First "real" Project

[ThoWaBu]

Hello,

I am Thomas - an hobby "EE" from Germany.
So my writing goes difficult.

All my projects only build on perfboards (lochraster) but still working.
These projects are simple steering and sometimes redistribut power from another ECU to the load.

Now I change to an more professional design.

I went up with EAGLE and an PCB with SMD devices.
I calculate the thermal worst case...
... redesign the board many times.
Compare to the prefboard.
Take care to get it more general,
implement some details...

But I think always I forgot something important... !
The fact to do/get a "profesional" Design is coupled on costs,
which pressurize me.

Is an design fully planable ?
When is it done, ready to produce ?

How do you deal with that ?

[Zero999]

But I think always I forgot something important... !

Yes, you forgot to post the schematic along with a basic description of the design.

What do you need help with?

[DJPhil]

Greetings
Don't worry about difficult writing too much, if in doubt it's ok to lean on google translate.

I think the general question is (correct me if I'm wrong): After much testing, how do you choose when a design is complete and ready for production?
I'm not sure what scale you're dealing in, so I'll answer for small to medium production first.

I think the real key for the electrical design part is testing an accurate physical prototype. Build at least one prototype that's as much like the final product as you can make it, and test voltages, currents, temperatures, noise, and performance. Try to test in environments similar to those the device will be used in. There's no good substitute for a thorough 'Dave-ing' (Dave, you are now a verb, woohoo!) to make all the bugs and weaknesses stand out. You won't always need to go this far, but be advised that a lot can change going from lochraster to PCB. The good news is that it's mostly for the better! A well made PCB will generally have much less noise emitted and received and a much smaller board. Probably the biggest downside is that you can't take the current capacity of traces for granted if you're working with small traces or thin copper. With point to point wiring it's easy to just assume everything up to a few amps is fine.

For the non-electrical, end user side of the question it's hard to answer well without knowing the market and the scale of your project. How many are you making? What's the worst case if the device fails, i.e. do you have to worry about high current causing fire or a malfunction making a vehicle inoperable? What level of skill can you expect of the user, i.e. can you be fairly certain they'll use it properly and be able to troubleshoot?

If I had a project I was selling I'd build two or three prototypes and test the heck out of them, perhaps even test them to destruction. Once you're sure your design is solid you're left trying to think of all the insane things people might do, and you'll probably have to draw the line of expected use somewhere. Dealing with a hobbyist market might be a little easier than dealing with the general public, as hobbyists are likely to read and understand instructions and warnings.

All this changes if you're just talking about one board for using yourself. Personally I etch my own boards, and I can imagine how stressful it would be if I started out working with a PCB house. I'd be nervous to commit to that final design and spend the money! I think the best way around that would be to post your design here in the forums. You'd have the benefit of many experienced designers to help point out any problems before you spend your money. It's a good sign that you've refined your design several times and worked out the thermal details.

I hope this helps some.

[Kiriakos-GR]

Welcome Germany ...   

Greece salutes you ...   I can not help at this inquiry of yours , but we have lots of trained members so to help you.

[DJPhil]

Thanks for greetings !

Interessting who do you meet around the world 

We have quite the international family here, I'm always amazed.

The full truth:
...
Thats for short.

Wow! I think I understand. Allow me to repeat back to be sure. You've designed a device to change brake light and blinker signal information from any vehicle so it will meet German automotive standards. I don't quite understand the details, but it looks as though you have the basic electrical circuit solidly developed.

I see why you'd want to move to a PCB, I'd feel much better about reliability over having a perfboard (Lochraster) circuit controlling auto lights! Eagle is a highly recommended EDA, but you might consider trying KiCad. It's a bit different, but the components and footprints are separate, which can save some sanity. It's understandable if you don't want to redo everything, just a thought. I know it's no fun to have to relearn everything when you're starting. KiCad can output Gerber and drill files, though I've never personally tried to make them work with a PCB manufacturer. Others may know more.

Redoing the thermal calculations shouldn't be too bad. I'm not sure which manufacturer produced the VN820 you have, but ST's datasheet has some info on PCB copper plane cooling on page 17. Remember to plan for worst case conditions, for example a hot car on a hot day in the desert, with the brake lights on indefinitely.

I have a consideration for you regarding the circuit protection, but I'm not sure I understand enough to know if you've covered it already. Be sure that the current limiting in the drivers (or anywhere in the circuit) isn't creating a situation where a fuse can't blow to protect the wires from a short circuit. If there's a short in a bulb socket, for example, and the current limiting kicks in, you would probably not want it to hold nine amps on a ten amp fused wire indefinitely until the driver decides to have the burned out bulb looked at (if ever). If all the high current is handled by a relay and this circuit just controls the relay, then naturally you don't have to worry as much.

You might also consider using a conformal coating on the final design to protect against dirt, moisture, and the like. You can sometimes get this as an option from PCB manufacturers, or if you like you can buy spray cans and put it on yourself. It will help with durability, but remember that coating everything in plastic traps heat in the components. This is sometimes a nasty surprise and it can even be a factor with through hole resistors, as they radiate most of their waste heat from their leads. Be sure to plan for this in the thermal design, and be sure not to spray your heatsink by accident!

I will post my schematic and board soon.
It would be an pleasure to get it knocked down by you professionals.

To paraphrase Dave, negative feedback is the most useful kind.
I should add though that I'm not a professional myself. I'm just a hobbyist, and just about everyone here has more experience than I do!

[DJPhil]

By the way. No feedback is the worst....

I'd meant to return earlier, my apologies. Life has been very distracting lately.
Everything looks good to me, but I don't have much experience, so use caution with my opinions!
I don't have any experience with Eagle, so I may be pointing out things that are just fine if I only knew how to read the layout properly.
All my advice is therefore really basic, so please don't be offended if I'm stating the obvious.

(Each Pin ist designed to source 3A -- Lamps drain 2.xxxA)

If you'd like to double check your trace widths for current capacity here's a handy calculator I found some time ago. You've probably done this already, but I put it up just in case.

I think the GND Polygon left from the µC could be bigger !?

It looks like you enlarged it in the second set of pictures. I don't think it'll give you any trouble, but others might disagree.

I want to place thermal vias under the P2PAKs, can i use these standard open vias like done in the Polygons ?

I poked around a bit looking for info on thermal vias with Eagle. Looks like most diy folks just use standard vias, just make sure the thermal relief settings are how you want them (probably off). Thermal relief is usually used to make soldering to a large copper area easier, but using it on a thermal via would be counterproductive I think. If heat turns out to be a problem and you have the room in your enclosure you might consider using a low profile SMT heatsink, though it would add to your cost.

Got anyone experience in Soldering with them ?

Others here do, I'm sure, but I don't myself.

How big should I drill ?

I'm not exactly sure, but the same folks who have the pcb trace calculator above also have a via calculator with thermal resistance that might help.

Hope that does some good.

[sonicj]

But I think always I forgot something important... !

Yes, you forgot to post the schematic along with a basic description of the design.

lol!