Discussion on the Problems and Inefficiencies seen with Hardware Bring-up

[dday]

Hello all,

I am an electrical engineering undergraduate currently in the process of starting my senior design project (capstone) with an idea based around solving test and debug inefficiencies in the hardware bring-up realm.

I’d like to start a conversation about what really grinds your gears at your workplace, or on your downtime at home, while testing, debugging and validating new board designs. Any feedback is greatly appreciated!

To get things started here are some of the questions I'd really like to be discussed:

• What type of information (and medium) do you have about a board while you are testing it?
• Through your experience, what points in your testing workflow do you see as inefficient?
• What is the biggest problem you’ve faced while performing bring-up on a new PCB?
• What parts of your PCB design software did you find most helpful with bringing up a new board?

Thank you in advance to anyone that can provide feedback on this subject as it will really help my capstone team narrow what to build and help validate some of our potential ideas.

Cheers,
Darryl

[Bassman59]

Hello all,

I am an electrical engineering undergraduate currently in the process of starting my senior design project (capstone) with an idea based around solving test and debug inefficiencies in the hardware bring-up realm.

I’d like to start a conversation about what really grinds your gears at your workplace, or on your downtime at home, while testing, debugging and validating new board designs. Any feedback is greatly appreciated!

To get things started here are some of the questions I'd really like to be discussed:

OK, I'll bite.

What type of information (and medium) do you have about a board while you are testing it?

I have all of the schematics in their native CAD form, as well as the software needed to look at it, and also printed copies. I have the layout in its native CAD form, too. I have all of the firmware sources and the tools needed to build binaries that get loaded into the programmable devices. I also have all programming dongles needed.

Through your experience, what points in your testing workflow do you see as inefficient?

Lack of test points, or access to test points. Usually this is just a function of the mechanical design, as in "the ADC I need to check is on the bottom of a board that's in a vacuum chamber."

What is the biggest problem you’ve faced while performing bring-up on a new PCB?

Well, we don't call the initial power-on the "smoke test" for nothin'!

Often the reason something doesn't work is because there was an assembly error. The wrong resistor value was stuffed, or a part that should've been installed wasn't, or there's a problem with bad solder joints and pins not connected to the PCB pads.

What parts of your PCB design software did you find most helpful with bringing up a new board?

Usually, all I use the PCB tools for in bring-up is to help me find points to monitor with the 'scope or the voltmeter. So I'll have the schematic and layout up on the computer, and I'll click on the schematic part I'm interested in, and the footprint or net is highlighted on the layout. (This is with Altium and Kicad, I assume others have the same feature.)

[dday]

Thank you Bassman59 for the feedback!

It seems like it's common practice for people to print their board schematics on multiple or large pieces of paper. You mention that you also have the schematic open on the laptop, why do you need the printed version for then?

You also mention that you like how in the EDA tools when you click on a net or component in the schematic it then highlights it in the layout viewer. On larger and more complex boards, do you find it still to be troublesome sometimes to match up where you are looking at in the layout with where that is on the physical board? For example, if a pin is highlighted somewhere in the middle of say a 120 pin connector, do you have trouble finding it?

[Bassman59]

Thank you Bassman59 for the feedback!

It seems like it's common practice for people to print their board schematics on multiple or large pieces of paper. You mention that you also have the schematic open on the laptop, why do you need the printed version for then?

The printed version gets marked up! Despite years of effort by various software companies to make electronic note-taking work reasonably well, it’s just easier to write on the paper schematic. Let’s see, do I print the schematic to a PDF and then mark up the PDF, or just print it out on 11”x17” paper and scribble on that with a pencil?

At some point the digital schematics are updated (new revisions made, old never lost) with the results of the tests and scribbles. The marked-up schematic gets put into a folder along with other contemporaneous notes.

(NB: we do B-size schematics. Reducing a D-size page to B-size makes the result unreadable. I haven’t see a D-size printer/plotter in 20 years. B-size sheets sometimes means more pages, but so what.)

You also mention that you like how in the EDA tools when you click on a net or component in the schematic it then highlights it in the layout viewer. On larger and more complex boards, do you find it still to be troublesome sometimes to match up where you are looking at in the layout with where that is on the physical board? For example, if a pin is highlighted somewhere in the middle of say a 120 pin connector, do you have trouble finding it?

Seems like our designs keep shrinking in size (cramming more parts into a smaller package), so yeah, it’s sometimes hard to find the pin. But we all have magnifying lamps on our desks, and Panavise jigs to hold boards in place. Visibililty is important! With experience you figure out how to put your probe on the right pin. Things like, “there are three vias, then an empty area, then one more via, and the pin I want is next to that empty area.” And you get adept at holding two scope probes on two test points with one hand, while using your foot to hit “enter” on the computer keyboard so the thing does something interesting.

Other things you should probably do:
* fan-out all BGA balls to vias, even if the balls are no-connects. You’ll thank yourself when you need to add a green wire. Sometimes this involves convincing the layout tool to do a trace on a one-connection net.
* for first-article runs, don’t tent those vias. Having no solder mask on them makes it easier to probe and add the bodge wires.

[dday]

Thanks again for the response Bassman59! This feedback was very helpful!