After a week of CW hardware injections, it's now feasible to recover some of the stronger signals with confidence. The bottom line is that amplitudes and phases look about right across the full band of injections (up to 2991 Hz).

Below are examples for specific "strong" injections (meaning marginally detectable with a day's integration):

Pulsar 6 - near 145 Hz
Pulsar 1 - near 849 Hz
Pulsar 4 - near 1387 Hz

For each injection, four graphs are shown:
- Cumulative F-statistic (predicted and recovered) vs day, starting May 17
- Cumulative recovered h0 amplitude with the true value as a horizontal line
- Cumulative recovered phase angle with true value
- Cumulative recovred polarization angle with true value

A full set of results can be found here.  
Notes:
1) The true signal parameters can be found here. 
2) The predicted F-statistic tends to be unduly optimistic in non-stationary bands
3) The last two signal injections are in binary sources, which this signal recovery infrastructure doesn't currently support (future upgrade planned). 
4) The phase recovery graphs show both a solid line for the true value and a dashed line which can be an attractor when the polarization angle is poorly recovered
5) A new of injection recoveries based on May 24 data onward will supersede these ER15  results after today