[Evan, Sheila, Jenne]

Yesterday we had 3 locklosses in a row at AnalogCarm, which is quite unusual for us.  Sheila tracked it down to a big glitch when we enabled the IN2 of the A summing node.  The signal coming into that input looks the same as it had during a successful smooth aquisition sequence, so it's not a change in what's coming in. 

I ended up changing the guardian a bit, so that it starts with much smaller gain (-32 dB rather than -4 dB), hits the enable switch, then ramps up the gain to -4 dB before continuing the rest of the sequence as before.  The ramp up to -4 is quite fast - it has worked at least once, but if it ever fails, we may consider putting in a few small sleeps.