#!/usr/bin/python # SRF cavity analog state computer # Takes in cavity field, forward, and reverse vector measurements # and computes the cavity detune frequency, decay parameter, and # power imbalance for the purposes of a tuning loop and quench detector. # Keeps a history of the previous two cavity field measurements so it # can get dV/dt. # Like cgen_srf.py, but with filters adjusted to give time-alignment of # primary signals (v, k, r) and dv/dt. fir_gain drops from 80 to 32. # Output of this program should be both valid c99 and valid input # for the scheduler/mapper. # See the rest of the Digaree infrastructure for details. from cgen_lib import cgen_init, given, mul, sub, cpx_sub, cpx_mul from cgen_lib import cpx_scale, cpx_dot, cpx_inv_conj, cpx_mul_conj from cgen_lib import cpx_mag, set_result, cpx_persist, cpx_copy, cpx_add cgen_init("cgen_srf.py") # History of measured cavity voltages, used to compute dV/dt # Initial values in simulation are read from init.dat or init2.dat. cpx_persist("k1") cpx_persist("r1") cpx_persist("v1") cpx_persist("v2") # These lines declare the input variables, # first six streamed from the radio given("k_r") # forward given("k_i") # forward given("r_r") # reverse given("r_i") # reverse given("v_r") # cavity given("v_i") # cavity # next eight host-settable given("beta_r") given("beta_i") given("invT") given("two") # needed by 1/x macro given("sclr") given("sclf") given("sclv") given("powt") # Get (still unscaled) derivative # Implements [-1 0 1] FIR cpx_sub("dvx0", "v", "v2", 3) # note multiply-by-4 cpx_add("dvx", "dvx0", "dvx0", 2) # multiply-by-4 # Result is the amount that V will change in 8*T. # Including the second-order CIC used to generate input samples, # this computation has a 2*T group delay. # State-variable computation of the complex number a, # yielding detune frequency and decay rate # use v1 and k1 to time-align with dvx cpx_inv_conj("x5", "v1", 0, 3) cpx_scale("dvdt", "dvx", "invT", 1) cpx_mul("x3", "k1", "beta", 1, 1) cpx_sub("x4", "dvdt", "x3", 2) # some evidence this shift should be 1 cpx_mul_conj("a", "x4", "x5", 2, 2) set_result("ab", "a_r", "a_i") # Power balance measure of cavity dissipation; uses magnitudes only cpx_mag("magr", "r1", 0) # reverse mul("powr", "sclr", "magr", 0) cpx_mag("magf", "k1", 0) # forward mul("powf", "sclf", "magf", 0) sub("wgnet", "powf", "powr", 1) # net power transferred by waveguide cpx_dot("dvsq", "v1", "dvx", 2) # 2 * V * dV/dt = d/dt(V^2) mul("dudt", "dvsq", "sclv", 3) # dU/dt = power to stored energy sub("diss", "wgnet", "dudt", 1) # est. of dissipation in cold cavity sub("perr", "diss", "powt", 1) # allow for measurement error set_result("cd", "diss", "perr") # trigger quench fault if perr > 0 # Watch these like a hawk: order of execution matters, # unlike everything else here cpx_copy("v2", "v1") cpx_copy("v1", "v") cpx_copy("k1", "k") cpx_copy("r1", "r")