ABSTRACT We present an analysis of Epoch of Reionization (EoR) data from Phase II of the Murchison Widefield Array using the simpleds delay spectrum pipeline. Prior work analysed the same observations using the fhd/εppsilon imaging pipeline, and so the present analysis represents the first time that both principal types of 21 cm cosmology power spectrum estimation approaches have been applied to the same data set. Our limits on the 21 cm power spectrum amplitude span a range in k space of $|k| \lt 1 \, h_{100}\, {\rm Mpc}^{-1}$ with a lowest measurement of Δ2(k) ≤ 4.58 × 103 mK2 at $k = 0.190\, h_{100}\, \rm {Mpc}^{-1}$ and z = 7.14. In order to achieve these limits, we need to mitigate a previously unidentified common mode systematic in the data set. If not accounted for, this systematic introduces an overall negative bias that can make foreground contaminated measurements appear as stringent, noise-limited constraints on the 21 cm signal amplitude. The identification of this systematic highlights the risk in modelling systematics as positive-definite contributions to the power spectrum and in ‘conservatively’ interpreting all measurements as upper limits.