Best PCB Design Practices and Books/Articles

[c4757p]

How could anybody possibly insist that a book would be useless in PCB design? Sure, you don't become an expert PCB designer by reading a book, but it's not all "art". Metric fucktons of electromagnetics stuff and manufacturing stuff in there that can definitely be taught...

[Iliya]

This reply is off topic. The topic is Best PCB Design Practices and Books/Articles
Please post your views to the topic you created.
Thanks

Free-electron is (was?) in the process of writing a PCB design book. If you ask nicely perhaps he might share some of it with you!

Thanks, but I am interested in books and PDFs that exist. I personally do not need a book on Altium Designer. This topic is for existing information.

Judging by the grammar used in his posting, I would expect it to be very difficult to read.

I find Complete PCB Design Using OrCAD Capture and PCB Editor by Kraig Mitzner is a good book.

Perhaps he could post his draft for all to see as he is keen to most comments and videos of his ability to use DEX. This would be great and we could all add our comments and experience to make the book the standard PCB design book.

JohnnyBerg states:
Designing pcb's is a skill, you can't learn that from a book. Some may even say that it is art, and not science.
So the free-electron book is doomed to failure.

So , if pcb design can't be learned from a book why are you asking for books ?

And no, this is not an 'Altium' book . The book is written tool-less.

Maybe i should add a chapter about crappy pcb design software. I can use DEX as prime example. If you are going to flame me i will flame back ! You comments are way out of line .

Your reply is also off topic. 

Now go read the washlist of problems i posted in the review topic and watch the videos i made to document them. Fix em. And please do explain how to link one pcb stmbol to multiple schemaric symbols and the reverse , adde soldermask layers and paste mask layers so i can move on and try to get some proper parts made in DEX.

Please calm down and devote more of your time to your book.
I have changed my mind; please send me a copy of your book. It's OK is it's not finished. I would love to learn from you (can run a spelling checker and grammar checker on it for you). Washlist! Stmbol, adde

Thanks I advance.

[nctnico]

Designing pcb's is a skill, you can't learn that from a book. Some may even say that it is art, and not science.

There is a lot you can learn from a book about clearances, layer setups, creating parts, solderability, preparing for production, etc, etc. BUT creating an actual PCB is like solving a 3D maze and that definitely takes talent. I have worked with people who really struggled to get a board routed.

[Iliya]

Designing pcb's is a skill, you can't learn that from a book. Some may even say that it is art, and not science.

There is a lot you can learn from a book about clearances, layer setups, creating parts, solderability, preparing for production, etc, etc. BUT creating an actual PCB is like solving a 3D maze and that definitely takes talent. I have worked with people who really struggled to get a board routed.

As I said in one of my videos, manual routing is like maze solving. You need good visual skills and great peripheral vision as well as a sense of 2.5D (layer swapping).
Mathematically, routing is NP complete. That's why people bitch about other people’s routing.

However, sometimes the tool can help you get out of the Maze.
Here the yellow arrow is telling you the best end point.

[free_electron]

Sorry, no previews of the book. That is author-publisher privilege.

[Iliya]

Sorry, no previews of the book. That is author-publisher privilege.

What a shock.   Does it exist? I would have a bet with you but I don't know how I could collect my winnings.  Perhaps we could put the cash in escrow.

P.S. We all have plans to write a book.

[c4757p]

Don't be an ass. That's how publishing works. You're not just "writing a book", you're writing it for the publisher.

[Iliya]

Don't be an ass. That's how publishing works. You're not just "writing a book", you're writing it for the publisher.

You write books for the reader.
There are many ways to publish a book.
I suspect he doesn't have a publisher, perhaps he can tell us so we can all monitor it.  It was Dave who suggested I ask him. So is he an ass as well?

Please let's get back on topic.

[c4757p]

It looks to me as though he has no trouble finding a publisher.

And no, you're an ass for feeling the need to insinuate that there might not even be a book, despite being about a five second search away from verification that the man has already written multiple.

You write books for the reader.
There are many ways to publish a book.

Do these words mean anything in the real world?

[free_electron]

I don't know how I could collect my winnings. 

in your case i believe winnings is spelled with an h and a single n as opposed to a double n . that seems to be all you do.

Please go work on your non existing paste layer or solve a couple of my bug reports i filed, i'll go work on my non-existing book.

[free_electron]

Here the yellow arrow is telling you the best end point.

Which , if you had any know-how of PCB design, you would realize is wrong, for at least 3 obvious reasons. I'm sure i can find even a few more than that.
(hints why: sideways pad connections is not allowed because of bridging and AOI ,angle of attack is wrong and the only remaining channel is already occupied by a trace... )

And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

[Iliya]

Here the yellow arrow is telling you the best end point.

Which , if you had any know-how of PCB design, you would realize is wrong, for at least 3 obvious reasons. I'm sure i can find even a few more than that.
(hints why: sideways pad connections is not allowed because of bridging and AOI ,angle of attack is wrong and the only remaining channel is already occupied by a trace... )

And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

sideways pad connections is not allowed because of bridging
But that's the intension, 6 joins to 7!
Even it I routed vertically out from pad 6 there is nothing to bridge to!

