One of the most important components of modern electronic systems in the fields of medical technology, automotive, 5G communications, etc. are microchips containing analog-to-digital converters (ADCs) on a single chip. Aging effects, environment and mechanical vibrations have a significant impact on the operation of electronic systems, so microcircuits need to be periodically verified, calibrated and even replaced. For this reason, in order to control metrological and technical characteristics of hardware and software of the "Multimag-M" complex, which contains diagnostic channels for measuring pulse wave, respiration process, blood pressure and blood oxygen saturation (saturation), it was proposed to introduce an automated block of control, testing and self-calibration into the structure of the chronomagnetotherapy complex [1]. Application of BIST (Build-in Self Test) - built-in means of self-testing is a promising approach to solving problems of reliability improvement of modern microcircuits [2]. Most of the known BIST methods of ADCs require on-chip highly linear test signal generator (TS) [3-4], which is the most complex block. The ability to test ADCs with a large number of digits is limited by the linearity of the TS generator and as the resolution of ADCs increases, it becomes a difficult task to provide linear TS generation on-chip. The aim of the work is to develop a method for determining the integral and differential nonlinearities of ADC independent of the linearity of the test signal. A method for determining the integral and differential nonlinearities of the ADC, taking into account the nonlinear component of the TS, is proposed. This is achieved by sequentially feeding to the input of the tested ADC a periodic TS of triangular shape, a level-shifted down TS and a level-shifted up TS. The first histogram of ADC codes distribution at absence of TS offset, and also the second and the third histograms at TS downward and upward shifting. Using the three linked histograms, the contribution of the nonlinear component of the TS to the integral nonlinearity for each ADC code is determined. Given the nonlinear component of the TS, the integral nonlinearity for each ADC code is further determined, and then the differential nonlinearity of the ADC is determined from the integral nonlinearity. In addition, the method determines the displacement voltages of the TS, which reduces the requirements for the accuracy of the displacement signal reference. The practical value of the research result lies in the application of the method reducing requirements to TS linearity at self-diagnostics and self-calibration of microcircuits containing ADCs will allow to reduce the complexity of crystal design.