Hacker News Reader

Comment by spott | Hacker News Reader

I'm a little curious... why?

It is much harder to visualize, I don't think it is appreciably shorter to write, it isn't necessarily clearer.

I suppose you can do interesting things like parameterizing parts on the value of other parts, but I'm not sure that is particularly useful.

If this was going to be the underlying core of a graphical netlist generator, I could see it being useful for that...

So, anyone have any ideas what is the big reason I should use this over KiCAD or similar?

View on HN

Since your question is about visualization -- the kind of schematic this is good for looks like this:

http://papilio.cc/uploads/Papilio/fpga_schematic.png http://papilio.cc/uploads/Papilio/duofpga_schematic.png

The drawing tool is really just getting in your way.

And you would use this with KiCAD -- this just takes the place of eeschema.

Those schematics though are just throwing everything into a bus... It isn't really differentiating signals at all.

What is the difference between just labeling all the FPGA outputs (which is done here), and then labeling the inputs where they go rather than having the bus line in between?

Or in other words, is this really the best schematic to be comparing against?

>And you would use this with KiCAD -- this just takes the place of eeschema.

Yes, I misspoke.

That's the point -- if the schematic is just connecting pins to buses and other pins, how is that improved by drawing a rectangle and placing the pins on its perimeter? It can be clearer as a simple textual netlist, or even clearer as a simple textual program that generates the netlist.

I guess my point was more along the lines of: "Is that schematic really a good schematic, and if not, should it be the goal of a project like this?"

Though I get your point as well.

> I'm a little curious... why?

_imho_ it is pretty cool, specifically because, writing circuits as code would make the whole thing _composable_ not unlike functions in programming languages.

canonical pcb design e.g. in KiCAD happens in a couple discrete steps:

      - draw the schematic

      - assign footprints to symbolic parts and

      - place them

now, the netlists is what ties all these phases together. an ability to generate this connectivity list programmatically is very useful (imho).

for example a filter circuit generator could take f.e. order, cutoff and type as filter inputs and give you a correct netlist. for your next design, when you need a different filter, changing couple of parameters should be all that is required !

i _suspect_ that we have become inured to seeing circuit diagram as a means (or crutch) of thinking about them.

designing stuff with this should be akin to sketching out block-diagrams where each block is bit of python code that generates the circuit module...

It looks like someone wanted to make an electronics version of OpenSCAD [0]. I agree, it makes more sense integrated into a graphical schematic capture package. There have been a few times when doing large FPGA pinouts that a tool like this could have been useful.

[0] http://www.openscad.org/

My hobby experience might not be similar to professional usages. In past thought about such tools, considered python for its ease of use and being dynamic. I was leaning towards electron/node.js for native html/css/svg based model generation and display. Also a hybrid approach with point and drag as well as programming facilities generating pcb or components.

It is not a 1-1 comparison but one distant example would be plain text html/css vs frontpage and dreamweaver. Over the time i have seen for a serious work done using plain texts instead of studio based design (may be specific to HTML?).

I think parametric design could be used here, but to what extent that would be useful, I'm not sure.

Couldn't that just be done directly on the netlist, though? I suppose this makes it easier to operate on the netlist, but I imagine there are netlist manipulation libraries that would also accomplish that task.

I am with you. This looks particularly bad for electronic engineering. Why they would define a schematic without any drawing.

Think that FPGA used to be programmed with schematic input. I don't think many people would like to go back there. As someone who went through schematic input for FPGAs, I wouldn't :-)

Reply on news.ycombinator.com