Functional electrical stimulation is an effective means of rehabilitation treatment for patients with motor dysfunction after stroke. As most of the existing functional electrical stimulators use passive or parameter invariant stimulation methods and the effect is poor, a biofeedback electrical stimulator with adaptive adjustment of parameters based on surface electromyography was designed. Root mean square (RMS) of surface electromyography (sEMG) in the relaxed state and the maximum voluntary contraction state was taken as a reference. The RMS of sEMG at the interval of stimulation was used as the feedback signal to adjust the pulse width and amplitude of the next stimulation cycle. In order to avoid fatigue and injury caused by excessive stimulation, a fatigue detection algorithm based on characteristic Q was proposed. It was detected in real time to terminate electrical stimulation when the muscle reached the fatigue threshold b. Experimental results show that the instrument can output stimulation waveforms with frequency, pulse width and amplitude ranges of 1～10 Hz, 0.1～5 ms, 5～30 mA respectively. The absolute errors between the actual output value and the set value of pulse width and amplitude are 2.3% and 2.5%. The correlation coefficient of the linear fitting results of voltage and load resistance on 12 groups of resistors under the same stimulation amplitude is 0.999 7, and the constant current output is realized. The instrument can adjust the pulse width and amplitude according to the RMS value of sEMG, and the electrical stimulation stops when the muscle reaches the fatigue threshold (50%) .