Alternatice UI

[PlainName]

because it makes the common 'place a resistor', 'place a capacitor' tasks fast and easy

I think the important thing there is how it makes it fast and easy. Consider someone used to, say, Kicad and not LTSpice - they will be placing things almost without thinking (perhaps - it seems to me the interface demands too much consideration, but go with it). You could then say Kicad has made things fast and easy, but without saying how and why it's not very meaningful.

There is also fast and easy for new, middling, and expert users, each being different. The new user will appreciate perhaps a popup list of components to pick from whereas the expert might prefer a magic keystroke. Again, it's not really enough to say that one is faster and/or easier without explaining why it is so.

[Cerebus]

Pictures of common components on the toolbar. LTSpice is so ubiquitous on here I'm surprised I need to point this out; my wrong assumption, but it still surprises me. 

[PlainName]

LTSpice is so ubiquitous on here I'm surprised I need to point this out;

I think the thrust of my comment was missed, then.

[Cerebus]

No, but if you've used both it's clear which will get the job done faster. If you haven't then sorry, but you need to actually try it. The alternative would be for me to spend a day writing up  a usability report on the 'challenge task'. Guess what, that's not happening.

[delfinom]

Rejoice, the nightly / v6 now has click on component anchors to immediately start drawing wires.

Click-drag on components is also partially there now by default.

[PlainName]

Well done 

[JohnG]

For what it's worth, I find it funny that people want a Windows experience and hold up LTSpice as an example .

John

[Warhawk]

For what it's worth, I find it funny that people want a Windows experience and hold up LTSpice as an example .

John

Haha, true 

[poeschlr]

Regarding the "give me a toolbar of common components" suggestion:

Well what is a common component? Spice has it easy. These are just the primitives of it. But what is common for a PCB design tool? I mean a resistor could be, but a resistor is represented by some symbol in some library. You might be satisfied with the symbol in the Device library that comes with kicad but somebody else might have their own symbol defined (for example they have symbols for every possible resistor that they have in stock. Fully filled out with order information and a footprint pre assigned. Somebody else might have a set of partially defined resistors in the lib like for example one for a 0805 imperial; 1% tolerance; 75V and one for 0603 imperial; 1% 50V ... And yet another person has the crazy idea of wanting all libraries local to the project including the symbol for a resistor)

And a similar problem exists for every component out there. What you define as common is not very likely common to somebody else. So if there would be a tool to easily add common components, then this tool must be highly configurable. This increases the implementation complexity quite considerably. And to be honest I am not even convinced that it will really save that much time for most users (in the grand scheme of things). After all drawing the schematic itself is one of the fastest tasks no matter how bad the editor is. The longest task of designing a project (at least for me) is deciding which exact components to use, especially as this in most cases means making decisions about what tradeoffs to make. Anything that would help with this part is therefore much more likely to save time than reducing adding a resistor down from lets say ["ctrl+a"->enter "R" into already highlighted filter field->click on resitor or just press "enter"] to [click on resitor in toolbar]. (and in current nightly you don't even need the ctrl key. Just use "a"->"r"->"enter")

[Cerebus]

Hmm. Someone who doesn't regard a resistor as a universal component, found in every non-trivial circuit design ever. A [different] symbol for every resistor in stock? I think someone doesn't understand the very essence of a schematic diagram. Confounding "symbol" and "complete component specification in the database" is an indication of some very off thinking.

In my mind when I draw a schematic I go "I need a resistor", then once I've got "a resistor", I go on to draw the next component. Only later do I go "that resistor needs to be 10K", and then "1/8W", and then "0802", and "1%". What I don't do is go "I want to draw a 10K 1/8W 1% 0802 resistor from Vishay" in one breathless thought the first time I encounter the need for a resistor there in the circuit. Ditto "NPN transistor" not "2N3055 in a TO-3(P) case from Motorola, high h_FE variant, packaged in tubes, in 50s". People should not be made to think in such concrete detailed terms while doing a relatively abstract task. We call it "capturing the schematic", not "filling in every detail but the physical layout and the tracks on the board".

