Designing my first "real" PCB, what should i take in consideration.

Go To Last Post
71 posts / 0 new

Pages

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know that this subject is sort of over but I'd like to give my thoughts on PCB layout, as well as clearing up something about Eagle.

 

Chuck99 wrote:
The silkscreen (tDocu) package outlines of capacitors and resistors are on top of the solder pads - a no no.

I have been told that any package details/outlines that will cover any pads should be placed in tDocu, while all the text/outlines that should be printed on the PCB(and never ever cover any pad) should be placed in tPlace?

I have etched all my boards at home so I wouldn't know but I thought I could decide freely which layers is used for what while generating the gerber files, or does the particular manufacturer control that?

 

As for PCB layout, one lesson I have learned early on is to be sure to not forget to think through the return path for all the currents, or the return currents which may have a bad impact on the function of the circuit.

I'm not sure but as I have come to understand this is that in digital designs as this this doesn't matter at all in comparison to some sensitive analog circuits, what do you say?

 

I had etched a board with a digitally adjustable(3-resistor feedback divider, with a DAC driving the third resistor) linear regulator stage, this was layed out before I even read about return current path's importance. the result was a reference voltage rail that was pretty much useless, I'm not sure what all the components on it where but there where heavy oscillations and what looked like lots and lots of random noise which in turn made the data-converters inaccurate. I can't be 100% sure that this was the correct explanation but given the ridiculous layout it seemed obvious as I had designed the reference buffer with care.

 

The lesson I drew from that was "Mind the return current path just as much as you mind the supply current path".

 

Another thing I will try out in my next board is buying SMD jumpers, metal things that looks like this _/¯¯\_ in order to avoid vias and still be able to overcome traces crossing each other.

 

As you guys seems to be experienced at making PCB's I'm sure you have figured out neat ways of making you own vias for you home etched boards.

 

Would you mind discussing the pro's vs con's regarding the choice to prototype your board at home first as opposed to having them made in china the first run?

 

I'm not sure but since this relates to the OP's situation this wouldn't be considered hijacking the thread?

There is a time of waiting right now anyway, I can't tell but I hope this is fine.

 

Regards

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A digital circuit in the real world is analog. Things like 'ground bounce' can cause problems. Your logic levels must have sufficient margin otherwise your circuit will be temperature sensitive and unreliable.
There's also transmission line effects for signals with fast edges.

You need to consider the pcb as a component in your circuit rather than just an interconnect.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Microzod wrote:
I have etched all my boards at home so I wouldn't know but I thought I could decide freely which layers is used for what while generating the gerber files, or does the particular manufacturer control that?

Atleast on my case i was granted the freedom of generating set of gerbers myself, i assume this is the case with alot manufactors? After all you get what you paid for.

Microzod wrote:
Another thing I will try out in my next board is buying SMD jumpers, metal things that looks like this _/¯¯\_ in order to avoid vias and still be able to overcome traces crossing each other.

0 ohm resistors could do the trick aswell, or atleast i came to that conclusion while searching on the subject.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

The silkscreen (tDocu) package outlines of capacitors and resistors are on top of the solder pads - a no no.

I have been told that any package details/outlines that will cover any pads should be placed in tDocu, while all the text/outlines that should be printed on the PCB(and never ever cover any pad) should be placed in tPlace?

 

That's correct.  In EAGLE when you generate gerbers, tDocu is NOT included in the silkscreen layers.  EAGLE libraries are FULL of tDocu cover pads.

 

Are resistor and capacitor smt pads for the same sized package actually supposed to be different?  Why?

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Good advices above. I'd also go for all parts on one side, and a spare status LED. 

It's been very helpful in my job to leave a couple of spare pins to be used as a debug port (preferably tied to an UART periph. or bit-banged otherwise), and hooked up to a computer with this adapter.

 

while(!solution) {patience--;}

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I got plenty of those usb-ttl converters, really handy gadgets to have in hand. Only proplem i have had with these is the drivers when you buy cheap chinese clones, profilic did something on newer drivers which windows will automatically load on these devices, that make them not function correctly if you dont revert drivers to older ones manually.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

True. Been there: hit Win + X, Device manager, re-push the 2008 driver from Prolific, get the COM port, etc.

If someone can suggest a tweak for Win8.1 to fix that, will be much appreciated.

while(!solution) {patience--;}

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Status update.
PCB's on the way to my place, will be one long weekend for me if they dont manage to deliver tomorrow, i dont have too high hopes about that thougth.

