You could add that to https://bugs.launchpad.net/kicad/+bug/593944 (or at least ask if this could be done with pin swapping which has been planned for 6.0).
I’ve often wondered which end of a “through hole” resistor is “pin 1” (and do they need to be fitted the correct way round for them to work ? )
Sorry, just me still being a bit of a dinosaur. A large number of my designs are very simple, with just a few components, and the circuit is in my head. I just want a single-sided PCB and can design it qicker than I can draw the schematic.
Perhaps I should look for a DOS 6.22 PC and go back to my old version of BoardMaker ?
Seriously though, I do like Kicad - it seems to be the best of the bunch around at the moment. I just wish someone would allow me to design my PCB without resorting to inputting a schematic for which I have no future use.
At the moment I am designing a PCB which will be used for THREE different purposes depending on which parts are fitted (each with its own schematic) - just how do you get around that without laying it out “on the fly” ?
DRC needs to somehow know what should be connected to what. You can work without DRC by switching to highlight collision mode and turning on “allow DRC violations”. This is the easiest option to work without a schematic but well you loose a lot with it.
The way to tell DRC what is connected to what is (right now) done either via the schematic or by some other means that results in a KiCad netlist. Your best option right now is therefore the use of wirelt (linked above)
Fully integrating this into kicad is non trivial as it kind of runs counter to the core workflow. It would need special handling within pcb_new and its file format in a way to be still compatible with using a schematic. (This is one of the core requirements of any new feature addition.) Part of the required backend feature set might come with the pinswap feature that is intended for version 6.
Also consider looking into how much time you really need to draw up a schematic. My guess is that the time needed is much lower than you expect.
The same way as you would on a pcb only implementation. You draw up one single schematic with all the connections in the same manner as you want them on the pcb. The added bonus is that you can include fields for BOM extraction that can tell you which part to use for which version.
In other words: worst case you have a schematic that looks just like the pcb. Best case you have a nicely organized documentation of your pcb that makes it possible for anybody to understand what your system does.
All of my PCBs are now for personal use one-offs so nobody else needs to know what they do. My new “triple” PCB will output either 0-24V analogue or 0-15V PWM both from an unregulated DC power supply or 0-12V PWM from battery power so I need to vary the input supply or the output circuit or both - the PIC in the middle is the same for all three versions. The PCB is still single sided and roughly 75 x 40mm. The problem is that some pads for wire-ended components are shared by different components going to different locations in different versions and some SMD parts are different values. I could have designed the PCB in the time it has taken me to describe it but non-familiarity with schematic drawing would take me for ever.
From those of us that have lived long enough, which is sometimes a couple months, to look at a project and wonder “WTF?”
Seriously, the schematic is your non-ambiguous notes.
It sounds like you are describing a pcb with multiple malfunctions.
Seriously I do not not fully understand what you are describing. IMHO pin numbers on common 2W resistors and most nonpolarized capacitors seems to be bogus. Exception which proves the rule…do you remember “outer foil” on some wound film capacitors?
But I think that one thing I have done is to draw a schematic which includes all of the component locations, even though you would never use all of them at once. There is also the possibility of two parts sharing a pin or hole (or which might otherwise physically interfere) except that you would never use both at the same time.
Thank you eelik.
I had a look at https://bugs.launchpad.net/kicad/+bug/593944. It looks like my thinking is pretty much covered (including your recent addition) and I do not have much new to add?
I looked at your link and read it with interest about being able to connect a non-polarised two pin item either way round, basically ignoring pin numbers. If I read the article correctly, it would appear that the main purpose is to allow a neat finished PCB with all of the annotation facing the same way. I did comment earlier that I also etch my own boards and being for personal use I don’t (can’t) screen print them or solder resist them so the need to have the printing “the correct way round” doesn’t apply. Yes, I do remember “outer foil” on capacitors - it was important when used in RF environments for shielding purposes. I also must point out that I hand assemble and hand solder my PCBs including SOIC and smaller “spider” ICs so I can fit my components for my convenience, through hole or SMD.
There is one other subtle and rather esoteric consideration, in addition to orientation of annotations on the silkscreen.
Consider a library that strictly adheres to the standard where “Pin 1” on any footprint is implemented as a square or rectangular pad, while the other pins are circular or oval pads. It may be desirable to have all of the square pads “facing the same way” - e.g., the square pads of all passive parts are closer to the top of the board, or toward the right-hand edge, etc. Admittedly, approximately 90% of this is for visual aesthetics and has absolutely no effect on the electrical performance of the board. However, there are occasions where it’s desirable to have a rounded pad at a particular location because it is easier to route a trace around a rounded pad, than a square pad.