I may know all that detailed stuff in advance, I probably do, but when drawing a schematic I don't want to be bothered with the details of each and every component, just the big picture. I'll do the details once I've got a whole to pin them on. You draw the schematic, then you specialise it as to component values, then as to other characteristics. You don't think about the individual footprints and the minutiae as you draw the whole. Just as when I wrote this screed - I picked the words, and I only went and added the formatting flourishes once there was a whole linguistic picture.

Just to get you thinking. Someone produces a schematic in KiCad. Someone else comes along and says "I need you to convert all the symbols in this schematic into IEC standard ones" (Which by the way is completely realistic. An American would probably use ANSI symbols in first instance, their new European customer might insist on IEC symbols.) How do they do that? Remember, the actual circuit hasn't changed one jot, neither has any component characteristic. All that needs to change is the symbolic representation of the components in visual form.

[julian1]

Hmm. Someone who doesn't regard a resistor as a universal component, found in every non-trivial circuit design ever. A [different] symbol for every resistor in stock? I think someone doesn't understand the very essence of a schematic diagram.

Except, go to wikipedia and lookup the article entry for a resistor. You will find there is no universal symbol for a resistor. Instead there are two distinct, and widely accepted symbol representations that are commonly used.

So a choice must be exposed somewhere, otherwise the end-user is unnecessarily constrained. Next you have a decision as to size - should a 5W current sense resistor receive the same visual emphasis in a schematic as a bias/pull-up. What if the resistor is part of a resistor array component. etc.

A workflow that works for me - is to pull my most common representations and choices into my own library. That has two advantages. Firstly I get the representations, designators and choice restrictions I/my org want and use most frequently. And secondly it provides a navigation shortcut for fast selection when populating items on a schematic.

I don't want to be bothered with the details of each and every component, just the big picture.

Following the same approach - Use or create generic symbol representations for popular components  -  eg. bipolars, fets, generic dual/quad op-amps and comparators, etc.

That way - you can complete a full PCB indicating only the necessary footprints - ie sot-23 versus dpak, soic versus dip etc, but without having to drill down and specify actual manufacturers parts or even values if you want.

I do this all the time. Leave the choice of current limiting resistor until later, when I actually pick a led and color with associated voltage drop. Or delay the op-amp choice until I have test data about performance. In most cases only the footprint matters.

That's a ton of design flexibility, while not particularly increasing gui/workflow complexity. 

[julian1]

Instead of some of the examples I gave, I should have used your example for a resistor.

The genericity / specificity tradeoffs that you prefer, are quite easy if you are willing to create your own library. And by that, I mean mostly just copy/paste from defaults - while restricting choice to preferred representations, and editing for size, designators etc for common components.

I have a generic resistor that I use everywhere - it gets the minimalistic designator "R" in my library as a default - so it's quick to find, and use.

It has the sawtooth view, because I hate the ugly square box, and small because I almost never want to give a resistor visual emphasis. But stylistically, these are personal preferences.  If I hand the project on to someone who wants something else, they can edit, and the changes will propagate across project schematics.

I do complete schematics using only this "R" symbol - without ever specifying actual values - resistance, tolerance, rating, manufacturer.  super generic.

I have also done complete pcbs with the only additional criteria being a smd footprint - all other component choices are delayed until I have the physical board in my hands and am ready to solder.

That is a high-abstraction workflow. And it is easily supported.

[Cerebus]

Hmm. Someone who doesn't regard a resistor as a universal component, found in every non-trivial circuit design ever. A [different] symbol for every resistor in stock? I think someone doesn't understand the very essence of a schematic diagram.

Except, go to wikipedia and lookup the article entry for a resistor. You will find there is no universal symbol for a resistor. Instead there are two distinct, and widely accepted symbol representations that are commonly used.

Yes, two, count them, two, but here are not different symbols for each and every possible value and type of resistor. Read things in context, please. Vis:

...but somebody else might have their own symbol defined (for example they have symbols for every possible resistor that they have in stock. Fully filled out with order information and a footprint pre assigned. Somebody else might have a set of partially defined resistors in the lib like for example one for a 0805 imperial; 1% tolerance; 75V and one for 0603 imperial; 1% 50V ...

A sane system is, click on resistor tool, slap down a resistor, default to whatever basic graphic is set globally {ANSI, IEC, Man-From-Mars}, worry about the details later. A non-sane scheme is start placing a component (either clicking on an op-amp symbol or hitting the 'a' key), and be faced with this choice:



and have to dig through 14,854 possible components to get to the humble omni-present resistor. What is quicker, making common components immediately accessible, or requiring this palaver for every component?

