another little 4-layer board

I remember when people were compulsive about running horizontal or vertical traces on specific layers. That made sense when every thru-hole pin was a via.

My SMT boards seem to be totally haywire... do anything that works. What surprises me is that it *does* usually work. There's usually a way to route things. There must be some theory there, topology or something, but I don't know what it is.

Coding an autorouter must be interesting. But so far they don't seem to be very good.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

As you know, it's known to be an epically hard problem for computers, and it's unknown how humans manage to do it somewhat better on small-enough problems.

.

Humans may use something like Annealing to plot paths through mazes and problems alike.

.

Joe

it is like airplanes using different altitude for north/south and east/west. It means no crossings

afaict they can be pretty good, but in the time it take to setup all it needs to know you could have been done doing it yourself

The advantage that an autorouter brings to the process is that it is rerunnable. You can rip up (and many routers do this as part of their routing heuristics) portions of the circuit and reroute them far easier (and more accurately) than doing a manual (re)route.

Routing an analog board is different than a digital as it tends to be more sensitive to noise -- but, can often require fewer layers as analog components inherently allow traces to "step over" other traces without having to stitch to another layer.

Also, analog designs tend to "flow" in a single direction; there are far fewer "backtracks" in analog signal paths. You can see this by "picking up" a ratsnest and noticing how "clumped" it is (or isn't). Digital designs tend to be more entangled.

Successfully using an autorouter typically involves understanding the different types of routing algorithms available to it and knowing how and when to apply each (e.g., it would be foolish to use a maze router on a *typical* entire board).

You also need to know when the "rules" (which are actually just guidelines) can be bent or ignored -- e.g., I often route certain groups of signals on the "power layers". This often lets me reduce a 6-layer design to 4 layers with no real consequences (for a board with no errors).

The days of a "single click" route are likely still far off -- for all but simple designs.

One advantage I have over an autorouter is that I can change the schematic if it makes the PCB easier.

Electronics is cool. I would have been a very bad farmer or goat herder or coal miner.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Sensible circuit designers have the printed circuit layout in mind when they design the electronics.

but are you an adequate electronic circuit designer? Pretty much everybody can work out what being a good farmer, goat herder or coal miner looks like. Good farmers produce bigger heaps of food, good goat herds have big herds of healthy looking goats, and good coal miners bring back lots of wagons full of coal.

Good electronic circuit designers put equipment together that is more compact, faster cheaper and more reliable than their competitors. But most people just see a black box that either works or doesn't. From time to time I've cleaned up legacy designs, but every time I've done it has been because the old design used parts that were no longer available. The fact that the old design wasn't all that good never came into it.

Maybe people give us purchase orders and awards because I'm so cute.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

We had a quick design review, as we always do before we release a PCB to fab.

Of course, we wasted the usual 15 or so minutes getting Teams to work for everyone.

Someone else didn't like the four big mounting holes each having three wimpy spokes to ground, so I paved them over with big copper-pour circles on layer 2.

I think JLC has a min order of 3 panels, which will be 12 boards. They are so cheap that having a few extra is fine. Only one of their

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

at that size why bother with panels and v-scores? just order separate milled out boards

My production people like panelized boards with breakaway edge strips for the machines to grab, and fiducials and stuff like that. They have time to build these so we'll make them happy.

We can buy loose boards if they are simple and reasonable to build by hand.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Or perhaps because you give them what they think they want, rather than what they need. One guy who ended up as my boss for a bit had driven his initially very successful company into bankruptcy doing exactly that. The LTSpice guy early in his career worked for the company that drove that boss into bankruptcy,merely by supplying a version of the same product that was easier to use and more reliable in operation.

Then why do they come back for more?

Electronic design is the ultimate reality check: it has to work.

The huge nasty physics thread above is hilarioius. Most of those guys couldn't build a doghouse.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Because you give them what they think they want.

