un-necessary flagging of start times in phase III

[d3v-null]

Back in Phase II and earlier it was assumed that the first few integrations are bad, and you should throw them away (quack time).

Phase III is so well behaved, I don't think this is necessary (although I haven't fully verified this yet), and it ends up messing with ssins because the first averaged timestep has fewer raw timestamps that go into it, which makes it noisier, and it gets flagged out.

What would be ideal is something that detects how many timesteps at the start are bad, but this would be tricky.

[gsleap]

Hi @d3v-null - quack time is still very real- it takes into account: beamformers repointing and receivers changing frequencies. @andreww5au does the quack time get calculated based on which receivers are in use in an obs? e.g. I'm guessing NI vs RRI vs SHAO vs CIRA receivers all take different amounts of time to switch frequencies. And if SHAO is very quick compared to RRI for example we should expect a much shorter quacktime for obs which only have SHAO receivers right?

[andreww5au]

There will always be a ~0.1 second period of noisy data while the beamformers are pointed, with comms traffic on the coax, delay bits changing, etc, regardless of receiver type. Then the RRI receivers take about 2 seconds to change frequencies (if a change is necessary), and about 2.5 seconds to change the analog ASC attenuation (if a change is necessary, but this is quite rare). Beamformer pointing, frequency and ASC attenuation changes happen in parallel, so the longest possible period of corrupted data is still only 2.5 seconds. BUT, this needs to be considered along with the averaging time - for example, with 2.0 second time averaging in the correlator, 2.5 seconds of corrupted data from the receivers will contaminate the first two blocks, so four seconds of data needs to be skipped, and QUACKTIM will be set to 4.0 seconds.
The QUACKTIM in the metafits file for an obs is calculated based on whether the attenuation and frequency channels for this observation are the same as, or different to, the previous observation, using the change times for the RRI receivers. I have no idea how long the SHAO's take to change frequencies or attenuations, I just send the command and Jake's code handles it.