So a choice must be exposed somewhere, otherwise the end-user is unnecessarily constrained. Next you have a decision as to size - should a 5W current sense resistor receive the same visual emphasis in a schematic as a bias/pull-up.

Now you're just being silly. What proportion of schematics have you ever seen with that feature?

[julian1]

A sane system is, click on resistor tool, slap down a resistor, default to whatever basic graphic is set globally {ANSI, IEC, Man-From-Mars}, worry about the details later.

See my other comment, that I believe addresses exactly this point.

[Cerebus]

A sane system is, click on resistor tool, slap down a resistor, default to whatever basic graphic is set globally {ANSI, IEC, Man-From-Mars}, worry about the details later.

See my other comment, that I believe addresses exactly this point.

No it doesn't. It effectively says "To make this tool usable you must heavily customise it, don't expect it to be easy to use out of the box for the very basic task of drawing a schematic. Please learn it and then fix it for yourself". I'd hate to see the UI you'd come up with for a general purpose drawing tool like Illustrator, CorelDraw or InkScape. Presumably there would be a line tool and a graphic tool, and then if you wanted a simple circle or rectangle tool you'd have to customise the graphic tool (after leaning how to use it) and your excuse for this would be along the lines of "It can draw any graphic, and it isn't for us to look at what people draw and make any judgements about common graphics that they draw".

I despair at any possibility of getting this basic "usability" idea into the collective heads of the KiCad crowd.

[delfinom]

Hmm. Someone who doesn't regard a resistor as a universal component, found in every non-trivial circuit design ever. A [different] symbol for every resistor in stock? I think someone doesn't understand the very essence of a schematic diagram. Confounding "symbol" and "complete component specification in the database" is an indication of some very off thinking.

I think this is a view of schematic diagrams from 4 decades ago. Modern professional designs detail component specifications to the dot, that information is transfered to design review, quality control, manufacturing and other departments. You don't waste time placing "resistors", instead immediately open a company library that has previously used parts with all that data immediately attached to the symbol.

Just to get you thinking. Someone produces a schematic in KiCad. Someone else comes along and says "I need you to convert all the symbols in this schematic into IEC standard ones" (Which by the way is completely realistic. An American would probably use ANSI symbols in first instance, their new European customer might insist on IEC symbols.) How do they do that? Remember, the actual circuit hasn't changed one jot, neither has any component characteristic. All that needs to change is the symbolic representation of the components in visual form.

I would consider it poor engineering practice to make such a radical visual change to a schematic for simple updates. The symbol standard should have been defined at the start of the job. Changing it afterwards comes with the risk of stupid issues that shouldn't exist arising and requiring full requalification of a design.

[pointhi]

I despair at any possibility of getting this basic "usability" idea into the collective heads of the KiCad crowd.

Sorry to say that, but you fell into the standard UX pit of: my approach is the right one.

KiCad is not a drawing tool for schematics as you would draw them for a book. Nor is it a tool to draw schematics for simulation, where you do not care if you can order that part. KiCad is there to build pcbs, and its long term goal is professional usage. And thats where database part libraries reside (planned for KiCad v7). Or nowadays, atomic libraries, when you want to have fully specified library symbols in KiCad as required by many business.

Optimizing you usecase makes KiCad behave more like a toy for electronic beginners, but not for people who care about the parts they need to order in the future.

For your use-case: place a resistor once, and duplicate it when you need a second one. Should be fast enough and works for every part, not only the "standard" ones.

[pointhi]

There is an open issue now discussing the toolbar and the implications of it: https://gitlab.com/kicad/code/kicad/-/issues/6028

[Cerebus]

Hmm. Someone who doesn't regard a resistor as a universal component, found in every non-trivial circuit design ever. A [different] symbol for every resistor in stock? I think someone doesn't understand the very essence of a schematic diagram. Confounding "symbol" and "complete component specification in the database" is an indication of some very off thinking.

I think this is a view of schematic diagrams from 4 decades ago. Modern professional designs detail component specifications to the dot, that information is transfered to design review, quality control, manufacturing and other departments. You don't waste time placing "resistors", instead immediately open a company library that has previously used parts with all that data immediately attached to the symbol.

