WH-3080-SOLAR new take on radio dropout
Posted: Thu 15 Nov 2018 12:54 pm
I thought I'd share my finding.
Like so many others, I have had serious problems with wireless comms. I moved from a 433 system to 868 because of the interference from next door's wireless security system, so I was dismayed to find the initially good 868 system gradually deteriorating. I have even brought out the console's down button wiring so that my PC could detect loss of signal and automatically force a re-acquisition. When I was losing signal for over 12 hours each day, I thought I should do some proper investigation.
Firstly, my sender is about 3m up the mast. I've extended the wiring for the sensors, so that I could put the transmitter as close as possible to the receiver, low on the mast. I use two consoles, one in the kitchen, about 1.5m from the sensor and one in the garage, next to the PC, another couple of metres away but through a couple of brick walls.
Being a radio ham, I checked the signal. I found that 'my' transmitter is centred on 868.320MHz but was audible for a good way each side of that frequency. The signal was very strong in the garage when it was working. I was getting solar information but not the other sensors. The most likely time for loss of signal was during the early morning but not always.
So I fetched it into the shack and froze it. It worked until the point at which the frost from the freezer spray melted and then started again a few minutes later. Due the delay from the real time clock, the FO panel didn't resynch for several/many hours later. Thinking therefore that the problem wasn't temperature but moisture, I then sprayed it with a fine mist of water. Bingo - it stopped and restarted later.
So something was being upset by moisture. Next I monitored the CPU clock by putting an oscilloscope on the crystal. The oscillations were strong until the point of frost melt and then they died out. Odd.
Fortunately, then, I tried to probe transistors around the CPU to see what was shutting the CPU down. One pad caused the CPU to shut down and restart, despite the meter being very high impedance. That is, it shouldn't have.
So digging around this transistor I find that it is linked to the lead from to the solar sensor and receives the 'reset' switch signal from to solar unit. Shorting this pad to negative shuts the CPU down - in fact triggering a restart. That's how the solar pod reset button works. The thing is, there doesn't appear to be any pull up to positive.
What I think is happening is that the lacquer coating is aging and cracking off, allowing moisture (we know that as dew!) through to the board and the film of water conducts enough current to trigger a CPU reset, or even a partial one. I had made it worse by connecting a long wire to it (the lead up to the solar sensor) which acts as an antenna and inserts any stray signals into this sensitive point.
I can't believe that this part of the circuit is intended just to float around aimlessly. Somewhere the pull up must have become detached. I have now soldered an 820k ohm resistor from the pad to the battery positive and for the first time in a year, it's run all through the night. In fact, I'm now 48hours+ with a reliable link.
Happy to supply photos if anyone finds this info useful.
Good luck, John
Like so many others, I have had serious problems with wireless comms. I moved from a 433 system to 868 because of the interference from next door's wireless security system, so I was dismayed to find the initially good 868 system gradually deteriorating. I have even brought out the console's down button wiring so that my PC could detect loss of signal and automatically force a re-acquisition. When I was losing signal for over 12 hours each day, I thought I should do some proper investigation.
Firstly, my sender is about 3m up the mast. I've extended the wiring for the sensors, so that I could put the transmitter as close as possible to the receiver, low on the mast. I use two consoles, one in the kitchen, about 1.5m from the sensor and one in the garage, next to the PC, another couple of metres away but through a couple of brick walls.
Being a radio ham, I checked the signal. I found that 'my' transmitter is centred on 868.320MHz but was audible for a good way each side of that frequency. The signal was very strong in the garage when it was working. I was getting solar information but not the other sensors. The most likely time for loss of signal was during the early morning but not always.
So I fetched it into the shack and froze it. It worked until the point at which the frost from the freezer spray melted and then started again a few minutes later. Due the delay from the real time clock, the FO panel didn't resynch for several/many hours later. Thinking therefore that the problem wasn't temperature but moisture, I then sprayed it with a fine mist of water. Bingo - it stopped and restarted later.
So something was being upset by moisture. Next I monitored the CPU clock by putting an oscilloscope on the crystal. The oscillations were strong until the point of frost melt and then they died out. Odd.
Fortunately, then, I tried to probe transistors around the CPU to see what was shutting the CPU down. One pad caused the CPU to shut down and restart, despite the meter being very high impedance. That is, it shouldn't have.
So digging around this transistor I find that it is linked to the lead from to the solar sensor and receives the 'reset' switch signal from to solar unit. Shorting this pad to negative shuts the CPU down - in fact triggering a restart. That's how the solar pod reset button works. The thing is, there doesn't appear to be any pull up to positive.
What I think is happening is that the lacquer coating is aging and cracking off, allowing moisture (we know that as dew!) through to the board and the film of water conducts enough current to trigger a CPU reset, or even a partial one. I had made it worse by connecting a long wire to it (the lead up to the solar sensor) which acts as an antenna and inserts any stray signals into this sensitive point.
I can't believe that this part of the circuit is intended just to float around aimlessly. Somewhere the pull up must have become detached. I have now soldered an 820k ohm resistor from the pad to the battery positive and for the first time in a year, it's run all through the night. In fact, I'm now 48hours+ with a reliable link.
Happy to supply photos if anyone finds this info useful.
Good luck, John