{this message bounced first time, outgoing mail server couldn't find "techservinc.com"....)

At 06:10 AM 12/10/2003, Laurie Biddulph wrote:
I hate having power supply pins as part of schematic component symbols (especially opamps and logic gate chips). I prefer to create an additional `component part' in the chip purely for the power supply pins. This makes it easier to assign decoupling components to the chip as well as reduce clutter in the main part of the schematic.

This is a very legitimate way of dealing with the problem, as is having the power pins be part of the symbol. Hidden pins have restricted application, some say that they should never be used, but that goes too far. If you have a digital design with standard logic, hiding the power pins may be acceptable.


However, if a technician is going to have any difficulty later figuring out which pin on a part is, for example, ground, it is better to be explicit.

Making symbols with power pins as a separate part of the symbol, while it is a little more complex -- in creating the symbols -- is really the best of both worlds. All the power parts can be placed on a page -- or part of a schematic page -- which shows power nets and bypass cap allocations. This leaves the rest of the schematic for signal flow and logic, and not having to deal with power connections and bypass on those pages saves both time and space, and results in a schematic that is easier to read. The only negative I can think of is that in a split-supply design the power assignments are not necessarily on the same page so an error in assignment might be less obvious.

I consider the improvement in general readability to outweight that; it just requires a little more caution, since, so far, there is no ERC for this.

(If component classes could be set up in schematic and assigned power supply classes, ERC would be possible, where a component was assigned the incorrect power supply, i.e., an analog part gets a digital supply. This, by the way, is a very common error in designs we receive as a service bureau, and we do try to notice it and query the engineer.)

Problem is Protel 99 doesn't like annotating these as it treats the power part as a real part and really gets messed up. I believe Protel DXP lets you assign the power supply pins to Part 0 and so, presumably, gets round the problem.

I haven't looked into that aspect of DXP yet. The problem in P99 (and earlier) is only with automatic annotation. I think one could get around the problem by having two libraries: one would be the components with no power pins (or with them as part of the main symbols), the other would have the same parts with power pins removed. The schematic would be drawn, at first, with the parts from the first library, and annotated. Then the symbols would all be updated from the second library, and then the power page would be added to the schematic. There are some caveats with updating symbols, but I'm a bit rusty on that topic....


Beyond that, manually assigning parts is normally not such a huge task. If there is a way in DXP to exclude a symbol part from the autoannotation task, this would indeed be an improvement.

But there is usually manual attention needed to annotation, to cluster logic functions, for example, on the same device so that signals remain local instead of running across the board and back just to run through an inverter. I'll often allow a few sections to be unused, more than the absolute minimum, just to keep signals together. Logic functions are generally cheap.

Hiding power pins is bad news especially if you use different power rails from, say, VCC and GND which are the common defaults for logic chips and so if you forget to unhide them you end up with a power net not going anywhere near your real power supply.

Protel does not handle this probem as well as DOS Tango did. Tango allowed sheet-wise net renaming. So you could place a power object on a sheet and a short piece of wire with a net name. This would rename one of them to the other (I forget which was which), allowing you to connect, for example, VCC to +5V. Whatever was VCC on that sheet, as a power object or hidden power pin, was reassigned to +5V. This had no effect on other sheets, thus allowing multiple power supplies with the same hidden pins. It was explicit and easy to understand.


Is there any recommended method in Protel 99 of handling the power pins on logic chips similar to my first method above that Protel 99 can handle comfortably?

Protel has no problem dealing with separate power sections, *except* for automatic annotation. It is a subset of the larger problem, which is that automatic annotation is a limited tool and often results in undesired assignments.


However, I think there might be a way, I don't have time to test it at the moment. The Annotation tool in 99SE has a number of controls. First, on the basic page, you can set it to ignore selected parts, and you can use part field match to control grouping (i.e., within multipart symbols, I think). On the Advanced page, you can include or exclude schematic sheets from the Annotation process.

*So*, I'd try this:

Having placed all my unannotated parts, excluding power sections, I'd run autoannotate (Tools/Annotate). I would not allow it to assign multipart sections, so every section would have its own unique number. (Uncheck all options in "Group Parts Together If Match By"). If I have placed power sections, I'd exclude them from annotation by one of the two methods provided. (This process assumes that multipart symbols being annotated have identical sections except for pin numbers and the power section, and that section 1 is the basic functional section excluding power pins).

Then I would pagewise renumber parts by selecting the appropriate number of parts, such as up to six inverters for a hex inverter package, and renumbering that part to one of the numbers already used for the six I have chosen. This will be a unique number.

I can use the annotate tool to auto-assign the sections by inverting selection (so that all parts but the ones I'm working with are selected and therefore ignored) and letting the tool group the parts together and assign sections. But it might be nearly as fast to manually assign the sections, something that is a good idea anyway if, for example, you are running a bus through an octal inverter and you want the pins to be logically sequenced, which can make for much better routing.

After this, I'd add or edit the power page so that there is the appropriate power section for each used part, having the same reference designator. If I have both pages open, I can do this edit as part of the original edit, i.e., I'd select, in the example given, the six inverter sections plus the power section, and globally edit them, keyed on selection, to one of the reference designators.

When I'm done, I'd have a schematic with unique and appropriate reference designators and sections on every page. Now, if I want to sequence the designators, I'd generate a list of reference designators by any one of several methods. I'd use a word-processor and spreadsheet, as necessary, to put those numbers in alphanumeric sequence, and I'd generate compacted set of reference designators to make a list that had, in place of U1, U5, and U9, the designators U1, U2, U3. I'd then run Back Annotate with the file I created. (I'd have to look at the manual to get the exact format, but it might just be:

U1 U1
U5 U2
U9 U3

I'd be done. It might seem like there is extra work in grouping the parts, but, in fact, that work would be reduced to what you ought to do anyway as a good designer, i.e., control section assignments to keep the physical layout simple.

By the way, it is my practice to always place unused sections on the schematic, and, as appropriate, to tie inputs directly (usually) or through a resistor (depending on mfr. recommendation) to a power supply. But I might do this at the end of a design, I don't want those inputs to be, for example, connected to a power plane. Instead, they'll be wired on the surface and thus easily available for blue-wire revisions. The unused sections might be put on the power supply sheet or area.

(Tango had an unused section report, something that has been overlooked in Protel, as far as I know; you can get a list of reference designations and part numbers -- i.e., section numbers -- as spreadsheet output, but there is no quick way from that list to distinguish, for example, a part which is not multipart from the first section of a multipart component, both of them will have 1 as the part number. I ran this unused section report with every design, to ensure that all part sections had been placed.)




* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * To post a message: mailto:[EMAIL PROTECTED] * * To leave this list visit: * http://www.techservinc.com/protelusers/leave.html * * Contact the list manager: * mailto:[EMAIL PROTECTED] * * Forum Guidelines Rules: * http://www.techservinc.com/protelusers/forumrules.html * * Browse or Search previous postings: * http://www.mail-archive.com/[EMAIL PROTECTED] * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to