I am talking about the usability aspects of drawing the schematic, in a tool, not the whole production process. Are you really going to tell me that in drafting a schematic, you always specify each and every possible parameter of a component (unless a tool forces you to , like KiCad)? Are you so 'modern' that you never draw a schematic with a paper and pencil, and when you do, is it logical to draw the schematic as a whole and then annotate it, or completely annotate every component symbol as you draw it?

Just to get you thinking. Someone produces a schematic in KiCad. Someone else comes along and says "I need you to convert all the symbols in this schematic into IEC standard ones" (Which by the way is completely realistic. An American would probably use ANSI symbols in first instance, their new European customer might insist on IEC symbols.) How do they do that? Remember, the actual circuit hasn't changed one jot, neither has any component characteristic. All that needs to change is the symbolic representation of the components in visual form.

I would consider it poor engineering practice to make such a radical visual change to a schematic for simple updates. The symbol standard should have been defined at the start of the job. Changing it afterwards comes with the risk of stupid issues that shouldn't exist arising and requiring full requalification of a design.

Aside from the fact that in real life 'unreasonable' demands like this intervene (you would not turn down a £100 million order from some countries military who insist on a particular symbology because "Our processes don't allow us to change the symbol set on a drawing") this is merely to get the programmer in question thinking about the mechanics of this. This ought to be a drawing level setting of "use symbol set \$x\$" both for usability/functionality reasons and also because (for anyone who realises that the parts are going to end up in a database) it's the sane way to build the data structures. (SELECT symbol_drawing FROM standard_symbols WHERE basic_part_type = "R" AND symbol_set = "ANSI"; ).

Even though the whole 'requalifaction' thing is a red herring, it should not need to be said that a process that insists on a complete requalification of a design for the drawing equivalent of 'using a different font to print this out for customer X' is a broken process.

[Cerebus]

I despair at any possibility of getting this basic "usability" idea into the collective heads of the KiCad crowd.

Sorry to say that, but you fell into the standard UX pit of: my approach is the right one.

No I didn't. It's just convenient for you to say that so that you can continue thinking that your approach is the right one. Yes, it's hard to have your baby criticised, especially when someone says that your whole UI is riddled with usability problems and needs a radical overhaul.

KiCad is not a drawing tool for schematics as you would draw them for a book. Nor is it a tool to draw schematics for simulation, where you do not care if you can order that part. KiCad is there to build pcbs, and its long term goal is professional usage. And thats where database part libraries reside (planned for KiCad v7). Or nowadays, atomic libraries, when you want to have fully specified library symbols in KiCad as required by many business.

Optimizing you usecase makes KiCad behave more like a toy for electronic beginners, but not for people who care about the parts they need to order in the future.

There is no tension between "make the drawing bit easy" and "make it possible to fully specify components". To get the latter it is not necessary to make "drawing" an exercise in "selecting components from a database". You're thinking of the schematic capture tool as a "produce a netlist containing wires and references to a database of components" tool, which it is under the skin, but on that skin you've quite deliberately put a schematic drawing interface. If you're really so wedded to the idea that it's being a "select parts from a database" tool is overridingly important and the drawing bit is an annoying irrelevance, why bother with this drawing at all. Make 'em select all their components from the database and only then let them draw any wires or move components about. That would be ridiculous wouldn't it? Why so resistant to making the drawing easy with good usability? Looks to me that you can't see the wood for the trees.

For your use-case: place a resistor once, and duplicate it when you need a second one. Should be fast enough and works for every part, not only the "standard" ones.

For a start it's not a use-case, it's a UI usability case, inability to tell the two apart is telling; as is dropping into business analyst-ese with phrases like "use case".

Every time, every bloody time, someone says to fanboy contributors of program x "It would be better if the UI did this like that" the fanboys say "Oh, you can do that [the functional equivalent] by <series of moves in existing UI>" thereby completely missing the point. Just because something is functionally possible doesn't mean that it satisfies usability criteria.

