Breaking News
Blog

Why I Prefer to Create My Own PCB Symbols

NO RATINGS
1 saves
Page 1 / 2 Next >
View Comments: Newest First | Oldest First | Threaded View
<<   <   Page 3 / 6   >   >>
antedeluvian
User Rank
Blogger
Re: Paper models?
antedeluvian   6/6/2014 2:47:52 PM
NO RATINGS
GSKrasle

If you can get seven-segment fonts, then why not an overbar font?

Good Point! That is more of a windows thing, so surely some thrird party can do that- any volunteers?

Can Altium load fonts?

Yes it can, but I would be wary of it. If you change the fonts on a symbol and then do a Force Update, it reverts to the default font. Apparently an acknowledged bug!

 

GSKrasle
User Rank
Freelancer
Re: Paper models?
GSKrasle   6/6/2014 2:39:54 PM
NO RATINGS
Overbar?

It's possible letter-by-letter in [one common word-processor] using proprietary character-combining commands, but is awkward and nonportable. It can be done in equation editors too, but again, nonportable.  "Combining Overline" in Unicode is supposed to work, but this is what appears here when I try it: 1.

http://www.fileformat.info/info/unicode/char/0305/index.htm

http://wordfaqs.mvps.org/overbar.htm\

If you can get seven-segment fonts, then why not an overbar font?

Can Altium load fonts?

(A seven-segment font source: http://www.twyman.org.uk/Fonts/)

(Seven-segment fonts are useful for writing instruction manuals when a controller with such a display is being discussed: Instead of saying '"Press the button until the display says "start seq,"' you can show what the display would actually look like. I originally used a drawing and manually turned segments on.)



 

 

antedeluvian
User Rank
Blogger
Re: Paper models?
antedeluvian   6/6/2014 1:32:20 PM
NO RATINGS
GSKrasle

 

but why not overbar?

At least both PCAD and Altium have that. In Altium though you have to do it for each letter in the line and not once for the whole line.

I seem to remember it was possible in Autocad, but I reaslly doubt anyone is doing schematics in that anymore.

 

zeeglen
User Rank
Blogger
Re: Paper models?
zeeglen   6/6/2014 1:21:08 PM
NO RATINGS
@GSKrasle a way to indicate active low signals?

Good question.  A few more:

~signal

signal~

-signal-

signal#

markkk
User Rank
Rookie
I find CAD package symbol libraries aren't very helpful
markkk   6/6/2014 12:18:16 PM
NO RATINGS
I almost always draw my own symbols which replicate the physical layout of the chip(s) involved.  Usually a PIC in my case plus whatever else is required around it.  This means that as I'm doing the schematic I'm planning the PCB layout at the same time.

GSKrasle
User Rank
Freelancer
Re: Paper models?
GSKrasle   6/5/2014 5:52:25 PM
NO RATINGS
Oh, and can we PLEASE settle on a way to indicate active low signals? The gods of word-processing give us underline, strikethrough and double strikethrough, bold, italic, super and sub, but why not overbar? It seems like the attribute set is optimized for lawyers only.

\signal\

/signal

*signal

signal_N

....

   

zeeglen
User Rank
Blogger
Re: Paper models?
zeeglen   6/5/2014 5:04:18 PM
NO RATINGS
@elizabeth I especially agree with your comment about "hidden" power pins.

Ditto.  The librarian can hook up hidden power pins wrong such as in http://www.eetimes.com/author.asp?section_id=30&doc_id=1284662 .  When power and ground pins are accessible one can place a few ohms between each device and power plane to reduce digital noise, and force each power pin to have its own decoupling capacitor.

I had one library component that showed two diodes in a 3-pin package.  The specs in the corporate data base looked good so I used it.  When I got the prototype boards back I discovered that the part was actually a single diode with 2 pins connected together!  Then when I got the symbol corrected, everybody else that used only one diode of that same component had it disappear from their schematics.

elizabethsimon
User Rank
CEO
Re: Paper models?
elizabethsimon   6/5/2014 4:45:22 PM
NO RATINGS
Excellent summary of rules. I pretty much agree with everything.I especially agree with your comment about "hidden" power pins. When i was using PADS, I re-did many of their standard symbols that had hidden power connections. It wasn't until later that I fully embraced the separate power block concept.

The company tht i work for has a PCB group that does the clean up on the schematic before starting on the PCB layout. they enforce most of the rules you mention. We do allow both off-page and on-page connections. The off-page symbols indicate direction and are labeled according to sheet location. The on-page symbols are different so they can be distinguished. Also, we usually only use on-page symbols to avoid having a lot of wires run from one side of the page to the other.

Another thing that I learned at my current job is the benifit of labeling resistors with "value, power, tolerance" and capacitors with "value, voltage, tolerance"

 

GSKrasle
User Rank
Freelancer
Re: Paper models?
GSKrasle   6/5/2014 2:29:47 PM
NO RATINGS
I'm old enough to remember tape and film, big photoreducers, diazo machines and the like. I cut my teeth on early versions of PCAD and PADS, and saw many examples of very very bad use of tools. I have strong opinions, some that people will disagree with, but if you think about it, the Library is the most critical part of the CAD system. It behooves us to make sure that all library parts are _perfect_ and that conventions are clear, consistent, and enforced. A "Librarian" with approval/editorial authority (or complete authority) is one way to do this.

