Hello,
We monitor the ADPA7009-2 temperature through the VREF pin. VREF is pulled up from 3.3 V through 100 kΩ and read by an ADC. The PA is biased in class A at about 4.3 V and 644 mA, so it dissipates about 2.75 W with no RF.
Our board thermal model and the fast VREF step at PA turn-on suggest that VREF reads only about 5–7 °C/W × P above the exposed paddle. The datasheet θJC is 13.6 °C/W.
So we cannot tell which of these is true:
- (A) VREF sits at a cooler spot than the hottest channel. Then the channel is about (13.6 − 5) × 2.75 ≈ 24 °C above the VREF reading.
- (B) VREF is close to the channel temperature, and θJC in our condition is lower than 13.6 °C/W.
Questions:
1. When the PA heats itself, which temperature does VREF follow: the hottest channel, an average or local die temperature, or something closer to the paddle? Is θJC = 13.6 °C/W measured from the hottest channel?
2. Is there a typical VREF-versus-temperature curve (value at 25 °C and slope) for the ADPA7009-2? We use the ADPA7002 values from AN-2578 (0.894 V at 25 °C, −1.21 mV/°C). Checked against thermocouples with the PAs unbiased, these read about 11 °C high on our parts, while the slope agrees within about 10 %.
We want to set a firmware over-temperature limit on VREF. Your answer decides whether that limit needs about 24 °C of extra margin.
Thank you!