Hi, I have been experiencing the same issue, except my controller is the 100 | 30 version.
My setup is simpler; I have 2 panels connected in series (~37V Voc each), an old 12V 100Ah AGM battery and a Victron 12V 500VA Inverter connected via a fuse box with short lengths of 8AWG cables. The cable from the fuse board to the battery is longer than between the inverter and MPPT, so the voltage drop (measured with a multimeter) is higher which may change how the power from the MPPT is distributed between the load and battery charging?
If a load was connected (~150VA but I think it was still happening with ~100W too), the tracker would gradually reduce the voltage of the panels until the output was about ~50% of their initial output (e.g. 316W would reduce to 165W) and then briefly jump back up before repeating the cycle.
Initially I tried turn MPPT charging off and on again to reset the MPPT tracker, but disconnecting the load allowed the battery voltage to jump back up and the tracker to stabilize with the expected voltage drop every 10 minutes to re-calibrate the tracking point. If I connected the load again I would see the same behavior as before.
The battery was depleted at the time (~11.5V) because of use the day before so it was less resilient to spikes in the load which may have been contributing to the issue?
To me there is a clear link between the load on the battery and the controllers ability to track properly. Low battery voltage may also be contributing to the issue. Like you my controller was under 40 °C at the time but I cannot confirm there is a correlation.
Here you can see the difference in PV voltage before and after disconnecting the load:
Even after the PV voltage has stabilized, the battery has been charging for a while and the load is reconnected, I still see the tracking getting worse when the load is increased:
