High-Resolution, Multi-Channel FPGA-Based Time-to-Digital Converter

TL;DR

A novel high-resolution multi-channel FPGA-based time-to-digital converter (TDC) whose overall accuracy is several orders of magnitude greater than the accuracy of the FPGA used in digital mode is presented.

Abstract

In this paper we present a novel high-resolution multi-channel FPGA-based time-to-digital converter (TDC). We designed and implemented a complex electronic circuit on the FPGA, whose overall accuracy is several orders of magnitude greater than the accuracy of the FPGA used in digital mode. Our sensor device contains simple circuit elements that are cheap and easily accessible (Xilinx Spartan 3 and Spartan 6). Using our design, many channels (80-100 channels) can be implemented on a larger FPGA. The prototype of our TDC has been implemented and functionally verified by experiments and measurements. By a certified pulse generator 20 ps precision has been measured over the range of 3 ns. Using more precise clock signal this range may be extended. The achieved resolution is 5 ps. Its resolution, channel number and range can be configured dynamically, which makes it suitable for effective use in industrial purposes.

Figures (6)

• Figure 1: Schematic of the prototype board.
• Figure 2: Implemented prototype.
• Figure 3: We measured the pulse delay between two signals generated by a two channel function generator (Type: Agilent 81130A) A personal computer triggered the signals, then read out the output values of the TDC conversion via USB interface.
• Figure 4: FPGA-A does the time-to-digital conversion. FPGA-D is the control FPGA responsible for the temperature-stability, control processes and measurements of the reference signals and their parameters.
• Figure 5: On the upper side we see the output of the TDC conversion depending on the delay set on the function generator. At the bottom side, we see the error between the theoretical and the measured delays. As we can see, the accuracy of the measurement is about 20 pico-secundums, which is the accuracy of the function generator. It is assumed that our method is capable of even more accurate results.