Semester 26A Stability Announcement

Precision radial velocity measurements which are comparable across long time scales (months to years) require a stable, well calibrated instrument. KPF has had several problems in the past which have impacted stability and calibrations and these will likely impact data taken in the 26A semester.

We discuss a few details below, but the high level summary is that we currently do not have confidence in the instrument sub-systems to maintain temperature stability and provide reliable calibrations. As a result, users of KPF who have science goals which require long term stability of the instrument should take this into account when choosing which science programs to execute with their time in 26A.

In addition, the KPF DRP is not performing well on faint object extractions. Somewhere around Gmag of 15-16 is where the performance degrades significantly, so we advise users to avoid faint stars if they need immediate results. We do expect the DRP to handle this case properly in the long term, so data taken now will likely be useful in the future, but the results coming out at the moment will significantly underperform the ETC.

Temperature Stability

The most impactful stability factor for KPF has been the detector temperatures. In order for RV measurements at different times to be compared, the system must not change temperature significantly during that time span as the hysteresis induced by a change and return to temperature results in a zero point change for the RV measurements. The goal is for the system to be stable at the ~1 mK level, while changes of ~1 K are almost certain to break the RV zero point.

The Green and Red CCDs in the main spectrometer are likely to experience thermal events of a few degrees C (or more) above the -100 C detector setpoints lasting a few hours. While the number and timing of these are difficult to predict, it is likely that KPF will not maintain long-term RV stability over 2026A. Thus, time series RV measurements over long periods (weeks or months) are likely to be compromised by changing RV zero points at the ~10 m/s level. Experience has shown that these zero-point changes are difficult/impossible to calibrate out and are different for each star observed. Science projects that rely on a time series of KPF RVs or spectra over a short timescale (a night to a few nights) are unlikely to be affected by the warmups (which would have to occur during the observing sequence to be a problem). Users who have science goals which require high precision (a few m/s or better) on long time scales should take this into account when choosing which science programs to execute with their time in 26A.

The WMKO and KPF Build Teams are currently planning for an additional servicing mission to address the cooling problems, however that will take time to plan and implement.

Calibration Reliability

In addition to stability, the instrument also requires frequent, high precision calibrations. The most fundamental calibrator for KPF is the Laser Frequency Comb (LFC). The LFC has had periods of significant unreliability and the wavelength coverage for the LFC has been inconsistent. This has resulted in long periods of poor calibrations which requires bootstrapping the wavelength solution from other calibrators or over longer time periods which impacts RV measurement precision.

Detector Noise

In addition to the stability concerns, KPF has also faced challenges with detector noise. The elevated read noise present since early 2024 has been partially mitigated. Currently, the Red detector is performing at a nominal level (4.3 electrons read noise) and although the Green detector noise has been significantly reduced from the 25-35 electrons seen in the past, it remains elevated at about 10 electrons. An intervention to address this is being planned.