How often does something you put together work first time?

[Dubbie]

My current project was a first for me. The assembled pcb worked first time, and what is more astonishing, the firmware I wrote while I waited for the boards, flashed and worked first time as well.  It was a relatively simple board with about 40 components and an stm32

I think the main thing that made my PCB strike rate go way up was The Fedeval Academy video about making boards that work first time. I have a checklist I run through now which catches most dumb errors.

[mavu]

Needs a 50/50 option, because it certainly feels like 50/50

But I suspect its less than that, because I'm always excessively happy when something works the first time

[BradC]

As you get older and gain more experience the percentage of things that work first time grows ever closer to 100% as you learn more and more about how things work.

I find this, but it's more than just learning about how things work. I also find the older I get the more theory I sort up-front, and the more considered my approach (read - slow).

I was astonished recently when lashing up an uC centric bit of kit after spending a day or so writing the prototype code to find that both the software and hardware worked first go. I shouldn't have been though. 20 years ago I would have done 40-50 crash and burn cycles putting it all together and getting it working. Now I spend that time drawing, documenting and reading about gotchas in the data sheets. Same time, same result, less solder and less time in front of the compiler.

[m98]

Most of my problems are actually weird manufacturing defects or some crucial hints burried deep down in a datasheet.
Just recently had a prototype where the uC won't respond to the programmer. Turned out that despite the otherwise beautiful solder joints, one pin of the uC wasn't properly soldered, the reset pin. That was literally the last thing I had suspected.

[basinstreetdesign]

   In 30 years it happened to me ONCE.

I was working for a small company when the marketers came to me with an idea for a gizmo they thought would be a good addition to the product line.  How long would it take to make a prototype of it so it could be demo-ed to a customer who also seemed to be interested, I was asked?  I thought for a few minutes and after a couple of doodles on one of the ubiquitous quad-ruled notepads, said I could have one working in about four weeks.  I figured one week to design the circuit, one week to hand wire the proto and two more weeks to debug it.  That sounded reasonable, at the time.  And fairly generous on the debug time.  The plan was, I would have the able assistance of a technician, John, who would make the prototype by hand, then we would both debug it together.

I spent the next week ruminating and drawing out a circuit design on paper (this was before PCs were a common feature of the desk-top landscape).  I tried my best to think of everything.  I gave it to John after I had given up trying to think how it could fail.  He went away and I didn't see him for the next few days.  When he came to me, late into the second week, I expected to hear about a long list of weird behaviours or even that it did nothing, but what he said made me nearly fall off my chair.  He said “It works!”.  “What do mean, 'It works.'?”, says I, flabbergasted.  “It works”, says he.  “Just the way it's supposed to.  In every mode!”.  I didn't believe him, of course, so I went to see it for myself and there it was doing what it was supposed to do and doing it very happily.

I only saw the unit that once.  After that it disappeared into the hands of the marketers.  I don't think it was ever manufactured.  However it did have a profound effect.

They started expecting me to do it again.    I became known as the guy who could design a machine from scratch, on paper and make it work with no debugging, on the first try.  For months after that I had to work against the tendency of the marketers and some management types to expect me to come up with prototypes in record time.  For me this was worse than being known as someone who's stuff was a little bit late.  (Not a lot late.)

Moral: Always use up your allotted design time even if you don't need it.

[Pack34]

Define "work". Can function to spec without any modification or could it require a small change, jumper wire, etc?

I would say that >95% of my work will function the first time but whenever you're doing something new there's always that small change that's needed before I would be comfortable with a production release. That's why I always prefer to next-day some bare boards, hand-pop, then I can find those little odds-and-ends that need to be polished before I can put in that order for 250 eaches.

[Cyberdragon]

Generally when something doesn't work as designed it is because there is something you didn't properly understand about what you were doing.

As you get older and gain more experience the percentage of things that work first time grows ever closer to 100% as you learn more and more about how things work.

You have to distinguish between designs (that are supposed to work) and experiments (where you don't know what should happen and you are exploring). However, it is often said that failed experiments are just as valuable as successful ones. Every experimental result increases your knowledge.

"Things work first time when you have more experience" and other lies we tell ourselves.

Then there's times when you have some intuition and just fiddle with something and accidentally get it to work yet you still don't fully understand what's going on.

It's like those mechanical puzzles when you accidentally open it and go "what did I do?"

[Leiothrix]

Talking about software here as that's what I do more of:

Sometimes I'll write a page or two of code that does something.  Then it compiles and runs perfectly and the program does exactly what it's supposed to.

That always makes me suspicious of what I've missed rather than happy though 

[james_s]

It depends on how you define "works". Most of my projects these days do something the first time, but for a new design there's almost always some tweak I need to make. On a bad day I realize I swapped a couple of the pin names when creating the schematic symbol and that issue went all the way into the first run of PCBs. More often I find I need to tweak the value of a couple resistors. Occasionally a project is simple enough that it all just works right way but then if any sort of software is involved my code almost never works on the very first try. Always there's at least one typo or some other bug. I've never met anyone who could hammer out a whole program and have it work right off the bat without any bugs though.