PCB design review

[PL]

I need completely isolated DPDT relays. For 1 or 2 of them, they need rated for mains voltage ~240vac.

[Psi]

I can't afford to have them made and assembled to find out that there are critical errors.

Always test the PCB with a small sample before going to low volume production.  Don't bother getting PCBA done for a sample
Just get a panel or a few PCBs made using your final gerbers and then hand assemble them according to your final BOM.
Test and confirm everything is ok.

Skipping this step is something people do do, but only when they have many PCBs under their belt and have put it significant time checking over the PCB. Usually by more than 1 person. And even then, problems still slip through from time to time.

However, if you are contracting the development that maybe something they should do.  I dunno, depends on your agreement with them

[exmadscientist]

For more than one relay, consider a bjt relay driver. Takes care of the resistor biasing, and incorporates flyback diodes. One part, generic, common, inexpensive etc. eg. uln2003

https://www.mouser.com/datasheet/2/308/ULN2003A_D-1814978.pdf

Yes, I was looking for an example of mine to redact and share and then realized that I didn't have any that weren't done with ULN200x or NUD3xxx drivers! But the good old '2003 is harder to use than it looks. Most people follow most of the datasheets and connect its COM pin to the relay supply. This isn't great; as AoE chapter 1x reminds us, we actually like the voltage to spike up a bit when you stop driving the coil. So you want to stick a Zener diode inline with the supply going to COM, and allow this. I think there's one figure in one datasheet that shows this. The others all show the more basic arrangement.

[Psi]

I wouldn't criticize a schematic too badly when the criticisms are things that don't actually matter.
If they are errors, absolutely that is bad. And a badly layed out schematic or a schematic that is not very tidy is a bad sign.
but. There could be reasons for this.  eg, the contractor might be giving you a very good price and taking shortcuts on things that wont affect the final product. etc.

Not saying that's a good thing,  just that I wouldn't focus on things that don't matter.
The schematic doesn't really matter as long as it's functionally correct, the PCB design does matter.

[julian1]

I'm sorry, if I pull out my copy of AoE an entire stack of books will collapse. Super briefly... what's the reason for desiring a voltage spike on the rail? When I used uln2003, I just put 1uF decoupling cap right on the COM pin.

[PL]

If they are errors, absolutely that is bad. And a badly layed out schematic or a schematic that is not very tidy is a bad sign...

The schematic doesn't really matter as long as it's functionally correct, the PCB design does matter.

The errors have also showed up on the PCB traces, when EE said all was good to go. Gulp.

[exmadscientist]

I'm sorry, if I pull out my copy of AoE an entire stack of books will collapse. Super briefly... what's the reason for desiring a voltage spike on the rail? When I used uln2003, I just put 1uF decoupling cap right on the COM pin.

I'm, uh, not digging anything out from my own stack(s) right now, but from memory, allowing higher voltage here speeds up field collapse in the coil (by a surprising amount), making the whole thing open faster (by a surprising amount), which quenches any arc on the contacts faster and helps keep the contact material intact. This was very important to us to get right on a project with Very Special Relays whose deaths tended to be the main cause of service in previous versions of the unit. So we stuck that zener in there!

[Wilksey]

If they are errors, absolutely that is bad. And a badly layed out schematic or a schematic that is not very tidy is a bad sign...

The schematic doesn't really matter as long as it's functionally correct, the PCB design does matter.

The errors have also showed up on the PCB traces, when EE said all was good to go. Gulp.

He's done a PCB too?? 
Do share some snippets of PCB layout, I could do with a good laugh today.
Seriously, get someone else to do it, if his PCB is anything like his schematic and he takes just as much care(!) then it will be riddled with issues.