"Please Rate my Design/Schematic"... [RANT]

[Nominal Animal]

Is that just your convention, or the tool convention - and how is a new reader meant to know that?

Not a convention, something I'm experimenting with, to make it easier to understand how the parts interact.  The colors are only informative; nothing of significance would be lost if all were the same color.  I describe the colors in the post where I explain each part of the circuit.

Red for supplies and black for ground seems to be somewhat of a convention, though.

The software equivalent is code using "goto name", instead of if-the-else, while, exceptions. That has been deprecated for half a century.

No, it is not.  goto is still useful in structured cleanup, when nested do { ... } while(0) "loops" would require extra state variables or excessive nesting.

Splitting such cleanup into separate functions –– which is often suggested as a way to avoid goto (because they, like yourself, mistakenly believe it is "deprecated" somehow) –– on the other hand is analogous to splitting a schematic into too small pieces, making understanding the entire schematic difficult.

[Smokey]

I feel like hierarchical schematic design is also partly to blame.  All this is is an extension of the hierarchical design concept with the smallest "block" being each individual part.  And then putting all those blocks on one page instead of separate pages. 

I look forward to the next stage of this schematic design style where each part is actually on its own page. 

[pcprogrammer]

.... schematic ....

What would make drawing your schematic with wires easier, is when you use the proper logic symbol for inverters and buffers like the 74125 you used in different forms. (single gate, quad gate) Then you can have the left to right signal flow and place the gate in series with the wires, instead of having to bend around the rectangle with pins on either one or two or even four sides.

EasyEDA does allow you to create these logic symbols such that the power pins are hidden and the different gates in a quad package are identified in the designator like U2A, U2B, etc. How to do this should also be in the documentation you referred to. I have done it before, but would have to look it up again because it did not stick. 

[Benta]

Wow, lively discussion here. The topic seems to have struck a nerve (which is why I started it).
Professional engineers (yes, I'm one) hate dealing with bad documentation. We're paid by the hour, and the "Little Boxes" schematics are total time-wasters.

Anyway, I promised the original schematic and here it is (feel free to comment on that as well).
I repeat the "modern" version for easy comparison.

Cheers.

[tggzzz]

Wow, lively discussion here. The topic seems to have struck a nerve (which is why I started it).
Professional engineers (yes, I'm one) hate dealing with bad documentation. We're paid by the hour, and the "Little Boxes" schematics are total time-wasters.

Anyway, I promised the original schematic and here it is (feel free to comment on that as well).
I repeat the "modern" version for easy comparison.

Cheers.

Just so. Life is too short to waste time wondering whether someone creating a circuit and schematic is ignorant or lazy - or has worthwhile ideas. Nowadays there is so much stuff clamouring for our attention that it is necessary to quickly decide what to ignore. An incompetent schematic is a good reason to ignore something.

If I was to pick bits with that decent schematic, they would be...

Possibly having a separate Vcc label for each side of the schematic. Reduces wires, keeps info next to the pins. Yes, I am aware this is almost labelling nets on pins, but PSU wires are an exception.

I think I would have the two JK flip flops on the right of the diagram facing in the traditional direction, with just the single feedback wire looping backwards. The mentality would be "from the counter's TC we form a new signal, and that is looped back to the counter input".

But those are nits; it is easy to follow the signal flow through the circuit, and to see the intention.

[temperance]

Wow, lively discussion here. The topic seems to have struck a nerve (which is why I started it).

Yes indeed, here is an example from Infineon:

https://www.infineon.com/dgdl/Infineon-ApplicationNote_evaluation_board_EVAL_HB_BC_1EDN8550B-ApplicationNotes-v01_01-EN.pdf?fileId=5546d46266a498f50166ef75bd7e183f

Page 13:
-MOSFET symbols. The what?
-The person who made this drawing placed the pin numbers on the inside and the description on the outside of the symbols... Honey, this shit doesn't generate a netlist, must call the help desk.
-Some things are crammed together because other items are unnecessarily spread out.
-The fact that the driver uses Kelvin sensing really stands out. It's the first thing you should see clearly because that's what the app note is all about. Instead you see just a bunch of wires.
-The half bridge really resembles a half bridge if you look carefully.

Overall a sloppy job.

In my opinion schematics should communicate the function and intent for anyone involved. Software developers, mechanical engineers, PCB layout requirements,...

In symbols for micro controllers for example I use PA0, PA1,... On the connections I always tag a label stating the intended function such as TIM1-CH1-PWM and preferably identical as they appear in the data sheet. By doing so the software developer knows what to do without asking questions.

Besides, those differential input kelvin sensing drivers are very interesting.

[Benta]

Page 13:
-MOSFET symbols. The what?
-The person who made this drawing placed the pin numbers on the inside and the description on the outside of the symbols... Honey, this shit doesn't generate a netlist, must call the help desk.

