Resistor Puzzle Solution... and a rant on schematics

RE: Resistor Puzzle Solution... and a rant on schematics

rick chinn — Tue, 09 Apr 2013 00:09:47 GMT

Nothing makes a circuit more understandable than a well-drawn schematic. Here are my suggestions:

for circuit blocks that are specific, more or less textbook circuits, use the canonical form of the circuit, such as found in the textbooks.

avoid 4-way junctions. from a style standpoint, I don't use jumpovers; I draw thru, and use dots.

V+ should be up, V- should be down.

grounding should flow from the diagram, unless you are deliberately trying to obfuscate some circuit detail that is the difference between someone copying your circuit successfully and someone having difficulty. I belive in the free exchange of schematics, but there are companies out there now who will copy your work and sell it as their own, so you might as well not make it easy for them.

circuits should be drawn in blocks. Grouping the reference designator numbers by block also helps.

Large numbers of parallel lines that join blocks together should have a line identifier placed on them every so often, as well as at the source and destination. This helps prevent blindness on the part of someone trying to figure out where things go.

Off-page symbols should ident the page where they reappear, as well as where they originated.

Some draftspeople may not have circuit-sense, in that case, you may have to at least block the circuit out, or maybe you must do the schematic capture yourself to make the circuit the most readable.

RE: Resistor Puzzle Solution... and a rant on schematics

Jens-Michael Gross — Wed, 03 Apr 2013 16:44:45 GMT

Whitham, the problem with symbol standards is that there are so may of them. It starts with simple digital gates. The typical American dsymbol is totally different from the Eurpean ones. And I don't mean that the American ones are drawn in inches and the European in centimeters :) Both types are official standards.

It continues with more complex things like D-latches or so simple things like resistors. Eagle, the layot program we use, comes with separate libraries for both "worlds".

And when it comes to special ICs, the usual way is to draw a rectangle and arrange all signals in the order of the pinout. Standard OpAmps have teh typical trianle and if a chip contains more than one, they are separate elements (and the supply is separate too), but that's the end of the line. Everything above is just a rectangle with many pins.

I try to maintain our own internal symbol library, ut it is not easy at all and takes quite some time. So if a new part is required for circuit, there's usually not the time to think about a proper symbol first: back to the rectangle. To be fixed later (often means: never).

Alan: we once had outsourced doing the layout of our PCBs. And we had schematics as you described, with all the blocking caps on one sheet. And the one who did the layout did put all caps together in one corner of the PCB.

Of course this layout was totally useless.

Now we have the blocking caps directly (and visually) connected to the explicit (rather than implicit by pin/net name matching) supply symbol for each part and on the same sheet as the functional unit of the part. They are still out of the way of the functional signals, but at least it's obvious now where they belong.

RE: Resistor Puzzle Solution... and a rant on schematics

Mike Kaelin — Mon, 01 Apr 2013 16:22:18 GMT

I always thought the old-style transistor drawings (often seen in IC datasheets) were drawn pretty poorly. They had a line for V+ and one for ground and everything was attached from top to bottom… Unless you redraw, you have no clue about how all the transistors interact.

Also, a good schematic can (and should) lead to a good layout, with the signal flow and current paths matching what’s on the schematic. If your symbols are drawn right, it helps to prevent signals crossing over each other and minimizes trace lengths and layer jumps. Sometimes it's worth redrawing the stock library symbol to make the layout easier, and sometimes it also includes the choice of pins on your microcontroller. You can also add labels and boxes to the slikscreen art. A layout that matches the schematic flow is invaluable for debugging and repair.

Great article!

RE: Resistor Puzzle Solution... and a rant on schematics

Earl Jensen — Mon, 01 Apr 2013 15:59:25 GMT

OK, I'm my soapbox too. Schematics are convey information on functionality, not physical representation. A proper schematic will reveal how the signal flows (analog or digital) through the circuit and what happens to it. This is invaluable for debug and test. It irks me when a see a IC on a transistor, directly copied from the pin diagram. Either automated form a CAD package, lazy or the person just don't understand electronics!

RE: Resistor Puzzle Solution... and a rant on schematics

Tim Sobering — Mon, 01 Apr 2013 14:17:58 GMT

Good suggestion Mike. I'll have to revise that into the next version of the doc. But I have to say I think I'm swimming against the tide with the students.

