Is this possibly the worst drawn schematic ever?

[c4757p]

At a time when there wasn't CAD, not surprising; the alternative of erasing many times or getting a new sheet and starting over is not really practical.

Really, though, how hard would it be to sketch it out roughly first to figure out where things should go?

I like your style, Glen. This is remarkably easy to follow, and having seen a good few of these old RF schematics myself, I can imagine how shitty it was initially

[krivx]

It wouldn't surprise me if drafting was a different person's duty than designing. Have you ever drawn out a schematic of a circuit you didn't design?

[c4757p]

Of course it was, but the designer would have had to sketch out the circuit first so the drafter knew what he was supposed to draw...

[GK]

On the topic of badly drawn schematics, some of the worst are from the vintage era of radio. There seemed to be a phase where the draftsman would use his stencil to draw in the major components, evenly spread out across the page, and then proceed to wire everything up.

At a time when there wasn't CAD, not surprising; the alternative of erasing many times or getting a new sheet and starting over is not really practical. Some of the early TV schematics weren't too bad, however.

Well sure, but economic rationalizations in wartime production had a great deal more to do with it than the lack of CAD. Have a look at the some of the early Tektronix manuals for examples of properly drawn schematics by the traditional draftsman who had the time and equipment to do it right.

[KJDS]

Of course it was, but the designer would have had to sketch out the circuit first so the drafter knew what he was supposed to draw...

One place I worked at the drafter would do the schematic from the designers sketches. It was the designers job to keep it updated and it would usually be necessary to add a few blocks, turning a nicely spaced, well drawn schematic into a mess.

[woodchips]

Lots of little things can add to make a schematic worse, for example:

Manufactures who shun standard components and call them by their own part number, like HP. If a part is selected then say so, and why it is selected.

People who draw interconnecting lines that join at a cross, is that a dot or not? Just how many hours have been wasted by that silly method of drawing?

Inputs on the left, outputs on the right, usually. But would you split a microcontroller port so some on the left, some on the right of the symbol if the port has inputs and outputs?

Am I alone in finding the new, 1980s, logic symbols to be meaningless? That a symbol has external connections to the symbol outline seems just foolish.

Meaningful signal names.

Programmable devices described by what they do.

[David_AVD]

Am I alone in finding the new, 1980s, logic symbols to be meaningless? That a symbol has external connections to the symbol outline seems just foolish.

Not sure what you mean here.  Can you give an example?

[mikeselectricstuff]

That's just lazy and dumb

What is? Sorry, it wasn't clear.

Assigning pins based solely on what looks neat in the schematic

Software is easy to fix.

Well, it depends. There are often restrictions on things like what kind of interrupts are available on a given pin, how many timers are available per port for PWM output and so forth.

That's about fully understanding the device, and what things can and cant be swapped ( and if necessary hacking up a quick prototype to test anything that's not clear in the datasheet)

When your code is filling up a 256k flash ROM and has to do some very complex stuff things like that can make it excessively complex and fragile. We are not talking Arduino sketches flashing some LEDs here.

If pin assignments make a significant difference to a 256k code base, you're probably doing it wrong

The hardware guys never read the detail in the datasheets either.
For example on AVRs with SPI ports the Slave Select pin forces the SPI peripheral into slave mode when pulled low, so you can't use it as a general purpose input.

I've been caught by that one, but fortunately it was just a DIP switch so I could work around it by setting  it as an output when SPI stuff was happening.

All  of this is about knowing and understanding the part, and where having the same person doing hardware and (at least the low-level) software makes more sense than seperating the two.

At the very least, hardware people should be able to write enough code to demonstrate that the hardware does what it should before throwing it over the wall to the code monkeys.

[Hypernova]

Well, my half serious comment was in reference to the schematics I posted.


-------------------------------------------------------------------------------------------------------------

On the topic of badly drawn schematics, some of the worst are from the vintage era of radio. There seemed to be a phase where the draftsman would use his stencil to draw in the major components, evenly spread out across the page, and then proceed to wire everything up. One rainy weekend a while ago I got fed up tracing the vintage wiring schematic spaghetti for some old units I was working on restoring, and re-drew them a bit neater:

Those are some nice sheet mans, IMO one very under appreciated metric of any engineering diagram is whether they remain informative when converted to black and white. This can be critical if they need to be photocopied. I've read too many IEEE papers where the graphs are useless because the author submitted them converted to B&W.

[tom66]

Good news is, the power supply is now fixed.
I accidentally disconnected one of the wires in the cable bundle going from the main board to power supply. Looks like it's related to backlight on/off as the backlight now stays on except during standby. Ah well. Tracing out exactly where that wire went is a nightmare because the colours are not coordinated. Different colours for the same voltages. Red and orange frequently used for ground. Ugh.

[mzacharias]

Many of the pictures posted of horrible schematics look very much like what I deal with every single day in consumer electronics. Large multi-channel receivers, processors, etc. The service data is at least usable for the most part, but much time is often wasted merely getting one's bearings and learning a given manufacturers quirks of notation, naming protocols etc.

Sometimes they are simply evil, in schematic and in physical execution, which by the way is what some of the designers deserve. The most failure prone parts are often hidden under a maze of CAD-designed inter-connected boards when they need not be. For example, crap 7812 and 7912 failure-prone VERY CHEAP Korean voltage regulators which could have been located where it does not take three hours in and out to replace. Merely finding test points for them was a lengthy and frustrating process. Eventually had to give up on a couple of the test points and just tear into the thing and replace them all.

Recent Sony electronic service data is excellent, many with clickable links to zoom you from one connector to it's mate on another board, from the IC position on the board to the parts list and part number, back to the schematic, and so on.

[echen1024]

I would say most disconnected looks like my scratch paper in the lab.

[woodchips]

The new, 1980, logic symbols.

Look at a suitable TI TTL data book, try the 7446/7/8 or 74184 etc etc. The symbol doesn't stop at the outer boundary line, but has wires and joins.

Look at the 74161, a basic counter, where do the pin names go? In the good old days you just labeled pin 15 as RCO, now it seems to be called 3CT=15.

Whilst the new symbols are all logically constructed, they are not obvious and will get worse as circuits are copied.

Have to say that never really seen them used much though, but that is probably reserved for companies that are required to use them on all documentation.