Review of first 4-layer PCB layout

[crustyAuklet]

I would appreciate any feedback on the attached PCB. It is my first moderately complex 4-layer, working with a size constraint of 30mm X 70mm. It is a data-logger with GPS, IMU, Light sensor, and Temperature sensor attached to a SAML21 running at 12MHZ and logging to an SD card or SPI flash chips. This is a prototype so there are large test points on the bottom and two different memory interfaces (deciding between SD card or daughter board with SPI NAND).

there is not much silkscreen due to size constraints so I colored the nets. Bright red VCC (3V3) while pink is the USB-VBUS and rusty red is the positive battery terminal. Dark blue is GND. Light green traces are I2C buses running at 400KHz, light blue is SPI buses at 1MHz.

The main questions I have are:

• The two power circuits at the bottom of the board. The left one is a Lipo charge IC (MCP73831T) and the right one is a buck-boost voltage regulator (TPS63051RMWR). I tried to follow the suggested layout as closely as possible. Since this is a 4-layer board I have solid VCC and GND planes on the two internal layers. Should I expand the pours around these power circuits? are there enough vias connecting the existing pours to the inner layers?
• Should I have more pours in general on the outer layers? I was avoiding external pours to avoid more vias stitching it all together, and possibility of loops. I ran into my fab house drill limit on a previous board and I wanted to avoid that.
• The IMU, light, and temperature ICs are QFN parts, with a fairly small pitch (especially the IMU). I had issues soldering on previous boards with these chips, and so I removed the solder mask from the traces just outside the chips, with the idea that is would make rework easier since I could get an iron and some flux in the reflow a stubborn pad.

Generally anything else I may have messed up and am unaware of? I attached my "top" schematic sheet that contains all the power stuff, and a printout of TOP and BOTTOM with colors as described before.

thanks!

[thm_w]

You can expand the pour if you think temperature will be an issue. Top/bottom pour is not necessary if you don't want to deal with setting it up, that is fine.
The length matching is overkill but ok.

Extending mask is ok, usually I would do this with the footprint pads so you don't need to remember to re-do it each time. Assuming you will only be hand assembling the boards.
You can also look into a finer soldering iron tip if you do not have one yet.

[crustyAuklet]

Thanks!

by "setting up" you mean stitching any outer GND pours the the internal plane with vias right? I may make them slightly larger, I am not super worried about heat but the whole thing will be potted so there isn't a lot of opportunity for dissipation.

I extended the solder mask out on the pads but NOT the paste mask. do you think this will be an issue? I am doing the first run by hand, but using an assembly service after that. I am slightly worried that extending the paste mask will result in too much solder, but ALSO worried about not extending them causing weak connections as the solder spreads over the entire pad.

[thm_w]

Thanks!

by "setting up" you mean stitching any outer GND pours the the internal plane with vias right? I may make them slightly larger, I am not super worried about heat but the whole thing will be potted so there isn't a lot of opportunity for dissipation.

Making them larger, if its potted heat will just go into the potting and more heat going inside the board with vias might not make much difference. If you aren't worried can leave it.

I extended the solder mask out on the pads but NOT the paste mask. do you think this will be an issue? I am doing the first run by hand, but using an assembly service after that. I am slightly worried that extending the paste mask will result in too much solder, but ALSO worried about not extending them causing weak connections as the solder spreads over the entire pad.

I would tend to go for larger paste mask, but they can always tweak it if you see weak joints.

[diyaudio]

I saw this today, I think this is very neat work. well done.

[asmi]

Generally multilayer boards tend to have at least one ground plane and so they can dissipate a lot of heat (for that reason soldering anything to them in pain in the back), but I usually create a small ground pour on the opposite side and clear it from soldermask so that I will be able to fit a heatsink should thermals ever becomes an issue:


This board have two ground planes, so it's unlikely it will ever need a heatsink, but adding an option to install it doesn't cost anything extra, so I've done it anyway.

[In Vacuo Veritas]

12 MHz and delay tuning? 

[maginnovision]

12 MHz and delay tuning? 

It doesn't cost anything.

[stefanh]

The VBUS and USB traces are are very close to the mounting hole.  These could be possibly be damaged/shorted with the screw head (the circle indicates the screw head?)

You do not really need thermal relief on the stitching vias in the GND pours.  They would be better off as direct connect.

[In Vacuo Veritas]

12 MHz and delay tuning? 

It doesn't cost anything.

It adds routing time and PCB space. If this costs nothing, good for you.

It also costs false knowledge. You're just doing a cargo cult at that point. It's not engineering, it's not science.

[thm_w]

It adds routing time and PCB space. If this costs nothing, good for you.

It also costs false knowledge. You're just doing a cargo cult at that point. It's not engineering, it's not science.

Next time just say "its not necessary" instead of laughing at someone who is asking for assistance.

[maginnovision]

12 MHz and delay tuning? 

It doesn't cost anything.

It adds routing time and PCB space. If this costs nothing, good for you.

It also costs false knowledge. You're just doing a cargo cult at that point. It's not engineering, it's not science.

The routing time is typically negligible and so is the pcb space if it would change size at all without it. I'm not sure what the false knowledge is either. He may not need it but if he wanted all the relevant signals to arrive at the same time he did it the way you do it. It might be more useful to check the actual delay time to determine its necessity but that's not what you suggested. In fact you didn't suggest anything. Loads of help you are.

https://www.protoexpress.com/blog/signal-speed-propagation-delay-pcb-transmission-line/