This has touched on before. If you're too lazy/incompetent to draw a symbol, just use the PCB footprint instead. Job done. Who reads this stuff anyway.

[Nominal Animal]

Again, the following is not intended as an excuse; only an attempt to explain why and how these bad schematics happen.

What would make drawing your schematic with wires easier, is when you use the proper logic symbol for inverters and buffers like the 74125 you used in different forms. (single gate, quad gate) Then you can have the left to right signal flow and place the gate in series with the wires, instead of having to bend around the rectangle with pins on either one or two or even four sides.

This situation seems analogous to comments in source code.  To me, source code and schematics are the implementation of the desired logic.  Source code comments describe the intent.  For schematics, that is in external documentation, like the posts here where I describe the ideas and reasoning behind each part of the schematic.

You are saying that instead of doing it that way, I should split the ICs into their logical symbols; for example TS5V330 into a four pairs of SPST switches with common active low enable and selector inputs, TS5A3159A into a pair of SPST switches with active low enable and selector inputs, SN74HCS125 into four tri-state buffers (buffer and switch) with Schmitt trigger inputs, and 74LVC1G125 into one tri-state buffer with Schmitt trigger inputs.   (These logic symbols are usually shown in the IC datasheet somewhere.)

I do realize that this will make the schematic much easier to read.  Definitely so; splitting the ICs into (logic) subsymbols makes the wiring simpler, too.  Then, at a glance, one can see not only the intent but also the implemented functionality.  I can understand why this should be my goal, and will try to do this in the future –– I've already discovered that for transistors and opamps, especially dual and quad opamp packages, this is basically a necessity.  (With those, I've struggled between having the supply pins in all subsymbols (problematic with EasyEda), in one of the subsymbols, or in a separate opamp subsymbol without inputs and outputs.)

However, this is a lot of extra work that does not help whoever created the schematic, when their main purpose is to create the PCB.  Splitting the ICs into the logical components they implement shows some, but not all, the reasons for picking specific ICs.  It is like having the source code properly commented, explaining the algorithms and choices made for each function: good to have, but not necessary when most users will simply compile it without any modifications, or extend its functionality without changing any of the choices made.  Doing it the stupid way is not wrong per se, just less useful and less clear than it could be; incomplete as a product, but understandable for hobby stuff.

(Me, I'm starting to think I might want to create my own schematic editor, because it would definitely be possible to construct a graphical UI that encouraged "good taste" more than EasyEda and KiCad do.  In particular, combining and splitting symbols into packages and vice versa would make this easier, and also solve the pin selection issues.  EasyEda already shows browsers are fully featured enough and fast enough so HTML+JavaScript should suffice, even without a network connection... )

[schmitt trigger]

Poorly drawn schematics pre-date semiconductors.
I remember seeing an all-tube (valve) controller, where the tubes were drawn like circles with the pin numbers. Not showing the heater, the cathode, the plate or the grid(s).
Although there were a few tubes which almost everyone knew them by heart (the ECC83-12AX7 being one of them), most tubes one would have to look them up in a dog-eared and worn RCA receiving tube manual.
By the time you had found the correct tube, your thought process would have been derailed.

[tggzzz]

What would make drawing your schematic with wires easier, is when you use the proper logic symbol for inverters and buffers like the 74125 you used in different forms. (single gate, quad gate) Then you can have the left to right signal flow and place the gate in series with the wires, instead of having to bend around the rectangle with pins on either one or two or even four sides.

This situation seems analogous to comments in source code.  To me, source code and schematics are the implementation of the desired logic.  Source code comments describe the intent.  For schematics, that is in external documentation, like the posts here where I describe the ideas and reasoning behind each part of the schematic.

There is a programming fashion that code should be written in ways that make it self-explanatory, and that comments are unnecessary. While it is true that code should be self-explanatory, it is religious dogma to believe comments are superfluous. Comments are necessary, just as good naming and traditional formatting is necessary - not for the machine but for the humans.

The dual is that schematics should be drawn on a way that makes it easy to read them, using traditional design patterns for components in a subcircuit.

[Benta]

However, this is a lot of extra work that does not help whoever created the schematic, when their main purpose is to create the PCB.   

...

Doing it the stupid way is not wrong per se, just less useful and less clear than it could be; incomplete as a product, but understandable for hobby stuff.

I understand that "quick 'n dirty" work is OK for a hobbyist, if you don't want to revisit the project after some years because a new idea has come up.
It's fine with us all, your private thing, keep it at home.
The problem comes up when you publish it as the latest and greatest. Just don't. There's enough web noise as is.