Silkscreen annotation is independent of the footprint orientation. I think what @Videostar meant was that the text printed onto the resistor itself will be aligned that way as it will depend on the footprint orientation.
If the workflow works for @Videostar (and it seems it has for some of our life times); I am happy that KiCAD fullfill even his way, as it does any other ways.
If I understand him correctly, when he is laying a board with a bunch of components, he put his components as he sees fit and then, he connects them using tracks, if suddenly he notices that a track would be better placed somewhere else, he just deleted the track and connects it any other way. In this case if the schematic has the connections defined (track goes from U1:P1 to R2:P2) he would need to go to the schematic, change the connection as he just occurred to him that it would be better (U1:P1 R5:P1) otherwise PCBNew would not let him connect the pin (which mostly is ok to avoid mistakes. But now in his work flow).
I think everyone does this; I laid out a board with three banks of connectors and a hex inverter IC; while drawing the schematic I connected bank 1 to inverter 1, bank 2 to inverter 2, etc. When routing the board I realized this was making my job harder, so I changed the bank-to-inverter connections in the schematic so that the trace routing would be more logical.
This is the idea for pin and gate swapping. AFAIK pin and gate swapping is in development (or at least on the milestone list) for v6. But, I’m not sure if it is planned to implement PCBNew driven pin/gate swapping or if it will only be done in the schematic.
Not for a pick and place machine, which will carefully rotate a smd resistor to match the assembly file.
Also not for my QA inspector 40 years ago, who would have rejected a board because the THT resistor colour bands were the wrong way round
They were just to lazy to decode it bassakwards!
That is not a QC reject for flight hardware as far as I know.
I hope that everyone concerned is grinning (as I am) as they discuss this. I am a person who delights in designing something while ignoring cosmetics. But for functionality…I do like silkscreens to be placed in a way that makes them easily understandable.
But my reason for wanting to ignore resistor pin numbers is that I have sometimes connected the wrong end of a resistor momentarily during layout. If it costs me 10 seconds to fix it, then that is 10 seconds of the software telling me to do something silly and I have a philosophical objection to doing that. It just should not matter.
I am not so intimate with pick and place machines but I think that when the resistors are placed onto tape, they may be randomly disoriented end-end. What might make more difference is placement of some chip ceramic capacitors which are as tall as they are wide…now there are 8 positions in which it could reasonably be mounted onto its pads. Half of those options have the ceramic layers perpendicular to the pcb. That might actually affect series inductance slightly!
I think that some pick and place machines start with loose parts in a cup of some sort…disoriented randomly…but then we have to make sure that the pick and place head does not swap the ends. This seems like a lot of nonsense. I am shaking my head and grinning as I write this…no worries.
To tell you truth I don’t understand whay you need these three schematics at one PCB. I would just do three PCBs. PCB manufacturer we use counts some price for each PCB documentation and then price for dm². To pay once for documentation I would put them all at one PCB to cut them myself.
The use for this one PCB is worth inputting a schematic. It takes about 10…20% of whole time of PCB design (I don’t count time used for element selections/decisions typically made during schematic inputting).
Yeah, you are right, I do it like that also, but I believe Videostar (whom has said his job was drawing the pcb layout for many years) doen’t like that the software “forces” him to go back to the schematic to make such a change.
Earlier I did say that I will be making the PCBs myself (not using a manufacturer). The board will have a “break-off” section either end - one end will have a bridge diode and smoothing cap (only required for the AC supply input) and the other end will have an FET “H-bridge” output stage (only required on two versions). I am doing it that way to allow me to only draw the board once yet etch all three together. Finally I will break off the non-required sections during hand construction. The centre section with the PIC will have some pads which are only used in certain versions (so I need to add “blank” pads in some schematics ?).
You say the inputting of the schematic is only 10-20% of the time to design the PCB - I would struggle to input a simple scematic in under a day. My 3-way PCB took me less than 1 hour to draw but I already knew how I wanted it laid out and what connected to what, which I suppose is what a schematic is doing but I had already worked everything out in my head. Incidentally, I made the 3 PCBs over weekend and have started building the first one which doesn’t have any errors.