With soldermask


And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

Uhh , you just did. You can always ignore the yellow hint arrow. Does Altium know that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading.

So, when's the book out?

[free_electron]

Yes altium can know that if i inject a net-tie object on the part that requires the kelvin connection. In that case altium will not let me tie it at the component. Net ties are made for that so you can create things like sense lines kelvin points or star points. Simlly drop where you need it and altium will do that. Like i said : i wont complain if dex cant do that. I will mention such a thing does exist but i wont comain if dex does not have that capability.

The connection you made in your first drawing is not allowed. Such commections lead to cracking of the copper and have a possibility to create false posisitves during AOI (automatic optical inspection) both during etch and post soldering. In case the mask needs modification due to the mask sliver to be too small then this stub of copper will wick away paste leading to bad solder joints and again failing aoi. These are manufactuirng problems that must, i repeat MUST, be solved during layout. The layouter meed to know this and the layout tool should not give 'bad ' suggestions. It is bad practice to route a board that way. Manufacturing terances , copper thick ess and kther factors may require that sliver of mask to disappear during production and then the problems begin. Such defects are caught in the DFM portion of the board layout process . dFM : design for manufacturability. There is also dft design for testability which involves creating testpoints for ate bed of nails or flying probe. That too is the task of the layouter and the cad tool must understand the aging rules like the probe pitch , detecting connections without testpads and other rules.

Good software has that understanding and provides the tools to the layouter to solve the dft and dfm requirements. During the design rule check the software must flag violations against dft and dfm rules. That is why we pay the big bucks for such software.

[free_electron]

Yes altium can know that if i inject a net-tie object on the part that requires the kelvin connection. In that case altium will not let me tie it at the component.

The connection you made in your first drawing is not allowed. Such commections lead to cracking of the copper and have a possibility to create false posisitves during AOI (automatic optical inspection) both during etch and post soldering. In case the mask needs modification due to the mask sliver to be too small then this stub of copper will wick away paste leading to bad solder joints and again failing aoi. These are manufactuirng problems that must, i repeat MUST, be solved during layout. The layouter meed to know this and the layout tool should not give 'bad ' suggestions.

As for when the book is out : that is not your worry. Actually it could be. I was hoping to give examples with DEx i the book.. If only i could make it work properly...

[nctnico]

The connection you made in your first drawing is not allowed.

It depends on the pitch but in general I agree. On a 1.27mm pitched SOIC you can get away with it but on anything with a smaller pitch the trace should exit on the long ends of the pad. Having a trace in between pads makes it hard to check for shorts and may cause soldering issues as well.
I don't mind a CAD not blocking this but it is something a PCB designer should be aware off.

[free_electron]

Indeed on an so pitch of 1.27 you can get away with it. My trained layouter eye spots a quad package with much smaller pitch so the alarm bell goes off immediately. It is bad practice no matter how you twist or turn it. Not done. Any pcb layouter will immediately comment on such layout structures

[Iliya]

The connection you made in your first drawing is not allowed.

But I did not complete it. I took the DEX hint and did it direct. See second drawing.

[Iliya]

The connection you made in your first drawing is not allowed.

It depends on the pitch but in general I agree. On a 1.27mm pitched SOIC you can get away with it but on anything with a smaller pitch the trace should exit on the long ends of the pad. Having a trace in between pads makes it hard to check for shorts and may cause soldering issues as well.
I don't mind a CAD not blocking this but it is something a PCB designer should be aware off.

I don't think the soldering issue on pin 5 would be much worse than pin 9. The routing was done with Electra.

[Iliya]

As for when the book is out : that is not your worry. Actually it could be. I was hoping to give examples with DEx i the book.. If only i could make it work properly...

I'm not worried, only interested. Mind you, it gives me an idea, I can do a book on PCB design with DEX. I can even ship it with DEX. (Which sadly you won't be able to). 
Went to my old University library today   Birmingham and picked up a pile of good books on PCB design. This topic really did not help. Seems nobody posting here really knows that many books on PCB design. At least one of Dave's links is broken.

[nctnico]

The connection you made in your first drawing is not allowed.

It depends on the pitch but in general I agree. On a 1.27mm pitched SOIC you can get away with it but on anything with a smaller pitch the trace should exit on the long ends of the pad. Having a trace in between pads makes it hard to check for shorts and may cause soldering issues as well.
I don't mind a CAD not blocking this but it is something a PCB designer should be aware off.

I don't think the soldering issue on pin 5 would be much worse than pin 9. The routing was done with Electra.

The problem isn't with pin 5 or 9. The problem is the direct connection between pins 6 & 7. If you route a trace between pads this way it looks like the board has a unintended short due to solder bridging between 6 & 7 after soldering. This is bad because it will trick people into thinking there is a soldering problem with the board OR ignore unintended shorts due to solder bridging. Always route connections between adjacent fine pitch pads from the long end of the pads.

[Iliya]

And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

Did a Google search on altium kelvin connection

Got this
http://premiereda-tech.blogspot.co.uk/2010/02/joining-nets-for-kelvin-connections-rf.html
I think they distribute Altium in the uk.