(Me, I'm starting to think I might want to create my own schematic editor, because it would definitely be possible to construct a graphical UI that encouraged "good taste" more than EasyEda and KiCad do.  In particular, combining and splitting symbols into packages and vice versa would make this easier, and also solve the pin selection issues.  EasyEda already shows browsers are fully featured enough and fast enough so HTML+JavaScript should suffice, even without a network connection... )

You don't seem to know much about KiCAD, but this is what your much-quoted 74125 looks like there. No boxes in sight.

[Nominal Animal]

You don't seem to know much about KiCAD, but this is what your much-quoted 74125 looks like there. No boxes in sight.

You don't seem to understand the difference between an user interface and a symbol library.

All schematic editors support multi-part symbols.  There is a BIG difference between multi-part symbols and an UI that supports combining and splitting symbols into a single footprint.

You seem to have quite a big mouth, Benta, but very little to show to back that up.  I am starting to believe you have very little to contribute, only a lot of complaints.

[Benta]

You seem to have quite a big mouth, Benta, but very little to show to back that up.  I am starting to believe you have very little to contribute, only a lot of complaints.

I'll leave that one here for posterity.

[pcprogrammer]

However, this is a lot of extra work that does not help whoever created the schematic, when their main purpose is to create the PCB.  Splitting the ICs into the logical components they implement shows some, but not all, the reasons for picking specific ICs.  It is like having the source code properly commented, explaining the algorithms and choices made for each function: good to have, but not necessary when most users will simply compile it without any modifications, or extend its functionality without changing any of the choices made.  Doing it the stupid way is not wrong per se, just less useful and less clear than it could be; incomplete as a product, but understandable for hobby stuff.

True, and that is more or less what I wrote in my first post in this thread.

It of course also depends on the intent of the schematic. If it is just for personal use to get to an error checked PCB, who gives a f***. When the intention is it to be for service and has to be read by a wider audience, than yes, you better make sure it is a good readable schematic. But there is a learning curve, and with new technology come new methods and options.

But just as with code, having proper comments makes it easier to understand the intent when you circle back on it after many years. Especially with aging, I find it to become more and more important. Recollection deteriorates faster then I would like. 

Wow, lively discussion here. The topic seems to have struck a nerve (which is why I started it).
Professional engineers (yes, I'm one) hate dealing with bad documentation. We're paid by the hour, and the "Little Boxes" schematics are total time-wasters.

Just so. Life is too short to waste time wondering whether someone creating a circuit and schematic is ignorant or lazy - or has worthwhile ideas. Nowadays there is so much stuff clamouring for our attention that it is necessary to quickly decide what to ignore. An incompetent schematic is a good reason to ignore something.

No one is forcing us to be on the forum and review other peoples work. If someone asks for a bit of help and the schematic is crap, than the options are to either walk away and ignore it completely, or just get over your own personal preferences and do a bit of work to understand the intent and give some guidance. Maybe a hint to how a schematic can be drawn better, but the full on bashing and pounding I have seen from some of the members is over the top.

Unless the recipient is "asking" for it of course. Some art and robot springs to mind 

Talking about crappy schematics that one holds the crown.

[tggzzz]

Just so. Life is too short to waste time wondering whether someone creating a circuit and schematic is ignorant or lazy - or has worthwhile ideas. Nowadays there is so much stuff clamouring for our attention that it is necessary to quickly decide what to ignore. An incompetent schematic is a good reason to ignore something.

No one is forcing us to be on the forum and review other peoples work. If someone asks for a bit of help and the schematic is crap, than the options are to either walk away and ignore it completely, or just get over your own personal preferences and do a bit of work to understand the intent and give some guidance. Maybe a hint to how a schematic can be drawn better, but the full on bashing and pounding I have seen from some of the members is over the top.

Unless the recipient is "asking" for it of course. Some art and robot springs to mind 

On a forum, I tend to agree. Unfortunately it occurs more widely that that.

I see it associated with things people are trying to sell, or where they are using the schematic to illustrate their coolness/competence/etc and/or I am trying to work out the crux of what they are claiming.

[pcprogrammer]

Just so. Life is too short to waste time wondering whether someone creating a circuit and schematic is ignorant or lazy - or has worthwhile ideas. Nowadays there is so much stuff clamouring for our attention that it is necessary to quickly decide what to ignore. An incompetent schematic is a good reason to ignore something.

No one is forcing us to be on the forum and review other peoples work. If someone asks for a bit of help and the schematic is crap, than the options are to either walk away and ignore it completely, or just get over your own personal preferences and do a bit of work to understand the intent and give some guidance. Maybe a hint to how a schematic can be drawn better, but the full on bashing and pounding I have seen from some of the members is over the top.

Unless the recipient is "asking" for it of course. Some art and robot springs to mind 

On a forum, I tend to agree. Unfortunately it occurs more widely that that.

I see it associated with things people are trying to sell, or where they are using the schematic to illustrate their coolness/competence/etc and/or I am trying to work out the crux of what they are claiming.