They did send a photo aswell, but as its taken with phone(?) hard to say anything about the quality yet.

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Microzod wrote:

 

Another thing I will try out in my next board is buying SMD jumpers, metal things that looks like this _/¯¯\_ in order to avoid vias and still be able to overcome traces crossing each other.

 

 

If you are getting a board made why would you avoid vias? 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Because Mr Zod makes his own boards.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The datasheet for the ULN2003 by Texas Instruments says "The ULx2003A  devices have a 2.7-k Ω series base resistor for each Darlington pair for operation directly with TTL or 5-V CMOS devices."

 

Try to reduce as much as possible the number of via holes in the design.   For example if the via from pin 11 were to be to the left of the trace from pin 23, the pin 23-trace would not need to cross the board with two vias.

 

Put lots of space between any ground plane and any pads for header pins.  They need to hold a lot of solder, and you don't want any difficult-to-detect bridges between the header pad and the ground plane.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

received the PCB's today and had time to place components for two of those, picture attached sorry for poor quality i dont have real camera at the moment to use. I will add good quality pictures from the PCB's whenever i have a change to borrow camera, in case someone is interested in the quality.

 

edit: Joni, I have reduced the size of your photo from 4MB to 78Kb because size costs a lot for some readers. Ross, Moderator

Last Edited: Mon. May 2, 2016 - 01:54 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Congrat's !  Your board looks great!

 

You'll have to report back on the "functional test" which is the next step in the process.

It determines if you have a great project... or a paper weight!

 

With the level of expertise you have demonstrated on your First Board, I would expect you will be laying out PC motherboards next week!

 

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:

Congrat's !  Your board looks great!

 

You'll have to report back on the "functional test" which is the next step in the process.

It determines if you have a great project... or a paper weight!

 

With the level of expertise you have demonstrated on your First Board, I would expect you will be laying out PC motherboards next week!

 

JC

 

Thx, we shall not forget all the great advices and tips i got on the road thought!

 

yeah about the testing, i was just coming to the fets and "boom" pwm/uart stopped working, oh well maybe i can figure it out. can still program/read the atmega so i dont think i did kill it, but something did happen.

Guess its time to go and get my scope from storage, i have too little space to keep everything at the "work" desk..

 

E: okay now its dead i assume, no connection anymore, hmmn.

 

E2: should be more carefully it seems, accidentally shorted my fet and the copper did fly out of whole input voltage circuit......well im glad the copper did fly out before my lipoly battery. guess im going to place components on new board tomorrow.

Last Edited: Mon. May 2, 2016 - 04:49 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

JoniS wrote:
accidentally shorted my fet

 

Ouch.

 

Had a project with a 3v processor and 5v display. Was just finishing when... I accidentally shorted the 5v supply to several processor pins. Time to cry, then remove the partly dead processor and replace.

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Get yourself a polyfuse and put it inline with your battery.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Today i had time to try with a new board, it did work fine for some time, i could connect to it, did change the fuses everything fine after that, started testing periephals pwm okay, uart decided it doesnt want to work so i did rewrite my uart init code(which have worked with mega168 on multiple small test projects) and at some point the MCU(?) did just fail as i could not connect to it anymore, as for now im clueless what should i do, to not destroy one more. I just dont understand how the uart did work on first one and not with this, and how c code can destroy the connection.

 

Someguy22 wrote:
Get yourself a polyfuse and put it inline with your battery.
 

Will do that for sure, ~250amps it can output on normal use is sure going to break something from the board if i fail with the fet again.

E: cause of all the errors/proplems migth have been identified, it seems that the 168pb which i got is not too happy to @16mhz in my circuit, will get normal 168 models and see if proplem persists

Last Edited: Tue. May 3, 2016 - 05:39 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Pcb design was succesful, as soon as replaced mega168pb with the ordinary mega168 everything is working as it should! Huge thanks to everyone helping with the PCB, now i can finally go forward with my projects.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is wise to check several times before ordering PCBs. Any potential problem could disable the whole boards, even the whole project, my experience.

All about PCB and electronics http://e-gather.ui3g.com/

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Try it on a small order from somebody like elecfreaks before you order a thousand wink

The largest known prime number: 282589933-1

It's easy to stop breaking the 10th commandment! Break the 8th instead. 

Pages