Cheat sheet for problems with parallel buffer dumps

These are some things to know that may be helpful resolving parallel buffer dumping issues:

Things to know:

  1. Exposures need to be longer than the buffer dump, in order for the dump to occur in parallel. The dump can overlap some part of the exposure overhead, but not the overhead that immediately follows the exposure. This overhead can be on the order of less than 10 seconds that cannot be overlapped by the STIS buffer dump, but it can vary in length.

  2. The ACS internal buffer can store the equivalent of 1 WFC full-frame exposure, or 16 SBC exposures. The buffer must be dumped if a commanded ACS exposure will exceed the remaining buffer capacity. A completely filled buffer can be dumped in parallel with the commanded exposure if longer than 339 seconds. If the buffer is only partially filled when the dump must occur, an exposure time proportionately shorter than 339 seconds will still accommodate the parallel buffer dump. If another instrument is used in parallel-mode observing with ACS, its parallel buffer dump exposure-time requirements must be added to those for ACS.

  3. After the first WFC3/UVIS exposure, each exposure must be >= 348 sec, or after two short exposures, the next exposure must be >=664 sec. (This is because the WFC3 buffer can hold two full frame images, but when it dumps it must be able to dump both images in their entirety during the next exposure to be dumped in parallel.) A fuller explanation can be found in the : instrument handbook.

  4. These times assume that CR-Split=No. If not, the pieces of the split must meet these minimums.

  5. For WFC3/IR exposures the buffer can hold two full frame 15 sample exposures. As with the UVIS, dumping one exposure in parallel will require a >= 348 sec exposure and dumping two exposures in parallel require a >=664 sec exposure. But note that unlike UVIS the IR data can be partially dumped. A fuller explanation can be found in the instrument handbook: instrument handbook.

  6. When instruments are used in parallel it becomes more difficult to ensure that buffer dumps are done in parallel. This is because the instruments have to take turns dumping their buffers to the solid state recorder.

  7. You won't be able to get more than 4 full frame exposures in a normal orbit with ACS/WFC and WFC3/UVIS. You may be able to get more with WFC3/IR. Consider using subarrays if appropriate for your science.

  8. If there is a short exposure, put that one first (since nothing has to dump in parallel to it) or put it last.

  9. COS and STIS have pretty short dump times and adding them to the parallels does not make things significantly more complicated. But if TIME-TAG is being used only one TIME-TAG exposure can be in each Prime/Parallel group.

  10. If you are doing a CVZ observation, the final dump needs to be fit into the final orbit because there is no occultation in which to to bury it.

  11. The last exposure for a particular instrument in a visit will be dumped in series regardless of whether there is room to dump it in parallel to the prime observation. So if you are ending parallels for an instrument early in a visit, consider adding a short parallel with that instrument at the end of the visit to delay the serial dump until the end.

Example Proposals:

Note that you can also retrieve any of these example proposals into APT using File - Retrieve from STScI.