Stabilization of the calibration signal frequency Vitaly Stepin Dmitry Stepin Jevgeny Boole

Problem to be solved T_EXT ~ 14us Period of the calibration signal T_VCO is varied in range … us Different measurement results of the RMS jitter at different frequencies of the calibration signal Period of the calibration signal must be fixed for satisfactory measurement results

Block diagram of the VCO period adjustment PC EPP (control + data + flags) Timer FPGA VCO module Calibration generator VCO signal FIFO Control EnWrite External signal ADC Logic

Requirements for realization. Setting VCO period Tvco_min = us Tvco_max = us N = = quantity of VCO periods. It influences duration of the period adjustment and accuracy. Tclk = 10 ns (time base) clkCounter = Tvco * N / Tclk clkCounter_min = Tvco_min * N /Tclk clkCounter_max = Tvco_max * N /Tclk clkCounter_min = u * / 10n = ,4 ~= 0x37DC86F clkCounter_max = u * / 10n = ,8 ~= 0x37DE179 0xC = xE = bit counter is required. 16-bit counter is used for convenience purposes. For example, Tvco = us clkCounter = u * / 10n = ,0 ~= 0x37DDB78 0xDB78 is sent to the VCO Period Register (clkCounterRef) VCO period step is accuracy +\- 0.15ps

Operation principles COMSEL (6) clkCounter clkCounterRef

RMS jitter dependence on VCO signal period. Test setup Calibration generator External signal T_EXT = const ~= 14 us Logic T_VCO is variable ADC Finding a good VCO frequency

Signal RMS jitter versus VCO period (a rough estimation) step 0.21ps 10 iterations per step signal period ~14us) T_VCO = us T_VCO = us T_VCO = us

Signal RMS jitter versus VCO period (a fine estimation) T_VCO = us (step 0.21ps, 400 iterations per step, signal period ~14us) Assumed as a good period/frequency

Testing good frequency over o C T_VCO = us ( iterations, signal period ~14us) VCO period error, ps Internal temperature, C Signal RMS Jitter, ps

RMS jitter dependence on external signal period. Test setup Calibration generator External signal T_EXT = us Logic T_VCO = us ADC A good frequency

RMS jitter dependence on external signal period. Experimental results T_VCO = const = us (good frequency) T_EXT = us (ETTG, 8-bit resolution, 20 iterations for each k = ) External signal period, ps External signal RMS jitter, ps VCO period error, ps

Conclusion VCO period couldnt be selected in an arbitrary way. It is possible to set period of the VCO generator with high precision. Precision of the VCO period depends on N and DAC resolution (in general, even 0.1ps is not a limit but so high resolution is not required). Adjustment time depends on N. VCO period can be simply updated from PC software via EPP port. Only several narrow gaps for good frequencies (not more than 4ps) are found. 800ps of the VCO period were tested (period of the VCO signal can be changed in ~2000ps interval). Wider intervals should be found for more reliable operation. Time base (10ns) must be very precise!!! Experimental results show that VCO period is fixed with expected precision. The good frequency seems to be OK. No changes in PC software are necessary for calibration data acquisition => compatible with older version RMS jitter may also depend on frequency of the external signal.