- but the resulting images are just too hard to read.

This is only practical when you are dealing with simple (or highly integrated) blocks. E.g., drawing a pipelined adder built of CSA's on multiple sheets quickly becomes impractical. You need more than a single bit per sheet (else you end up with 16, 32, 64, etc. sheets *just* for that adder!). Yet, you can never really get "a lot" of bits on that single sheet without making it look overcrowded.

So, settle for 4 bits? 8? In either case, you end up flipping through lots of pages that, together, represent one simple "device" (functional block) in the grander scheme.

less,

When you end up with very little on a page, then you *have* to label lots of signals that don't really deserve names in their own right. I.e., why label the carry and partial sum outputs from each CSA if they are only going to feed the next stage of the pipeline? (i.e., if that stage was present on the same page, you would never

*bother* to label them -- since their roles are obvious and the signals don't go anywhere else in the design!)

This is just stupid. Saving pages for no real reason. I vacillate in how I handle things like connectors -- put them

*with* the circuits to which they attach vs. group them on "connector pages" vs. some combination of the above. But, in either case, it is obvious what the role of the connector is in the circuit. If it is on another sheet (even sharing that sheet with other connectors), you can forget that detail and just remember "the offpage references here are just for the representation of the connector"

Someone else might be charged with fabricating the prototype, etc. But, I've never been anywhere where I "dictated" a design to someone else for schematic capture :-/

Hey Don,

Winston >> What I meant is that one can create components that Winston >> are an excellent match to the originals, where no Winston >> existing components have the desired rotation, size, Winston >> orientation, etc. Taken to extremes, one could Winston >> completely ignore the existing library components and Winston >> create new ones that would reproduce the routing Winston >> in the original schematics perfectly, for all intents Winston >> and purposes.

I've never run into that. I *do* rely on the native symbol libraries extensively, though. One can make an easy-to-use and aesthetically pleasing schematic using practically any capture program.

(...)

There is *lots* of competition for that honor. Some Linux open-source CAD packages I've tried appear to take their user interface queues from Salvador Dali.

Back at Xerox, the software guys would always complain when we were compiling ABEL code on our shared VAX 11/785 because it caused their work to slow to a crawl. :)

Why ISA? For the dongle?

--Winston

Virtuoso won't allow you to set the grid to anything you want? ..Or that setting isn't available under script control?

--Winston

That's the max size on the printer. I would reduce it to ledger but some text got hard to read.

Sure.

Um, if I'm "drawing" a pipelined adder, it'll be in VHDL. ;-)

For mainframes, back in my deep dark history, one bit. Today I wouldn't use a schematic at all. Such things are much better in VHDL. Even there, the lowest level of the hierarchy might still be one bit. Yes, even with schematics, I make good use of hierarchy.

less,

Then don't do that.

They're easier to "probe" if named.

Again, I'd never use schematics for such a project. Things have progressed in the last fifty years.

Plotter paper is expensive. ;-)

Yep. Me too. I do have one symbol for the connector and try to make it look something like the physical connector. I *never* split the connector into individual pins, then scatter them to the winds, like some designers. Ick!

I don't do the layout, either, but I am responsible for it and everything else. My deliverable is ultimately revenue.

That;s fine when you are drawing *with* a capture program! Note that I mentioned these drawings were done "freehand" (i.e., with a "logic template" and a "lead holder")

[Think: late 1970's]

Or Timothy Leary?

Exactly. I suspect I *might* be able to play games with the BIOS to get it to allow me to use a serial mouse on a "regular" serial port (and just map that port to the right I/O address).

It's fun to go back and revisit old tools on modern hardware. E.g., I have an old copy of BRIEF that is completely unusable on modern machines because the software timing loops (keyboard repeat, etc.) happen in microseconds instead of milliseconds. I.e., hit the "down arrow" key and you are at EOF before you can lift your finger off the key!

Shame. So much lost technology due to shortsighted implementations...