Actually, the reference voltage is not at all critical for the boiler interface. The current detection is done by R2 and OK1B. If the voltage over R2 goes above the voltage required to light the internal LED of OK1B (typically 1.2V), the output of the OK1B (pin 5) will be pulled high. Otherwise R4 pulls the output low.dirkjanw wrote:Any idea what could cause the gateway to intermittently need VR=2 or VR=3? Lately I regularly (every few days) need to toggle it for communication to be restored. The thermostat will report 'No communication' and no boiler messages are being decoded at that moment. So it seems the voltage is very critical there.
The voltage over R2 is related to the current from the boiler. For a logical low level (5-9mA), the voltage over R2 will stay below 0.9V. For a logical high (17-23mA), OK1B will limit the voltage to 1.2V. That means 12 mA will go through R2, the remaining 5+ mA lights the LED.
When OK1B isn't activated, the output is pretty much 0V. When it is activated the output goes to something like 3.5V. A really substantial swing. The output of OK1B is the input to the voltage comparator. So that should always work, no matter whether the reference voltage is 0.625V (VR=0) or 2.5V (VR=9), or anything in between.
If R4 is too small, the output may not go high enough and lowering the VR setting may help. But then there would be no need to ever raise it again. If any component in the boiler interface is faulty, changing the reference voltage would not make a difference. So I can't come up with a scenario that would explain the behavior you are observing, unless there's a loose connection somewhere. I would probably start by resoldering all component leads of the boiler interface. Next step is to measure voltages in specific places. Do you have a voltmeter? Is your IC1 in a socket so you can easily remove it?