Many parts from TI (and other chip manufacturers) have multiple packages for the same component. What is the best way to use the same symbol with multiple packages, when the pin numbers are different for each package?
Like it or not, some variant of this approach is probably the most efficient (and least error-prone) approach in the long run. Yes, it can lead to libraries that are large and unwieldy, with dozens of seemingly indistinguishable components. It’s possible that a particular component, in a specific package, may still have several footprint options based on how the board will be assembled (hand-soldered versus automated assembly on two or more production lines, each with its own requirements for pick-and-place courtyards, etc.)
It also clutters and disrupts the creative design process as practiced by many of us.
(Rather than try to give an elegant explanation of this point, just imagine a circuit designer sitting down to sketch out the first- or second-pass of a schematic and muttering to himself, “What’s all this alphabet soup dribbling off the part numbers? All I know is we’ll use an opamp here. Probably not a FET-input chip but I’m not sure of that, either. No, I don’t know the required precision yet. Temperature rating won’t be definite until the customer decides where he wants to use this product… Ya wanna know single, or dual, or quad? How the heck can I know? Depends on how many amps eve ntually get parked on this board!!”.)
I can understand the justification for not pre-assigning footprints to commodity passive parts. For example, I don’t think it’s unreasonable to wait until you start layout to decide whether a resistor will be an 0805 or 0603 package.
Well, be honest - Do you REALLY remember the differences between all those suffixes like “UA” and “AU” and “UAE”? The OPA551 Data Sheet ( http://www.ti.com/lit/ds/symlink/opa551.pdf ) lists ELEVEN (!!) variants of that chip alone. And since T.I. swallowed Burr-Brown and Nat’l Semi and several smaller semiconductor companies, is there such a thing as a uniform “TI part naming scheme”? When you consider another dozen or so major semiconductor companies it’s quickly obvious that there are no standards for these cryptic suffixes.
OK, the “Description” field is certainly helpful if you use it intelligently and consistently. The process of sorting through the Descriptions is still tedious clunky.
I think Andy_P is only saying to use different schematic symbols when pin numbering changes. The 10 different part numbers you get for something like a TL071 SOIC-8 due to temperature range, precision and taping options only affect the BOM, not the schematic symbol or footprint
If the pin numbers and functions are the same and the symbol (can be) the same, there is no problem to copy the symbol to a new one and make a different choice for the footprint in your parts library.
If you then decide - during design - that you want the other option, just delete the part you don’t want and put in the other. Only thing that is important there to not to forget to do manually is to keep the REF ID the same. I don’t see an option/menu for that function yet - swapping parts in EEschema (or I’m too dumb).
Oh. and My libs contain parts - not just symbols - where the name has got the housing included:
Thank you all for your responses. It’s interesting to see how others use the tool.
I recently started migrating from Eagle CAD to KiCad after about 10 years of professional and hobby use. The biggest change for me has been the part library. I really liked having the schematic symbol and footprint tied together at the library level in Eagle. It will take some getting used to KiCad’s approach, especially in cases similar to what I mentioned in my original post.
Thanks again for all of the advice. I look forward to KiCad’s bright future!