On 10 June 2011 02:13, Mandana Amiri <[email protected]> wrote: > Hi all, > > We have finally setup two dual-polarization antennas and we are acquiring > data using a Roach board and 2 iADC boards. We are running a slight > variation of poco design in tutorial 4, sampling at 850MHz. We adjusted the > equalizer gains for each frequency bin separately, but the same for all 4 > inputs (a, b, c, d). I realize now that one has to monitor the overall power > and re-equalize these gains to avoid clipping. Our current plan is to find > out experimentally how often we need to re-adjust these gains. > > Is there any guideline or advice on how to set these gains? > > I came across the corr_eq_init.py in the corr package, but I am missing the > idea behind the equalization polynomials (or is this because of a particular > setup?). > > Thanks, > Mandana
Hi Mandana For background to others: The 'equaliser' is a block that sits after channelisation and just before the requantisation in the signal flow in most (all) current CASPER correlators. It can be used to perform two important functions; 1. It can be used to shape the resultant spectra. Gain (and phase if complex correction factors are used) errors can be corrected here. 2. Its primary purpose is to prepare the data for requantisation, which involves a large reduction in resolution (often from 18 bits to 4). The data often needs to be 'shifted' up quite a few bits to sit within the most signficant 4 bits so as not to lose all of it. A polynomial is useful here as we mostly shift all of the channels up by the same amount which can be described by a trivial polynomial (eg. output = (4000 + 0n)*input, for n = 0:fft_size-1) It is important to ensure that sufficient signal is coming out after requanisation to ensure reasonable correlator efficiency while avoiding clipping. This requires a calibration step that can be done using autocorrelation products. Hope this helps. Regards Andrew