Altium has a serious weakness in its "free-form" component attribute system: I have accumulated component definitions from several sources; fields "MFRPN," "MfrPN," "Mfr_PN," "Mfr#," etc., and "W," "Rating," "Power," "Altitude" (?????!), and many others appear. It's a mess, and would not have been if a canonical set of basic attributes had been defined from the outset.

One set of conventions will probably never be appropriate for all users, but there is one choice that divides the world: Are resistors defined with a manufacturer's/supplier's/stock part number for each value+rating, or are they simply put-down on the design and then their values defined, so that "jelly-bean" parts might be ordered? Highly-regulated industries are forced into the former convention, but low-cost products are served by the latter. In the former case, choice of already-defined parts is often economically advantageous, in the latter, less-so.

Anyway, hearkening back to the PADS paradigm (excepting some of the as-supplied conventions), here are some things, great and small, that I like and dislike:

1:

Most parts should be defined to have at least two "gates." For Logic, this can be one "Power Block" for supplies and others for logic functions, which might be of several types: a control block and several individual channels (think 74245). For a transformer, each winding is a gate, and for resistor networks (remember those?), a "Common Terminal" gate with one pin, and several one-pin "Resistors." 'Same for connectors. This convention allows for the power sections and bypass capacitors to be gathered appropriately (as mentioned below), and allows flipping/mirroring to optimize the "flow" of the schematic. Normally/often the individual "gates" of the design will be shown contiguously, like a single symbol, but they don't HAVE to be.

Defining "hidden" power pins assigned to nodes "3P3V" and "Gnd" or any hidden pins at all, is wrongheaded. Parts defined as monolithic blocks, so wires have to swirl and cross, or you are forced into just labeling pins with textual node names are lazy-now, painful-later.

2:

Sloppy flow is bad work. The only place I have found for not orthogonal wires on a schematic is S-R flip-flops (exception for late-breaking mods). L => R, T => B, if you please!

3:

Cross-References! Make Off-Page symbols with "pointed edges" to show signal-flow, and node-names, and a list of sheets and coordinates of connected off-page symbols. And please please be careful or prohibit "off pages" that go elsewhere on the same sheet, as that invites missing them and ensuing confusion. Wires are much more understandable than textual references! Also show the node-names in several places next to the wires, for understandability. "Schematics" where many/most/all connexions are made through node-names are not even as good as lists!

4:

Meaningful part symbols:

I like a basic representation of the internal function (rather than a blank rectangle) to facilitate understanding and debug. Keep 91-1984/60617-12 away from me: if I only have a daisy-wheel or Selectric, I don't need to print it. Symbols are like letters. They should be as large as necessary, but not larger.

If it can be determined, make a distinction between "Not Internally Connected" ("NIC") and "Make No Connexion" ("MNC"). (MNC pins may include ones reserved for future functions.) It is often useful to ground genuine NICs to facilitate "blue-wires" and logic-analyzer, emulator, etc. grounding, and even heat-sinking.

It's also nice to have alternate (DeMorgan) symbols.

5:

"Redundant" information:

I like a sketch of the internal connectivity of a sensor or such: placed by the connector on the schematic, this is valuable in understanding the circuit's operation.

Likewise, I like to put package pinouts ("ebc") by the symbols to facilitate debug. I have been known to steal the pictures off D-K's site, label them (MfrPN, Pin1, etc.) and paste them.

Text notes wherever the operation is not obvious. Basic design equations.

I like to put some non-PCB parts on the schematic, even if they are just in the forms of stars or dots, to indicate heat-sinks, default jumper locations, etc.

Some regulatory schemes apparently prohibit the inclusion of non-essential information like these on schematics, and some "gatekeepers" enforce this rule, like pedantic grammar-school English instructors. I hate having to have the schematic, BOM and Datasheets all on my desk at the same time: it's too crowded. The documentation should include what you need to do your work, within reason, even if you have to use the label "Indication Only" a lot.

6:

Reference Designators can carry more useful information than simply uniqueness. "Rn" vs. "Cn" is an example. The order in which components were placed on the schematic is of course very interesting, but not very useful. Useful things can include numbering components in patterns when a functional block is repeated, having the numbering range reflect the functional block, or even having them reflect the coordinates of the component on the PCB. Why are so many people worried about gaps in the numerical list?

antedeluvian
User Rank
Blogger
Re: Paper models?
antedeluvian   6/5/2014 9:21:19 AM
NO RATINGS
Daedus

We use DxDesigner which allows sheet numbers to be entered as attributes.  This has the benefit that it can be updated automatically (should updates require insertion of new sheets)

PCAD had this, but Altium (the "upgrade" path) does not.

and that it can also be used with off sheet connector icons to add the reference link mentioned in the article.  Although it won't add the alphanumeric block reference, it will add the sheet number(s).  It also allows large fpga parts to be split among banks so we can break the symbols down to whatever size lends itself to better schematics.


Now this is nice. I took a quick look at DX Designer which now appears to be part of Mentor Graphics. I can find no mention of cost or download, although I am often flumoxxed by web screens. I did find a free download on Google, but there is little documenation if this is legal.

 

<<   <   Page 3 / 6   >   >>
Top Comments of the Week
August Cartoon Caption Winner!
August Cartoon Caption Winner!
"All the King's horses and all the KIng's men gave up on Humpty, so they handed the problem off to Engineering."
5 comments
Like Us on Facebook

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Times on Twitter
EE Times Twitter Feed
Radio
LATEST ARCHIVED BROADCAST
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.
Flash Poll