Click on play and got. 


How can this happen? Could do a free-electron review of this but my spelling checker won't let me.

Seems I'll have to look elsewhere. 

[Iliya]

The connection you made in your first drawing is not allowed.

It depends on the pitch but in general I agree. On a 1.27mm pitched SOIC you can get away with it but on anything with a smaller pitch the trace should exit on the long ends of the pad. Having a trace in between pads makes it hard to check for shorts and may cause soldering issues as well.
I don't mind a CAD not blocking this but it is something a PCB designer should be aware off.

I don't think the soldering issue on pin 5 would be much worse than pin 9. The routing was done with Electra.

The problem isn't with pin 5 or 9. The problem is the direct connection between pins 6 & 7. If you route a trace between pads this way it looks like the board has a unintended short due to solder bridging between 6 & 7 after soldering. This is bad because it will trick people into thinking there is a soldering problem with the board OR ignore unintended shorts due to solder bridging. Always route connections between adjacent fine pitch pads from the long end of the pads.

Sorry, I mean pin 6 and 7.

[free_electron]

The connection you made in your first drawing is not allowed.

But I did not complete it. I took the DEX hint and did it direct. See second drawing.

You dont seem to understand what we are talking about . The short vertical trace between pin 6 and 7 is what i am talking about. That is not allowed for the detailed reasons i listed.

Anyway you will mot learn this stuff from books , you need access to the standards from IPC and you will need tot talk to people doing board design for production. A lot of this stuff you learn by sending a design to manufacturing and having it shot down in flames. There are courses you can follow like the CID course from IPC but that is only the basics. It takes years to know all the intricacities.

There are some good courses at certain universisties like the emc course from mark montrose (which i did and aced it with top score)
Ather good books are the handbook of black magic (yes that is whats it called) and the books from the speeding edge: pcb layout done right the first time. There is emc for pcb design as well. I will post a list of the books i have laying around that deal with some of the complex stuff.

Pcb technology has evolved tremendously from when you did layout on rubylith. Accept that there is a whole bunch you do not know. That reflects on dex as well. I have listed several things that are minimum requirements to do a modern board. Stuff gets really complex once you start doing HDI boards . I just did a board that has 0402 parts with 4 mil track and gap as smallest feature and simultaneously traces that need to hold 100 ampere. Creating the structures for such currents alone is bery comex as you need to spread power vertically through the layers.
Via stitching rules for example, taking into account plating thickness , ohmic differences betwen ated and rolled copper. You cant ace voas too close together because the hole wall will warp. You need to know the weave of the material to guarantee enough strands and mot cause in hole delamination. All that stuff needs looking at and then design rules are set up so that once we start drawing we cannot violate any rule. Mechanical rules , placement rules , soldering rules. Electromagnetic rules testability rules and much more.
A good software can handle these conex rule sets and work interactively with the layouter to make sure no violations creep in.

[free_electron]

And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

Did a Google search on altium kelvin connection

Got this
http://premiereda-tech.blogspot.co.uk/2010/02/joining-nets-for-kelvin-connections-rf.html
I think they distribute Altium in the uk.



Click on play and got. 


How can this happen? Could do a free-electron review of this but my spelling checker won't let me.

Seems I'll have to look elsewhere. 

Look, your drivel is getting a bit long in the tooth... You seem to be good to spot the mistakes in someone elses stuff but you cant recognise the blatant mistakes and huge shortcomings of your own stuff.
Go on, read your books, write one yourself if you want to ,and go fix the problems in your software.

If you don't want any of my advice anymore then tell me, i can spend my time better than trying to help you improve dex.

[Iliya]

And i'm not even going to mention that this pin could be a sense wire requiring a kelvin connection to the node on the other end in which case the arrow should point to wherever that other track is leading. that is really advanced stuff which i don't expect DEX to cover. So i will not complain if it lacks that functionality or understanding.

Did a Google search on altium kelvin connection

Got this
http://premiereda-tech.blogspot.co.uk/2010/02/joining-nets-for-kelvin-connections-rf.html
I think they distribute Altium in the uk.



Click on play and got. 


How can this happen? Could do a free-electron review of this but my spelling checker won't let me.

Seems I'll have to look elsewhere. 

Look, your drivel is getting a bit long in the tooth... You seem to be good to spot the mistakes in someone elses stuff but you cant recognise the blatant mistakes and huge shortcomings of your own stuff.
Go on, read your books, write one yourself if you want to ,and go fix the problems in your software.

If you don't want any of my advice anymore then tell me, i can spend my time better than trying to help you improve dex.

I was following your advice and got the 404 error in my search. Appears Altium of all people screwed up.
Yes they do sell Altium.
http://www.eda.co.uk/
http://premiereda.blogspot.co.uk/
This is their Farnborough road show.


And this is their Edinburgh road show


This points out how you don't need to sell many Altium Designers to break even.

Anyway, I'll try elsewhere. Got a book to write.

Here's some advice for you
https://uk.answers.yahoo.com/question/index?qid=20101204063922AA93s6K