In this work, hardware unique fingerprints are extracted from differential nonlinearities of analog- to-digital converters. Transfer curves from measuring a large set of low-cost analog-to-digital converters with 12bit resolution are analyzed regarding their quality as system fingerprints. Additional post-processing methods are investigated for further improvements. The fingerprints are optimized to a perfect inter-hamming distance of 50%, close-to-maximum entropy and a minimum achieved number of 440bits. The intra-hamming distance could be reduced to less than 4%. Temperature measurements show that the error rate due to temperature drift never exceeds 13%.