Objective: Photoplethysmography (PPG) has become the standard technology used in wearable health and fitness monitoring devices. Although, PPG works well at rest, trying to incorporate it for continuous monitoring is challenging, since the signal is inevitably corrupted by motion artifacts. This paper explores the feasibility of a reflective PPG sensor aiming at reducing the changes caused by motion in the non-pulsatile component of the signal, e.g., changes in venous filling. Approach: The tested sensor consisted of two back-to-back connected photodiodes and a light source with a peak wavelength around 575 nm. Additionally, the second photodiode was covered by a 575 ± 5 nm optical notch filter. Main results: When compared to standard PPG (single detector) on one test subject, the tested sensor provided 4.38 dB of attenuation for a change in venous blood during a gentle lift of upper extremity. Moreover, over 40 dB rejection was measured in the baseline resulting in a very small 'direct current' (DC) component. Furthermore, the baseline wander due to respiratory modulation tended to be smaller for the tested sensor. The sensor was not either susceptible for the changes in ambient light or the drift caused by the light source heating up. Significance: We believe that this is the first work that proposes the back-to-back connected photodiodes and a notch filter to reduce in-band noise in PPG. The positive results encourage to further optimize and integrate the sensor. The novelty of the method also brings new perspective to the PPG measurements.