When it is in the commercial domain, or to show off, I fully agree and with pleasure supply the sledge hammer to do the pounding.

[nfmax]

What would make drawing your schematic with wires easier, is when you use the proper logic symbol for inverters and buffers like the 74125 you used in different forms. (single gate, quad gate) Then you can have the left to right signal flow and place the gate in series with the wires, instead of having to bend around the rectangle with pins on either one or two or even four sides.

This situation seems analogous to comments in source code.  To me, source code and schematics are the implementation of the desired logic.  Source code comments describe the intent.  For schematics, that is in external documentation, like the posts here where I describe the ideas and reasoning behind each part of the schematic.

There is a programming fashion that code should be written in ways that make it self-explanatory, and that comments are unnecessary. While it is true that code should be self-explanatory, it is religious dogma to believe comments are superfluous. Comments are necessary, just as good naming and traditional formatting is necessary - not for the machine but for the humans.

The primary function of the source code is to support human reasoning about the program behaviour. As a secondary function, it enables the compiler to choose the processor instructions needed to implement it.

[SteveThackery]

Because of this, I do claim that this kind of splitting is done because properly connecting the wires is way harder (without creating an even more horrible mess), when you don't have the experience yet in placing the components so the wires fall at least halfway neatly.  Using net labels like I have here is simpler; I just didn't know how deeply annoying this is to more experienced people.

It's not annoying per se, it's just that schematics like that are essentially unreadable to a human brain, thus leading to frustration.

A good schematic is a bit like a flowchart for the signal. I'm sure you are familiar with flowcharts and how their structure reveals the "logic" behind the flow, the decision points, the loops, etc.  Now imagine the same flowchart drawn as randomly placed boxes with labels instead of lines. I'm sure you would agree that it would be vastly more difficult to interpret.

In fact, I bet you would be sorely tempted to take a pencil and draw lines between the labels to help clarify it. 

[tggzzz]

Because of this, I do claim that this kind of splitting is done because properly connecting the wires is way harder (without creating an even more horrible mess), when you don't have the experience yet in placing the components so the wires fall at least halfway neatly.  Using net labels like I have here is simpler; I just didn't know how deeply annoying this is to more experienced people.

It's not annoying per se, it's just that schematics like that are essentially unreadable to a human brain, thus leading to frustration.

And they are completely non-scalable. Try reading a schematic like that with 100 semiconductors plus passives!

[2N3055]

That is not a schematic at all.
It is a table of what component pin is connected to what net.

Whole purpose of schematic diagram is graphic representation, where it is visually easy to "see" what is connected where.
You can have 35 pins connected to same node. How do you keep in head all of them?

That is why on classic schematic diagrams you do have labels for power rails, data buses etc.
That is logical.

But trying to figure out any kind of signal is connected where in analog circuit looking at that kind of netlist is ludicrous. And how do you even design a circuit that way?

[Analog Kid]

Unless the recipient is "asking" for it of course. Some art and robot springs to mind 

Talking about crappy schematics that one holds the crown.

For those who missed the fun, here's a sample of his "schematics" that I saved. They do indeed take the cake.

[Benta]

@Analog Kid, that seems to point at a certain user (I've no idea who), but I did NOT want to go there.
I started the thread as a discussion about a certain "fashion fad", not to call anyone out.
Better edit your post IMO.

[Analog Kid]

I won't edit it (why should I?), and I don't think it detracts at all from your point, which is different but somewhat related.

[tszaboo]

I attach my "modern" schematic. Comments appreciated.

"Unreadable trash that doesn't work, you will make a perfect software engineer at this company!"

[SiliconWizard]

Just a quick note regarding all those schematics made with EasyEDA.
Yes, many are made by people who don't actually have any background in EE, so for sure you won't get common engineering practice there.
But there's another factor: many of these schematics - in particular those often cited as an example of excessive splitting into small blocks - are pretty much the wiring of off-the-shelf dev/breakout boards. So that would explain the schematics: people are effectively just showing how they wired some boards together rather than make a proper, full schematic. And I think those people tend to get into this habit even once they start making their full own PCBs.

After that, the "right" amount of splitting and hierarchy just... depends. On the complexity of the design, on whether some blocks can reasonably be reusable or not, your own preferences, and other factors.
Some people just freak out at the first sight of a label, others will just put them everywhere. Extremes. As often, reason is in a "healthy" middle ground which depends on the project at hand.

One thing that irks me with labels is people stubbornly not willing to use the direction property (in/out/bidir/or none) appropriately. Many seem to be oblivious to direction and feel like the right thing to do is connect the wire to the pointy end of the label. Properly using the direction of labels (when appropriate, that is when there is a clear signal "flow") helps readability a lot. (Just like placing schematic blocks in a logical flow.) Not doing so makes labels a huge mess.