Can I get some feedback, please?

[ssashton]

Hi!

No pun intended, I'd love to get some feedback on my PCB layout. This is my first attempt at a single GND plane for a mixed signal design.

This project is an audio DSP based around the Analog Devices ADAU1701. My past experience would have me 'separate' digital and analog ground planes and return them to the bulk smoothing cap > power GND by a thick trace. The reason being, I can run the dirty digital return closest to the bulk cap, return the analog further back and also join input signal ground on the analog plane near the op-amps.

I have been advised however, that a single GND plane is best if implemented well. So this is my first attempt to do so.

I hope that you can view the full layout with this link. If not let me know and I'll post some images.

https://easyeda.com/editor#id=1279b1681b4e447cbc0df04207c3e3e0

You can see the schematic pages here:

https://oshwlab.com/ssashton/dsp-adau1701-online

[aduinstat]

I didn't go through the whole thing in depth, but I have a few notes:
* I don't like the pin 1 indicators on U1. One of them intersects with the C6 silkscreen.
* U4 and U5 should have the same name (LM1117F-ADJ). Right now, each one is it's own line on the BOM.
* C17 and C20 will slow down U4 and U5. I'd get rid of them. Some people would even put a small capacitance across out and adj to make it respond faster. Too much capacitance on adj may drive the regulator to instability.
* I like to put fiducials in the corners, but it's probably not necessary.
* You may want to replace C19, C21, and C53 with tantalums. The ones you have chosen are piezoelectric. Linear regulators can also be picky about ESR, but I can't find much about that in the LM117 datasheet (but TI's datasheet says tantalum is ideal).
* You don't have have mounting holes, but I assume you have a solution.

[exmadscientist]

(If you post plain PDFs instead of forcing people to use that strange software, you might get better results.)

U4 and U5 will fail immediately if they are in fact LM1117Fs, as it seems to say. I stopped reading about there. I agree with aduinstat that you should check your capacitors.

[Terry Bites]

It looks like you've laid it out well but you wont know till you fist power it up.

[ssashton]

I thought that it should just open in your browser and allow full access to the layout and schematics. How was it for you?

Can you tell me why U4 and U5 will fail?

The linear regs are this - https://lcsc.com/product-detail/Linear-Voltage-Regulators-LDO_HTC-Korea-TAEJIN-Tech-LM1117F-ADJ_C126021.html

As far as I know basically an LM317.

[exmadscientist]

The 317 and 1117 are different parts. Sure, they're similar, but the 317 is more robust and the 1117 has a few fussiness issues (that makes me, personally, prefer to avoid it but I'm not dogmatic about that).

You don't quite have an 1117 there, your part has a few asterisks. Like VIN,max being 12V. Probably not the best choice for 15V inputs!

I stopped there because as soon as I see part labelling issues, I stop trusting anything. And since no-holds-barred reviews are my day job, I'm not really interested in doing that in off hours. Nothing personal. The circuit doesn't look crazy at all.

[strawberry]

1117 LDO (low dropout)
317 not LDO, easy to stabilize