Radiometric fingerprinting systems leverage unique physicallayer signal characteristics originating from individual hardware imperfections to identify transmitter devices. The pure passive nature of such mechanisms entirely relieves the overhead of identification and authentication operations from the end devices, which fits well with resource-constrained applications such as wireless sensor networks. However, existing systems are limited by the need for specialized hardware and non-trivial computations to extract fingerprinting features, hindering their pervasive deployment. For the first time, we ask the question whether it is feasible to implement an entire radiometric fingerprinting system on a low-cost and low-power embedded system on chip (SoC). We introduce Orf, which demonstrates the feasibility of such a system on an embedded SoC that costs less than 6 dollars. Our experiments in an anechoic chamber show that Orf achieves over 92% average accuracy in identifying one out of 32 different transmitter devices.