The new version of KiCAD is a fact.

[c4757p]

This may seem cumbersome, but it's actually quite nice because it makes it easy to have multiple layouts for the same schematic in one project.

Not anymore.

[RJFreeman]

You need to assign footprints, then generate a netlist.  The netlist is imported into the layout editor.

This may seem cumbersome, but it's actually quite nice because it makes it easy to have multiple layouts for the same schematic in one project.

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.
Whereas to be honest, that was what did annoy me about Eagles when I chose a component, I had to decide on the footprint, so for example I place an electrolytic capacitor in a schematic, now I have to stop and look up what size, and dimensions the capacitor I want to use is, check my calculations to cover voltage etc.

[Wilksey]

If you look in the download folder under stable, there is one that says "almost there" 4.0.0, this binary was released recently, I suspect it isn't fully packaged yet, just tested and committed as the "final" release.

The PCB editor will show a blank screen if you don't import the netlist, this part is not as seamless as EAGLE, and i'm sure this will be addressed in future releases as the components become more integrated and share data, for now, a few extra clicks of the mouse doesn't really make much difference (to me anyway)

Certainly cannot complain for the price!

[nctnico]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right. Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

[firewalker]

They aren't even separate anymore, though. That went away with the _last_ release - they sit in the same process now, so they can easily pass each other data. They just... don't.

A was referring to the process the user must take. Not the computing term. 

Alexander.

[tggzzz]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right. Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

Everybody I have ever seen has been sufficiently pessimistic to have a different database/library for each design, and to copy everything relevant into that "design local" database. Makes it easier to archive and hand over the design, and avoids the problems of subsequent changes to a component affecting previous designs.

None of that is specific to any schematic/layout tool, of course.

[Bassman59]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right.

Kicad for quite a long time has allowed you to embed a footprint reference into your schematic library symbols. No need for the error-prone CvPCB process. All of the symbols in my schematic libraries have footprints set in this manner.

Yes, the schematic libraries are separate from the footprint libraries.  While the devs could possibly cook up something like Altium's integrated libraries, I am not sure that there's a benefit to doing so.

Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

Kicad does let you generate a BOM with one click. OK, two clicks. You end up with a CSV file, which you can import into whatever you like.

[Bassman59]

Last time I tried Kicad I completed the schematic, switched to the layout editor and got an empty screen. This is where I went back to Eagle. Is this still the case? I would like eventually to switch to Kicad.

You need to export the netlist from the schematic, and then import it into the PCB editor.

[nctnico]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right. Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

Everybody I have ever seen has been sufficiently pessimistic to have a different database/library for each design, and to copy everything relevant into that "design local" database. Makes it easier to archive and hand over the design, and avoids the problems of subsequent changes to a component affecting previous designs.

You don't quite understand how it works. Search for how Orcad Capture CIS works and you'll see this workflow makes a lot of sense. Orcad allows to select which fields (part numbers, manufacturer, etc) are copied into the schematic from the database (which in my case is an MS Access database) so it can be handed over without needing the component database.

I have written it before but I will repeat myself: many low end CAD systems don't seem to understand that a component consists of a symbol, footprint and several other fields (properties if you like). You can't get away with just linking a footprint to a symbol because that just doesn't make it a component! For me it is necessary to have a database which has the components which then can be selected by the schematic package. No way I am going back to putting BOMs together manually or discover I choose the wrong footprint for a component I might have used many times before.

BTW: I don't mind having to create a netlist first and then read it into the PCB package. The ancient Orcad package I'm using works that way as well.

[RJFreeman]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right. Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

I disagree, I find the create the component once work flow cumbersome and unwieldy, so there you are deciding what component you will use, and then you also have to finalise power rating, package etc?
so you guess you need a 1/4 Watt resistor, sketch up the schematic, do the sums and then decide a 5W fits the bill?
or you have to decide on one package, then get to thinking about PCB layout and decide another package would work better?
I far prefer the work flow of:
1) work out schematic, get the circuit right
2) workout layout and component size
3)  design PCB
- and besides Kicad does do a BOM, pretty well in one click, once you have designed the circuit.
i find I get more errors and end up stuffing about way more, If I have to decide on component and footprint at the schematic stage, so I do have to strongly disagree, the Eagles work flow is far more prone to errors, and way cumbersome than doing the design in a more logical order.

[RJFreeman]

Kicad for quite a long time has allowed you to embed a footprint reference into your schematic library symbols. No need for the error-prone CvPCB process. All of the symbols in my schematic libraries have footprints set in this manner.

Yes, the schematic libraries are separate from the footprint libraries.  While the devs could possibly cook up something like Altium's integrated libraries, I am not sure that there's a benefit to doing so.

No don't do it - the schematic, then PCB design work flow works much better for me, and makes much more logical sense.