Sadly, just working, and working well and reliably aren't necessarily the same thing. My exercises in cleaning up legacy designs really did make that clear to me.

They can't recognise a spurious theory when they see it. Building a dog-house is a different skill set. Your incapacity to recognise climate change denial propaganda could be seen as equally comical, but when Donald Trump is equally gullible and takes deliberate action that will make for more anthropogenic global warming, amusement has to give way to alarm. Plausible nonsense has been responsible for quite a lot of death and misery over the years, and when the US education system doesn't seem to teach people how to recognise and reject it, the country has a real problem. Donald Trump's claimed "common sense" may be widely shared but that isn't a good reason for taking him seriously.

One can understand why people whose businesses depend on manipulating consumer demand with implausible advertising don't want the US education system to spend time on training kids to recognise nonsense as nonsense.

They don't seem to appreciate that this defect has a downside.

Unless a design is fundamentally botched, it's not usually difficult to see what's wrong and fix it.

The thing that needs guts is to really redesign it right, bottom up. The sunk-cost emotional issues often prevent that.

Laptops are cheaper than lab benches. TAs are cheaper than profs.

Long-term, market forces sort all that out.

Short-term, things can wobble a lot. Right now around here, hardware designers and PCB layout people are in demand and are respected as magicians. Coders are plenty available, cheaper in bulk.

We were just working with an FPGA designer, programming a chip for a simple product. We showed him the board schematic and he basically panicked. He can't read schematics. It's that text-brain vs pictures-brain thing, I guess.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

John Larkin wrote: |---------------------------| |"Most of those guys | |couldn't build a doghouse."| |---------------------------|

"One way of thinking about the semantics of math that was based on logic, is as interlocking gears. Everything kind of has to fit together, and if it does, and everything is compatible at the end, you get the final turning of the shaft that you want.

An analogy to these programs of the sixties is a dog house. If you take any random boards, nail, and hammer; pound them together and you've got a structure that will stay up. You don't have to know anything, except how to pound a nail to do that. Now, somebody could come along and look at this dog house and say, Wow! If we could just expand that by a factor of a hundred we could make ourselves a cathedral. It's about three feet high. That would give us something thirty stories high, and that would be really impressive. We could get a lot of people in there. The carpenters would set to work blowing this thing up by a factor of a hundred. Now, we all know, being engineers and scientists, that when you blow something up by a factor of a hundred, its mass goes up by a factor of a million, and its strength, which is mostly due to cross sections of things, only goes up by a factor of ten thousand. When you blow something up [by] a factor of a hundred, it gets by a factor of hundred weaker in its ability, and in fact, what will happen to this dog house; it would just collapse into a pile of rubble. Then there are two choices you can have when that happens. The most popular one is to say, Well, that was what we were trying to do all along. [Laughter] Put more garbage on it, plaster it over with limestone, and say, Yes, we were really trying to do pyramids, not gothic cathedrals. That, in fact accounts for much of the structure of modern operating systems today. [Laughter and applause]" says

Cf. @inproceedings{Kay_against_C++, author = {Kay, Alan Curtis}, title = {{The Computer Revolution Hasn't Happened Yet.}}, booktitle = {{The 12th ACM SIGPLAN conference on Object-oriented programming, systems, languages, and applications.}}, year = {1997}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, note = {\url{

(S.

All of which profundities leaves the dog out in the rain.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

This is because software relies on operating systems and software is always looking to leverage the past instead of abandoning it and starting over from scratch. With hardware designs, even the second VERSION of a design can be completely independant of the first as there is much less effort invested. Almost NEVER the case with version 2 of a software product.

Look at the folks designing with 50 year old OS technologies at the core of their products, completely ignorant of the advances that have occurred to improve the quality of those designs. Usually, because the folks making design decisions are terrified of a "blank sheet of paper" (which you face each time you start drawing a *schematic*)

Has anyone designed a new OS in the last 35 years?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

I know of one. Its proprietary and is used in certain network routers.

John