I would have them fix up the schematic to make it human- readable.

Seconded. There's at least one major fault in the schematic that I suspect will stop this working. Did they supply a parts list/BOM with JLCPCB or PCBWay part numbers? If so, post it to confirm my suspicion.

Absolutely awful PCB layout, IMHO. The WiFi signal is likely to be severely compromised by that central location, especially as the display cable appears to exit the FPC connector directly over it.

Although the routing is basically OK (I've not checked all the detail) it smacks of a library of pre-routed elements just dumped on the PCB with only a cursory thought about the wider aspect. See screen shot below of the on-board buck/boost converter - the layout is probably OK, but its not great. I personally always put the job #/name, version, client name/ref on the front and rear silk screens, but there's nothing here. There is a distinct lack of thermal vias between top and bottom layers. I've not checked thermal footprint so there may be enough copper on just the top ground-plane, but normally Tja for most chips assumes/states heatsink area based on both sides of the PCB. I found, I believe, at least one hidden via - one that doesn't go all the way through. My understanding is that these are not supported by JLCPCB or PCBWay for low-cost prototypes.

 

The WiFi signal is likely to be severely compromised by that central location, especially as the display cable appears to exit the FPC connector directly over it.

Thanks for the detailed response, it helps a lot.
Is that the kind of thing I should ask for a revision for? (meaning specifically, about the wifi)

There was a BOM, I'll attach whole folder here.

 

Thanks for the detailed response, it helps a lot.
Is that the kind of thing I should ask for a revision for? (meaning specifically, about the wifi)

There was a BOM, I'll attach whole folder here.

I should also note, I'm not looking for a robust design right now.
Just something that works and won't break after a few uses.

Needing something to test my software out on that can use buttons / speakers / nfc / bluetooth

 

I should also note, I'm not looking for a robust design right now.
Just something that works and won't break after a few uses.

Needing something to test my software out on that can use buttons / speakers / nfc / bluetooth

Well, as of now this won't work due to a fundamental inability to read a datasheet.

I always supply my clients with the actual schematic and PCB files so they can, if they want, revise it themselves - unless they bought a turnkey service which is end-to-end support.

Right now, you are stuck with getting this individual to figure out what's wrong and fix it.

 

Well, as of now this won't work due to a fundamental inability to read a datasheet.

I always supply my clients with the actual schematic and PCB files so they can, if they want, revise it themselves - unless they bought a turnkey service which is end-to-end support.

Right now, you are stuck with getting this individual to figure out what's wrong and fix it.

Oh, I mean asking the original designer to do a revision.
Would this be difficult for them to redo?

 

It's just for prototyping, so I can test the software out on a physical device, and also to just generally have a rough design of the final product. Is it normal to find someone to design the hardware + casing?

You need one does the PCB design and another one for the case design, for prototype the first maybe can do or no, depends on if he has the tools for prototyping or only do design and pcb work
You can try upwork, there are many freelance guys there

 

There are two MAX98357A pcm–class d amp chips. As far as I can tell, ALL of the connections to the ESP32 are the same. A minute of detailed analysis says the chip can use the left channel data, right channel data or both, depending on the connections of the /SD pin. I'm guessing you want a stereo output since you have two chips. With the two /SD pins in parallel, there's no way to tell one chip to use right channel data and the other to use left channel data. I'm not an expert on these chips, but this sure seems like an error.

 

Good catch, and yes, I think you're right since all of BCLK, LRCLK, /SD_MODE and DI are commoned. At least one of those, probably two, must be different!

That's in addition to the error I highlighted

I have a feeling this might have been drafted by AI, at least initially, as it lacks the nuances clearly identified in the datasheets.

 

So, if the audio is in mono and not stereo, it's not really what I was expecting with 2 speakers, but it's also not the end of the world.
But if the wifi is likely to break, that's kind of a big deal.

 

Oh, I mean asking the original designer to do a revision.
Would this be difficult for them to redo?

It shouldn't be, and they should do so for free. This isn't a revision, these are errors they should have dealt with before release.