RE: Resistor Puzzle Solution... and a rant on schematics

Alan Rich — Mon, 01 Apr 2013 12:56:35 GMT

I agree 100% about your comments regarding the importance of a well-drawn schematic. A schematic should be like reading a well written book. It should flow from start to finish - a schematic should flow from left to right; it should tell a story so that 10 years later someone can pick up the schematic and understand it.

As a former field application engineer I was asked to review many customers’ schematics. The worse was from a major networking company; it was for a standard synchronous buck regulator with a controller, external FETs, inductor, input/output capacitors, and a discrete resistors and capacitors. The schematic was spread over four sheets! Controller on one, FETs and inductor on another, all the discretes for the board on another, and all the board's bypass capacitors and input/output capacitors on a fourth sheet! The connectivity for the net list was correct, but the schematic was impossible to read.

Another company's schematic had many missing visual connections on the drawing, but the net names were included and correct so that the PCB layout was correct according to the net list! But, unless you very carefully looked at the visual schematic, it was wrong.

When I am asked to review a schematic, often the first thing that I do is redraw the schematic with my own well trusted schematic capture program. It is amazing how many mistakes I find in the original schematic, and how much I learn by this process.

RE: Resistor Puzzle Solution... and a rant on schematics

Randy Luck — Sun, 31 Mar 2013 23:46:23 GMT

The schematic on page 37 of the datasheet for the ADS1258 EVM is an excellent example of how NOT to draw a schematic. It is a cramped, confusing jumble of jiggles and jaggles. I can only say two things positive about it. 1) it is correct, and 2) there are no 4-way junctions. www.ti.com/.../sbau126c.pdf

RE: Resistor Puzzle Solution... and a rant on schematics

Mike Beach — Sun, 31 Mar 2013 18:09:34 GMT

I too strongly agree with the thrust of these comments. A detail I didn't notice in Tim Sobering's document is the desirability of minimising the number of corners, crossovers and joggles in the connections. Each one adds, if only slightly, to the effort of reading the diagram, and its removal can make the result fit more comfortably on a given sheet size. But again, it takes more time to organise the layout of the diagram - well worth it, in my opinion.

RE: Resistor Puzzle Solution... and a rant on schematics

Whitham Reeve — Sun, 31 Mar 2013 17:08:40 GMT

My main gripe is inconsistent circuit symbols. There are standards, you know, but how come people don't follow them? Also, what about the resistor symbol? Should it be a rectangle or sharp, wavy lines?

RE: Resistor Puzzle Solution... and a rant on schematics

Tim Sobering — Sun, 31 Mar 2013 14:00:25 GMT

Bruce...you are ranting "to the choir". I nag my students endlessly about drawing schematics. It wastes my time to have to figure out what each of them is doing when they don't follow a basic set of rules, as you mentioned. I finally wrote a guideline...not perfect perhaps, but a starting point. www.ksu.edu/.../Technote%208%20-%20Guidelines%20for%20Drawing%20Schematics.pdf

RE: Resistor Puzzle Solution... and a rant on schematics

Jens-Michael Gross — Mon, 25 Mar 2013 20:11:56 GMT

Today, many schematics are automatically generated from libraries. The software generating them may put everyhting together nicely, and the router will be able to do its work, but the result is lacking human readability.

However, once more the things said about schematics also apply to software.

While high-level programming languages give you, and even force you to use structured elements, peopl emanage to mangle their code until nobody understands what's going on. And of course helpful annotations are scarce. (the ones that are present often only state the obvious, or, worse, contradict the code). Too bad that some of the most popular layout programs do not really support free text input. (possible, but uncomfortable to handle)

Thanks again for some valuable gems of wisdom about hardware design.

RE: Resistor Puzzle Solution... and a rant on schematics

Avi Chami1 — Thu, 21 Mar 2013 07:03:26 GMT

You can learn a lot about an analog circuit only by looking at well drawn schematcs

Sadly, that is not the case for today digital circuits, that usually include huge behemoth with hundreds and even more than a thousand of pins each. But even for these complex devices, many things can be done to improve the schematics

Give meaningful names to signals and buses

Divide the devices in smaller blocks, group functional signals of the same kind on each block (power, memory bus, interfaces, etc.)

do not heasitate to redraw the symbols if that helps to straight the connections between two symbols

If possible and if it doesnt affect clarity, prefer real nets to named nets connections

Avoid using names like Fpga-tx and fpga-rx. I prefer fpga-to-pci and pci-to-fpga.