I've found a new interesting behavior with some of our sites after updating to 3.6 under Linux.
Brief summary, looking for help:
(1) Connected to solar inverters, 9600, via optically-isolated RS-485 adapter. This unit works fine (Generally) with all my other modbus devices I've tested it on. This particular inverter is doing something, that sometimes breaks 3.6. It worked fine with 3.5.
(2) When using the Modbus Tools, Mango never sees that data come back from the inverter. (No Response from Slave XX is the error message)
(3) When using the Enable/Disable data source option, Mango never sees the data come back from the inverter. (No Response from Slave XX is the error message)
(4) When just polling the device, Mango DOES see the inverter responses, but only AFTER the system is restarted or the computer is restarted.
(5) If you disable the data source, and restart the data source, it doesn't see the respones.
(6) If you restart mango entirely, it DOES see the responses from the inverter.
This behavior is (apparently) new in 3.6, as I had another site that was working fine with 3.5 and is now showing this behavior with 3.6.
My hypothesis is that these inverters are maybe responding too quickly (or something), and that's tripping up the RX/TX control circuit in the 485 adapters. I've tried messing with the flow control, and the behavior described above seems dependent on the resetting of Mango entirely.
I'm using Linux, and the JSSC drivers as we've discussed here before.
I've got (2) identical serial converters on most of my sites- FTDI chip based devices. They work fine running my other (2) devices that are standard in my install, an electrical meter and a (self-designed) Modbus-talking RS-485 input module.
These inverters always seem to talk properly using Modscan32 or similar tools, but fail with Mango as described above.
If I get a chance to visit one of these sites, I'll take a scope out and see what's happening on the bus, to test my RX/TX switching hypothesis.