Ensuring reliable and safe performance of medical devices in healthcare institutions is crucial for the wellbeing of patients. For this, physiological simulators may be used, which provide reference signals to compare against. The usage of physiological simulators, such as ECG simulators, as medical device testers is widely extended in high-income countries. However, their high cost and need for frequent calibration by the manufacturer make their availability and usage in low-resource settings (LRs) unfeasible. Remote healthcare facilities, low- and middle-income countries (LMICs), and hospitals in general would highly benefit from new designs of simulators that can be produced, repaired and maintained locally. Therefore, we present the design of an affordable, portable Arduino UNO-based prototype ECG simulator with the novel feature of visual indicators for self-calibration checks. The ECG simulator was built with affordable and broadly available components and tools, retrieved from an educational electronics workshop. The current prototype can be used to test 3-lead ECG meters. It reproduces a lead II waveform extracted from MIT-BIH database, with 1.3 mV peak-to-peak and 0.1 mV negative offset. Heart rage simulation ranges from 35 to 130 bpm, in steps of 5 bpm. Signal data points defining a desired waveform and heart rate can be uploaded into the simulator via the user-friendly and open-source Arduino software. The simulator could also be modified to generate 10- or 12-leads by upscaling the electronics and adding more storage space. Analog signal generation is based on a pulse-width modulation signal, which is smoothened by a low-pass filter and decreased in amplitude with operational amplifiers, which also add a negative voltage offset. Visual indicators enable simulated frequency and amplitude checks without the need of external tools, while amplitude can be calibrated manually. Additionally, rechargeable batteries facilitate on-site testing of medical equipment. Evaluation of the prototype included performance and functionality tests. HR and amplitude accuracy of the simulated signal were quantified and compared to those from a commercial simulator. Lastly, the performance of the simulator in a real scenario was tested at the Green Pastures Hospital of Pokhara, Nepal. Functionality tests proved the correct detection of the simulated signal by 6 patient monitors of different models, manufacturers and purchase year.