[Bassman59]

I disagree, I find the create the component once work flow cumbersome and unwieldy, so there you are deciding what component you will use, and then you also have to finalise power rating, package etc?
so you guess you need a 1/4 Watt resistor, sketch up the schematic, do the sums and then decide a 5W fits the bill?

I would imagine that most experienced engineers know the power requirements for resistors by the time they start capturing the schematic. I'll do a sketch in the notebook and work all of that out, before I even open the CAD package.

Same thing with op-amps. Same thing with voltage regulators. You pretty much know what you're going to use before you start the schematic-drawing program.

or you have to decide on one package, then get to thinking about PCB layout and decide another package would work better?

Well, that happens. Say there's a mechanical constraint that requires a different-size connector, or maybe the purchasing people tell you that you can get a part in package A a lot easier than package B.

But here's the thing: LM317 in a TO-220 has a different manufacturer part number from LM317 in SOT-223. So, thinking about how you'd change from the TO-220 to the SOT-223, isn't it easier to have both LM317T and LM317EMP in your library? Each component in the library knows the correct footprint AND has a part number that the purchasing people can use to buy the correct parts. Changing from one to the other, at the design level, is a simple matter of deleting the LM317T from the schematic and replacing it with LM317EMP. 

Then you regenerate the netlist (in Altium, you recompile the design and then update the board, in Kicad, like the old Accel, you generate a new netlist and import it into the PCB). You then generate the BOM from the schematic. No hand-fiddling is necessary, and stupid mistakes are eliminated (parts in the library are vetted, right?).

i find I get more errors and end up stuffing about way more, If I have to decide on component and footprint at the schematic stage, so I do have to strongly disagree

How can you get "more errors" if you place components on a schematic, instead of symbols which have no part number or footprint information?

I honestly don't understand.

[donotdespisethesnake]

There are clearly two schools of thought regarding component workflow, and I would respect the fact different engineers have different preferences. Ideally, the user should be able to choose the workflow they prefer.

Kicad could support both I think, if there were "generic" libraries with unspecified footprints, and fully associated libraries.

[c4757p]

KiCad definitely supports both workflows - I use Bassman59's workflow. I just didn't comment on that because I know it always turns into a flamewar

[tggzzz]

KiCad definitely supports both workflows - I use Bassman59's workflow. I just didn't comment on that because I know it always turns into a flamewar

Probably a wise decision. Judging by this post of bassman59
https://www.eevblog.com/forum/microcontrollers/beginner-fpga-dev-board/msg811642/#msg811642
he does appear to have a tendency of knowing The Single True Way^TM of doing things.

[bson]

a component consists of a symbol, footprint and several other fields (properties if you like).

Actually, in many cases there's multiple footprints.  Not only this, but the different footprints (e.g. soic vs ssop) have different pinouts and pin counts.  While you can pick one up front during the capture ideally you want to wait to the layout phase to determine which package makes most sense.  At least where you don't have a hard need or requirement for one or the other.

[Rasz]

usability of ~2005 Gimp, whats not to like?

[nctnico]

I actually prefer this work flow, where I can focus on getting the circuit designed first, then work out the details required to decide on the footprints, such as power ratings of components, voltage ratings for capacitors etc etc.

This is a very error prone and cumbersome workflow. The CAD package I'm using has a database which connects a symbol, footprint, part numbers, etc together. Creating a new part in the database is a one time job. When I draw a schematic I pick components from that database so I know the footprint is right, the part numbers are right. Generating a bill of materials is just a matter of one click. It would be really nice if Kicad (eventually) has something similar.

I disagree, I find the create the component once work flow cumbersome and unwieldy, so there you are deciding what component you will use, and then you also have to finalise power rating, package etc?
so you guess you need a 1/4 Watt resistor, sketch up the schematic, do the sums and then decide a 5W fits the bill?

I click 'link database part' in Orcad and all the information for the part gets changed immediately without needing to remember I decided to use a 5W resistor instead of 1/4W. Circuits quickly consist of 100s of components so you can't just remember every one of them. Recently I cleaned a schematic in which there where several different voltage rating 1uf capacitors in 0603 size. Orcad has a part manager which lists all the components so changing many of them in one go is just a matter of selecting them and picking another part (in this case a 1uf with a higher voltage rating) from the database. Automating component management really makes life easier and results in less errors. Errors in components can lead a lot of time lost. A couple of years ago I was involved in a project where someone else designed the hardware. He used a manual design flow like you described. The design was for a battery powered device with a couple of hundred SMT components on the board. It turned out the device kept drawing 30mA in standby which was much more than was calculated. I think he spend at least a week trying to find the cause and it turned out he made a mistake in the BOM.

Besides that I find it odd to start entering a schematic into a CAD package before knowing which components are required. For a project I'm currently working on I already have a dummy PCB layout to try and fit the large components but no schematic at all. I do this a lot for space constraint designs: get a PCB mockup first and then work out the (final) circuitry.