I've been a programmer for 45 years, since the bloody punch card era. If I was as resistant to improving usability as you guys seem to be I'd still be insisting that my customers used punch cards. "No, look, you CAN change the customer field. Just pull the 3rd card out of the batch, walk over to the punch/duplicator, hit COPY for the first twenty columns, then type the new customer name, and then keep hitting copy until the end of the card. Then put the card back into the stack. See, no need for fancy mice or screens. Not as easy for you, you say. Well if you want a millennial toy rather then a professional tool I suppose it's alright, but we produce professional tools." Sound familiar?

[pointhi]

I despair at any possibility of getting this basic "usability" idea into the collective heads of the KiCad crowd.

Sorry to say that, but you fell into the standard UX pit of: my approach is the right one.

No I didn't. It's just convenient for you to say that so that you can continue thinking that your approach is the right one. Yes, it's hard to have your baby criticised, especially when someone says that your whole UI is riddled with usability problems and needs a radical overhaul.

I never said KiCad is perfect, just that your usecase is not the only one.

For your use-case: place a resistor once, and duplicate it when you need a second one. Should be fast enough and works for every part, not only the "standard" ones.

For a start it's not a use-case, it's a UI usability case, inability to tell the two apart is telling; as is dropping into business analyst-ese with phrases like "use case".

Every time, every bloody time, someone says to fanboy contributors of program x "It would be better if the UI did this like that" the fanboys say "Oh, you can do that [the functional equivalent] by <series of moves in existing UI>" thereby completely missing the point. Just because something is functionally possible doesn't mean that it satisfies usability criteria.

I've been a programmer for 45 years, since the bloody punch card era. If I was as resistant to improving usability as you guys seem to be I'd still be insisting that my customers used punch cards. "No, look, you CAN change the customer field. Just pull the 3rd card out of the batch, walk over to the punch/duplicator, hit COPY for the first twenty columns, then type the new customer name, and then keep hitting copy until the end of the card. Then put the card back into the stack. See, no need for fancy mice or screens. Not as easy for you, you say. Well if you want a millennial toy rather then a professional tool I suppose it's alright, but we produce professional tools." Sound familiar?

Yes, this is a workaround. But simply because I don't think what you want will be implemented upstream anytime soon. It starts with the simple problem: what is a Resistor? That thing is inside a library, which is >>OPTIONAL<<. KiCad does not know what a resistor is on the data-format level, and never will. Would you like to hardcode this? Would you like it configurable? If this will ever be implemented, this would likely be a plugin you install.

Let's start at the basics: What are the use-cases/user stories a proper symbol chooser should support:

Your usecase: I want to place Resistor, Capacitor with a single click on a Toolbar Button -- Problem: you already stated the solution of your problem which serves your use-case, but not other ones
Other usecase: I want to place a LM747 --> does not work with your Toolbar
Other usecase: I want to place this fully defined symbol from the database --> does not work with a simple Toolbar
Other usecase: I want to place the same symbol multiple times --> can be seen independent from the symbol chooser
Other usecase: I want to choose the next symbol after placing the current one
Other usecase: I want to filter the library by maximum supply voltage
Other usecase: I want to categorize the symbols
...

So, lets behave like proper programmers, and collect the requirements first. And what do proper requirements do not include: A solution. Finding solutions is a later step when you actually know what you want.

Anyway, what is a "UI usability case"? Google finds outstanding 4 results where this term is used.

[Cerebus]

It starts with the simple problem: what is a Resistor? That thing is inside a library, which is >>OPTIONAL<<.

Ah, I finally get it. I foolishly am discussing a program to do electronics with, where resistors are ten-a-penny and appear on every circuit board I own, every circuit board I've ever designed (except breakout boards for switches and other trivia) and in every schematic I have ever drawn, whereas you are clearly not. If you can't or won't get it into your head that placing the single most used individual component in the whole of electronics should be the 2nd easiest of all tasks (placing wires should obviously be 1st) in a schematic capture program then <throws hands up in despair rather than completing the sentence>. To posit that a schematic capture/PCB layout package should not need to know what a resistor is so wrong there isn't a word for it.

Anyway, what is a "UI usability case"? Google finds outstanding 4 results where this term is used.

English, work it out. That it's not industry jargon like "use case" or "user story" goes a long way to explaining why there's a lot of shitty software out there. It's quite clear that you're at the "being clever/snarky" stage now, rather than the "even remotely looking like someone who might one day actually listen stage".

