Rev 3 will be better

[NivagSwerdna]

If you cannot design... be good at bodging...



Have a good weekend!

[Benta]

Haha! We've all been there (and will be in the future).   

[NivagSwerdna]

Missed one 

[Miti]

If you cannot design... be good at bodging...

(Attachment Link)

Have a good weekend!

You’d be surprised, or not, but I’ve seen worse with very experienced designers on very complex boards. And those designers are still there and very valued after many, many years. If the board does the job, a couple of respins are normal and actually budgeted.

[coromonadalix]

Rev 3  of what  loll

[SiliconWizard]

Ahah, yes this happens to almost every engineer.

But but... still, I'm curious what you call "rev. 2" (current board) here. Was the rev. 2 already meant to fix issues you found in rev. 1, or was it an evolution with new "features", which introduced new "bugs"?

Because obviously, in the latter case, well again that's relatively frequent. But in the former case, that doesn't look very good and probably meant you rushed rev. 2 before thoroughly testing rev. 1.
Sure sometimes some issues you find may make it look like it's not worth it/or not even possible to test the board further, so you don't test it "thoroughly", but often this is still possible to test most of it with workarounds, and I tend to do that instead of multiplying the revisions. Of course that also depends on the cost and yield time.

[phil from seattle]

Green wires are always faster than a PCB turn!

As to V2 being rushed, is it possible that he discovered he needed to use Serial rather than I2C or some such?  Maybe a little benefit of the doubt is a good thing? I supposed we all should strive towards the perfection that some of us obviously already have.

[NivagSwerdna]

what you call "rev. 2" (current board) here.

It's a trick I learnt whilst working on product development at IBM... Nobody trusts V1.0 of anything so start your versions at 2.0 and nobody trusts .0 so 2.1 is always the first revision. True story! 

[SiliconWizard]

what you call "rev. 2" (current board) here.

It's a trick I learnt whilst working on product development at IBM... Nobody trusts V1.0 of anything so start your versions at 2.0 and nobody trusts .0 so 2.1 is always the first revision. True story! 

Lol, that's a good one. So your rev 2 was actually the first one?

[NivagSwerdna]

Lol, that's a good one. So your rev 2 was actually the first one?

2.1    First one

[SiliconWizard]

Lol, that's a good one. So your rev 2 was actually the first one?

2.1    First one

Ahah. Which shows that the trick doesn't work. The first version is bound to have bugs whatever you call it.

[NivagSwerdna]

OK I'm getting close now... and seriously considering a PCB revision.

(Note sure why this is 2.2... guess I changed my mind... this is my first PCB for this project)

[Vovk_Z]

It seems it was easier if you traced only power lines on PCB.
Those copper PCB wires only make a mess.

[NivagSwerdna]

It took me ages to work out my fundamental error... the connector on the RHS is wired correctly... but when I laid it out I made a 180 degree error! (confused by the fact the board this mounts on has this oriented the 'wrong' way).  It does work perfectly now.... so I will spin another board when I have a batch to send to China.

[b_force]

You know what's the worst?
When there is a serious error in the datasheet.
Omg, that meant spending WEEKS on troubleshooting and calling back a whole bunch of boards + wasting time bodging them all. 
(and being lucky to be able TO bodge them)

[TimFox]

Before retiring, my job had an engineering rule about how many wire mods could be tolerated before a new PCB was required.  What are the current rules on this?