[Bassman59]

KiCad definitely supports both workflows - I use Bassman59's workflow. I just didn't comment on that because I know it always turns into a flamewar

Probably a wise decision. Judging by this post of bassman59
https://www.eevblog.com/forum/microcontrollers/beginner-fpga-dev-board/msg811642/#msg811642
he does appear to have a tendency of knowing The Single True Way^TM of doing things.

Just the Single True Way of doing things that seem to used at most employers with thought-through workflows.

And yeah, two-process state machines suck balls.

[Bassman59]

a component consists of a symbol, footprint and several other fields (properties if you like).

Actually, in many cases there's multiple footprints.  Not only this, but the different footprints (e.g. soic vs ssop) have different pinouts and pin counts.  While you can pick one up front during the capture ideally you want to wait to the layout phase to determine which package makes most sense.  At least where you don't have a hard need or requirement for one or the other.

Right, often you'll wait until layout phase to decide which footprint makes the most sense, and maybe your purchasing person says that one package is readily available and the other is not.

But the point we're making is that it's not difficult to have both LM317T and LM317EMP in the library, and it is dead simple to delete the LM317EMP from the schematic and place the LM317T in its place. And then the component will have the part number for the BOM so purchasing can buy the right part and the kit people can pull the right parts from the bin.

If you just place generic LM317 on the schematic, then at layout time you have to choose the footprint AND go and make sure that you update the BOM so you buy the right parts. The "wait until the last minute" method is error prone.

[donotdespisethesnake]

Bassman59 says, "There is One True Way".

Bassman59 exemplifies the "God" approach to quality.

"There is one person in the company who is Never Wrong. So if we all do what He says, there shall be perfect quality. By the way, that person is me. Kneel before me, inferior beings!".

In the real world, there are alternative workflows and quality processes that are equally effective. You know, just in case God/Bassman59 is not one of the employees.

[bson]

But here's the thing: LM317 in a TO-220 has a different manufacturer part number from LM317 in SOT-223. So, thinking about how you'd change from the TO-220 to the SOT-223, isn't it easier to have both LM317T and LM317EMP in your library? Each component in the library knows the correct footprint AND has a part number that the purchasing people can use to buy the correct parts. Changing from one to the other, at the design level, is a simple matter of deleting the LM317T from the schematic and replacing it with LM317EMP. 

Here's a better, yet still trivial example.  There are far more complex chips that have multiple packages with different pinouts.



On the soic, the inputs are at the top and the outputs at the bottom.  On the ssop the outputs are on the right, inputs on the left and there are four nc's.  (Or, I should say uC-facing vs external-facing.)  You really want the freedom to pick one or the other during layout, because you won't know until then if there's any benefit to using the soic over the ssop.  Without context switching to the schematic.  No tool handles this adequately, but KiCAD and its manual netlist management at least permits multiple exploratory layouts: I can run two copies of pcbnew on Linux side by side, making two exploratory layouts in parallel, for the same schematic.  And it's not just pinouts, I can experimentally design for 2 and 4 layer and then see if the benefit of the latter outweighs the cost, or conversely use it to prove my assumptions.  All I need to do is generate appropriately named netlists or subdirectories for the different variants, and go to town.  In the documentation I then do a writeup of the different variants and why one was chosen over the other (more quantitatively, as in halving the number of vias, or made a 2-layer board possible within the dimensional constraints).

[kripton2035]

downloaded last kicad for os x today.
start a new pcb
program unexpectingly quit ...
so far I'm not using it further !

[RJFreeman]

You really want the freedom to pick one or the other during layout, because you won't know until then if there's any benefit to using the soic over the ssop.  Without context switching to the schematic.  No tool handles this adequately, but KiCAD and its manual netlist management at least permits multiple exploratory layouts: I can run two copies of pcbnew on Linux side by side, making two exploratory layouts in parallel,

But apparently (at least as far as I can tell from the comments in this thread) everyone else decides on the final board layout, BOM, and component choice, before they even design the circuit

[nctnico]

You really want the freedom to pick one or the other during layout, because you won't know until then if there's any benefit to using the soic over the ssop.  Without context switching to the schematic.  No tool handles this adequately, but KiCAD and its manual netlist management at least permits multiple exploratory layouts: I can run two copies of pcbnew on Linux side by side, making two exploratory layouts in parallel,

But apparently (at least as far as I can tell from the comments in this thread) everyone else decides on the final board layout, BOM, and component choice, before they even design the circuit

No, 'we' do that before entering the schematic into the CAD package. Going all the way from a schematic to a PCB layout only to find out the circuit doesn't fit or benefits from using different parts is just a waste of time. In my projects I often have a rough PCB layout before starting the schematic. This way I can optimise the schematic for the PCB layout from the very start and not as an afterthought. Sometimes limited PCB space drives other features of a design as well so it is important to start with the size constraints for the PCB layout and go from there.