I tire of this. There's really no point in me carrying on as it's quite clear that you are just dead set to do it one way and one way only and you really, really, don't care what your 'customers' think or are incapable of comprehending that there is a better way. I'm out.

For the avoidance of doubt, I'm marking this topic 'ignore' I really do feel that I'm wasting my time here. I tried...

[pointhi]

Well, it became clear this will not work out between us.

English, work it out. That it's not industry jargon like "use case" or "user story" goes a long way to explaining why there's a lot of shitty software out there. It's quite clear that you're at the "being clever/snarky" stage now, rather than the "even remotely looking like someone who might one day actually listen stage".

I tire of this. There's really no point in me carrying on as it's quite clear that you are just dead set to do it one way and one way only and you really, really, don't care what your 'customers' think or are incapable of comprehending that there is a better way. I'm out.

For the avoidance of doubt, I'm marking this topic 'ignore' I really do feel that I'm wasting my time here. I tried...

I come from research, and when you invent a term, you need to explain it. Even it looks clear to you, without specification, people will interpret different things into it. And industry jargon like requirements engineering is something everyone should have heard of, as well as some basic terms. Nothing wrong with it.

And even though you will not read this anymore: I can listen and change my opinion. I do that all the time. But it requires good arguments, and I did not see them in yours. Coming with arguments like "I've been a programmer for 45 years" is the opposite of an extraordinary argument. I know quite a few, way more senior engineers than me, which build crappy software from an engineering standpoint, and they don't see this. From my point of view, I only heard: this is the right way and you are wrong. No discussion anywhere to iterate design ideas and incorporating feedback of use-cases, other people might want. And I'm not the only one who stated them to you.

---

Lets start with an idea of me, if this could be a workable symbol chooser (idea was stated in https://gitlab.com/kicad/code/kicad/-/issues/6028). No guarantees given, more like a though how it could be improved.

What about a dock, which can be attached to the window like the layer-chooser in pcbnew:

1. You have a visualization of the current selected symbol/footprint in the dock
2. You see some searchable tree/etc. (what works well), where you can select the symbol
3. You can apply extensive filters on the symbols
4. You can add symbols to a bookmark, which are like a library, and which you can easily open anytime

Advantages:

* always open
* immediate visual feedback what library you are in
* you can (for example) open the devices library and place resistors as often you want, like in a toolbar (but more flexible)
* Works with the KiCad library, but also some future concept like database library
* scales well for bigger libraries
* you can select one library by default (Like Devices) to help beginners
* one solution for everything. Not something specialized.

Disadvantages:

* space could be too small to fit everything in
* bad for small screens
*

---

Any input? What good examples of other EDA's (not LTSPICE) can you think of in this regard, where we can steal the good parts?

[PlainName]

What good examples of other EDA's

Well, now you mention it...

(not LTSPICE)

Oh.

 

[JohnG]

For what it's worth, a library for simulation-only purposes would be a nice addition to KiCad, and could also satisfy those who just want to sketch schematics. Ideally it would consist of all the Spice primitives. It could even be an addon, and maybe there is one already that I haven't found yet (if anyone knows, please let me know!)

For background, I have only just started looking at KiCad, and so far strictly for simulation. Why in the world would I do that? Because I work for a semiconductor company, and therefore can't use LTspice. I also know personally someone who was publically yelled at by the author of LTspice in a seminar because he had been toying with an IC design and dared to ask some questions about it. KiCad and ngspice seem to have the most momentum behind them, and I am looking for a longer term solution.

I also think of KiCad as a good choice because of promising developments for PCBs. Right now I use Altium. I found the user interface hard to learn, but I learned it. The same for Eagle, Orcad, LTspice (which I used to use extensively), etc. So yeah, KiCad has some pretty funky interface choices, but it also does some things I really like. I still was able to go from downloading it to running a simulation in about an hour. It took me a lot longer to get anywhere with Altium.

Interestingly, I learned that Altium 20 also uses ngspice as its simulation engine.

You want a terrible example of a user interface? Try Windows 10 and Office 365. It tricks you into thinking you can be productive, but by the time you realize it's all a terrible lie, you are trapped in sticky secretions and slowly dissolved by MS's digestive enzymes...

John