Home » Blogs » rjreed's blog

EPD Run 23 configuration

Updated on Tue, 2023-05-23 18:53. Originally created by rjreed on 2022-07-17 12:31.

Executive Summary
Due to the 5 kHz rate of data taking, the existing 3 QT crate structure of the EPD will not work.  The plan is to add an additional crate, and then divide the boards in the 4 crates E and W so that all the crates are mixed.  The West side boards can then be read out by the DEP boards installed for the forward upgrade.  In run control, it should be possible to choose how many boards to read out via the QTs, so that if the rate is low enough for a given run, we can run in a similar fashion to previous years.

It may be that even at top rates we can read out all 8 boards per crate (see issue summary at the bottom), but we will not know until the data is here.

The report for the trigger group can be found at: drupal.star.bnl.gov/STAR/system/files/RReedEPDTriggerboard07192022v2.pdf

Each crate will get one QTD board... So the new mapping is:


New QT board mapping

The original mapping philosophy for the EPD can be found at: drupal.star.bnl.gov/STAR/system/files/RReedEPDTriggerboard03192018_1.pdf
We had 5 ring groups - 2 in the trigger acceptance and 3 in the rest.  For Run 23, we will have 4 crates, equally divided between East and West.  When the luminosity is high, we will not read out the QT boards connected to the West side (that information will come from the DEP boards installed as part of the forward upgrade).  Since at run control, we will be able to choose how many boards to "keep" and we need to recable everything anyways, we can devise a new strategy.  I propose to leave the 3 outer ring groups the same, and combine the inner ones into a single ring group - so that all the tiles associated with the triggers are the same.

To summarize:
Ring Group 1 = 108 x 2
Ring Group 2 = 72 x 2
Ring Group 3 = 96 x 2
Ring Group 4 = 96 x 2

We have:
18 QTBs (32 channels ADC, TDC), 14 QTCs (16 channels, ADC, TDC, TAC)
This means we have 9 QTBs, 7 QTCs per wheel for a total of 288 channels of ADC and 112 for ADC with timing.  In the old configuration we had 108 tiles linked to QTCs, corresponding to the first 5 rows, for the trigger. 

For completeness, the original mapping is: drupal.star.bnl.gov/STAR/system/files/EPD_mapping_12172019_1.xlsx

After some discussion, the version is:

Our final (I think) configuration as of 9/11/2022


Figure 3:  Now our configuration

Issue Summary

It's 399 ticks to read out a crate with no boards, an additional 177 ticks for the readout code with a single board, and then about 35 ticks per board.  Each hit adds about 8 ticks to the read-out time.  To read out at 5 kHz, the whole crate needs to be read out in under 2000 ticks, so if we have 4 boards with 32 hits each (i.e full occupancy) , that would be:

399+177+35*4+128*8 = 1740 ticks, under 2k

If we had to read out all 8 boards in this circumstance, it would be:
399+177+35*8+256*8 = 2904, too high

Occupancy may very well be ~50%, which with 8 boards would give us:
399+177+35*8+128*8 = 1880

So it may even be that we can run without removing any boards... TBD.  It is clear that there is no way we can run with 3 crates, though.

»