Incidentally, the connections to the amps for GAIN and /SD_MODE are directly from the ESP32 GPIO. But these input are NOT strictly digital. You have to make some decisions about what gain and L/R signalling scheme you want before you can connect them up - you have a dozen or more options to be considered to set specific resistor values and configurations. That's something you have to understand and plan for and instruct your 'designer' what you want it to do.

 

if the audio is in mono and not stereo, it's not really what I was expecting with 2 speakers, but it's also not the end of the world.

It appears to be a simple matter of hardwiring the /SD_Mode pins to appropriate resistors....

BUT, if such a simple point was missed, there may be other more serious issues.

 

I think what we learn from this exercise is that "design me a PCB", unless incredibily simplistic requires effective collaboration between the procurer and the designer. "Boiler plate" (I want one of these and ... ) just doesn't work if design decisions need to be made that affect the PCB layout. In such situations it's incumbent on the designer to stop until those decisions are made, or at least communicate the default option. Picking a chip from a catalog without understanding it's nuances is a recipe for disaster. Equally, if the designer chooses a chip for a specific "standard" function, eg regulator, converter, level translator, etc., it's their responsibility that it is appropriate and configured correctly.

 

It appears to be a simple matter of hardwiring the /SD_Mode pins to appropriate resistors....

BUT, if such a simple point was missed, there may be other more serious issues.

Exactly so John. Yes, you can set certain aspects of the configuration of those chips with fixed resistors but those values are either set in concrete at assembly, or must be laid out on the pcb in a location and format that can be easily changed. And once those resistors are set that has knock on impacts on what protocols etc you can use.

Basically we have a situation where neither the TS' nor the 'designer' have any understanding of the chip or how to use it.

 

They did a revision, but maybe this one makes even less sense.

 

that makes no sense... each file contains all layers... so copper layers are all shorted.



top copper:

bottom copper:

 

btw, you can check it yourself, no software install needed. just unzip the files into a folder and load them in one pf online gerber viewers like
https://www.pcbgogo.com/GerberViewer.html

 

I posted the revision on a different forum, because I couldn't see any tracks on anything on the pcb.
Someone mentioned it was a export error, and sent this pic of the layout.
At this point... I'm just trying to figure out if I should get a refund or not.

 

Sad to say, but the fundamental error I identified initially which will stop this board working is still present and clearly hasn't been touched.


Refund? He should be paying you for the time you've wasted on this! If you go to manufacture on these boards you'll be throwing money away.

 

once you get your money back, share this with him:

 

I'm developing a bit of software for a microcontroller, and I need help designing a pcb layout for prototyping (I have a list of components I want to be included on the board, like a certain mcu / screen / buttons etc..).
But I also want a design for a casing to go around the device.
So my question is, how would you go about freelancing this task out, when there's both electronic + casing to be considered in the design?
(This is totally out of my realm of expertise)

I think in the first place, you need to look for somebody who is experienced in both PCB designing and 3D designing. If you cannot find one who knows both. If you cannot find one person for both, hire a PCB designer and a mechanical designer who can coordinate together. PCB and enclosure are tightly linked, so they should not be treated as completely separate stages.

From whoever you get the PCB work done, you need to receive the full PCB layout along with schematics and the original editable design files. I do not know how many layers of PCB you need to design. If you design a 2-layer PCB, tell the designer to submit you the full design files along with Gerber files, drill files, BOM, pick-and-place files (if assembly is needed), and also preview PDFs for top copper, bottom copper and top silk in separate formats for reference. When you have these, you can manufacture your prototype from NextPCB or anywhere else depending on which country you live in.

In case of enclosure, look for someone who is experienced in 3D design software like SolidWorks, FreeCAD or AutoCAD. You should get not only the STL file of the design, but also the STEP file and editable CAD source files so it can be modified later if needed. Ideally the enclosure designer should work from an exported STEP model of the PCB assembly so everything aligns properly. Then you can get the enclosure printed from a 3D printing company.