Connectivity issues in damp/wet weather

[Jerod]

Looking for some advice on how to handle connectivity issues in damp/wet weather.

We've found that our NRF51 device works well indoors, but when placed outdoors overnight, the ability to connect to the device can fall to near zero percent chance in the morning.

Notably, if we take the unit back inside to warm up/dry out, it goes back to connecting flawlessly within 10-15 minutes.

We're encased in a plastic water resistant housing, but certainly the damp/moist air can get at our system. We're connecting to the device using a mobile phone, from within 3 feet away. Most of our embedded connectivity parameters are "default" from Nordic examples, including transmitting at 0 dBm. We are using a PCB-based antenna.

The design we're using is actually the same as another "non-Nordic" solution where we've just replaced the processor core with the NRF51. This problem isn't seen on the "other" (non-Nordic based) PCB.

I had considered possible problems with the crystals, so I changed the code from using an external crystal to internal RC. This improved things quite a bit, but still hasn't proven fully capable. Some devices are connecting 1 in 3 attempts, others are not. iPhones went from zero percent chance of working to non-zero, but still far from perfect.

We've attempted to put the unit in the freezer, and didn't see the same problems.

Wondering what I should be looking at more closely, or attempting to try?

It's pretty hard to get a unit back in the lab to look at more closely when 10-15 minutes after being indoors it goes back to working state.

Any advice/suggestions would be greatly appreciated.

Cheers!

[AmbystomaLabs]

For starters you should get someone on your team familiar with RF development.

But here is the short list of things you need to check:

1. Running off the internal RC is an untenable idea at best. RF needs ppm accuracy not percentage accuracy. You need to key up the transmitter on a channel and check channel accuracy over your temp range with a spectrum analyzer on an accurate reference. If you identify something is wrong with the external 16MHz crystal then it must be fixed.

2. You need to verify your field strength at 1 meter. Easiest way to check this is beacon app on nRF and sniffer app on smartphone. Your design should be at least as good as the DK. This should always be done with the device in the final housing. If it's not right then go back to the bench with the VNA and get it right.

3. Water is massively lossy at 2.4GHz and highly reflective due to its high value for Er. If you are letting water get into your device it will detune the entire thing and will only be exacerbated by failures in tuning the RF properly. Also, long term you may have issues with the various types of corrosion that affect circuits with static DC voltages. There are conformal coatings but none that I know of with low loss at 2.4GHz.