'Universal' ESP32 footprint - okay to paste unused inner pads?

[HwAoRrDk]

I have a PCB design that uses an ESP32 module, but at present it's not clear whether a WROOM or WROVER module will be needed, so I have put a 'universal' footprint on the board (see attached image). Currently, a WROVER is going to be used because of the extra RAM, but it may turn out not to be needed.

When it comes to the solder paste stencil, I'm not sure what to do. Do I leave paste on the inner WROOM pads, or will it affect placement of the larger WROVER module? Or should I get two different stencils? (Yes, if careful I could probably make do with one and just not paste those pads, but it'll probably be tricky, and one mistake and you have to clean the excess paste off.)

[thinkfat]

Don't put paste on pads under the module. Once you heat up, the paste will not stick. It's likely that solder balls will start wandering around under the module and shorten other pads. Or the excess solder might make the module float around. I had the same problem with smaller parts, used a footprint with a center pad while the chip had none. Solder crept out, shortened some pins.

It is OK if you place the module by hand, or apply solder by hand. No-go for automation.

[HwAoRrDk]

Ah, okay. I had a feeling the unneeded solder might make the module float, but hadn't considered that the solder might become loose.

Guess I'll just have to have a second stencil made if in future we go with a WROOM module.

[HwAoRrDk]

Oh, and speaking of centre pads:

If I have a QFN part where the distance between the centre and the pin pads is pretty small, such that the solder mask will be just a tiny sliver (and probably outside PCB manufacturer's stated minimum), is it best to just remove the solder mask there entirely? Like as attached?

Or could I do something like make the centre pad smaller than the actual exposed pad area on the chip?

[thinkfat]

Oh, and speaking of centre pads:

If I have a QFN part where the distance between the centre and the pin pads is pretty small, such that the solder mask will be just a tiny sliver (and probably outside PCB manufacturer's stated minimum), is it best to just remove the solder mask there entirely? Like as attached?

Or could I do something like make the centre pad smaller than the actual exposed pad area on the chip?

I'd say, make the center pad according to the chip vendors recommended land pattern, or just stick to the IPC footprint. The PCB house will either notify you being out of their capabilities or they will remove the solder stop when processing the Gerbers. If you deviate from the recommended land pattern, you might have problems with the chip not centering properly on reflow. Strictly speaking, the solder mask is not really required there anyway, the center pad is well defined without it. You need it to stop outbound traces wicking away the solder alloy from their pads. If the center pad is only connected to a ground plane for heat dissipation, through thermal vias, there's nowhere for the solder to go, it will stick to the center pad of the chip and the PCB pad, surface tension should keep it there.

[HwAoRrDk]

Okay, great. I'll leave it as I showed.

I've not dealt with a QFN as small as this before (28-pin 4x4x0.4mm) so hadn't encountered issues with solder mask width, and wasn't totally sure what to do.