nRF52832 BT832XE randomly disconnects

[mbates14]

So, after using my Devkit and everything working OK as far as I can tell, I decided to order up a bare BT832XE module and place it on my PCB. 

Flashed it with the same firmware inside SEGGER using the Nordic PCA10040 devkit. 

Seems to connect, everything works fine. for a little bit. Then I will get a random disconnect. It does not show up at all in the debug terminal. as far as it knows, it still thinks its connected. 

When I hit pause, it stops at DEADBEEF. With no rhyme or reason as to why. 

I tried recompiling as debug, and then tried again. Same thing. the disconnects are random. it can happen 30 seconds in, and it can happen as late as 4 to 8 minutes. But it ALWAYS happens. 

I tried adjusting the Clock to 500 with its internal oscillator (not using a crystal externally). No difference. I even tried adding extra bypass capacitance near the module on my PCB, the voltage going in appears to be stable so thats not it. 

Any thoughts? 

[mbates14]

Nope, nevermind. Yet another fallacy. I cant track down the problem at all. Its totally random, I disabled the communication error traps, and that fixed my crashing when grounding the RX pin, but it still did NOT fix my problem. I can simply just pickup the PCB and wave it around, and itll disconnect. But I can twist on the board, and it doesnt. I really dont get it. 

[Mojo]

May I suggest to try a different module.  Like this one https://www.crowdsupply.com/i-syst/blyst-nano

[mbates14]

Wont work. Not enough output power. the +19 to +20dbm that the BT832XE produces is crucial. 

[mbates14]

Any critical pull-ups or anything that are required that I may have overlooked? 

[run_ar]

I assume the module manufacturer has a reference layout for their module and that you have followed that? You could also check against the reference layout in nRF52832 product specification against the module.

But are you sure that you everything is soldered correctly and that you do not have additional pieces of wire/shorts or maybe poor soldering connections?

Just so it is clear, this only happens on the custom board? Not the nordic DK?

[mbates14]

I think I narrowed it down to the power supply. I used a 6V 300mA that was going through a 3.3V regulator. But I changed it out with a 9V 2A and the problem is gone. Funny thing though, my "power led" stayed green the whole time, but I assume an LED is not going to react to spikes in the DC bus. . 

[run_ar]

I assume you must have been hitting this: A step increase in supply voltage of 300 mV or more, with rise time of 300 ms or less, within the valid supply range, may result in a system reset. ref: Power-on reset

[mbates14]

Yea, Probably so. I didnt know that. Almost sounds like an anti-glitch, glitch detector of sorts. I even tried adding a nice large capacitor right off the BT module, but it still didnt help. 

Once I changed out the power supply completely though, all was fine.