Hi WFD readout fans, Here's a report on the WFD readout tests Bob and I conducted on Feb. 28: We created a new system (super.sys) in [bostonpub.kearns.pas.ve3_3] with the following modifications: (a) The WFD readout procedure in cadriver.pas was modified - The first event record (id = 1) now contains only the WFD header - this indicates the number of wfd channels to be readout. Subsequent event records (id=2) are created for 1st, 7th, 13th etc. channel to be read out. These contain the channel header followed by the WFD data for a maximum of 6 channels. (b) We set an upper limit on the maximum amount of data that will be read out for each channel. We set this initially at 200 bytes and later extended it to 10 K. (This includes the FADC data and the timeword-discriminator words) The ERP readout procedure from the ve3_2 directory was used since it is the newer version. Ed's code that allows setting of the readout window size in the camac list was also included. We tried out the system in the mini acq and then used it for the main acquisition on uVAX 1. The test runs are 9410 - 9433, all logged on $1$dia6. PHRASE was off on SM 1 until run 9426 (so no LIPs). We pulsed LEDs in the tanks instrumented for WFD readout and changed the camac list to cause WFD readout on TOHM triggers. RUN DESCRIPTION --- ----------- 9410-13 We start pulsing LEDs towards end of run 9410. During these runs the synchronization unit was flipped from uVAX 1 only on to all 3 uVAXes on by mistake. We had "hardware error executing equip 42" - the WFD equip no. - and the acq would hang during these runs. 9414 We reload the default system in uVAX 1 and use the default camac list. The run is fine - but with no LIP triggers, the default camac list causes no WFD readouts. 9415 We now use the new camac list with the WFD reading out on TOHM triggers - the run crashes with equip 42 errors. After a lot mini acq runs we determine that we get these errors if we are pulsing LEDs at the beginning of the run 9416 We reload the new system and use the new camac list. We start pulsing LEDs about a minute after the run begins. The acq hangs about a minute later with NET_SERVER complaining about trying to compact equip 0. Of the 30 events logged to disk, most look like perfect monopoles. Note that 8 channels are read out - two continuation records. 9417 We load a system from last weeks tests in which we had a continuation record for every channel pair. So we expect 4 continuation records for each event with WFD readout. The run begins OK but hangs immediately after we turn on the pulser and produce TOHM triggers - with the same equip 0 NET_SERVER errors. 9418 Now we decide to pulse LEDs so that only 3 channel pairs would get read out - so only one continuation record. This run lasted for 3.5 min (2.5 minutes of LED pulsing) and we have 501 events. There were a couple of hardware errors from equip 42 but the run continued eventually crashing with the equip 0 NET_SERVER errors. 9419 We reduce the LED pulser rate from 2 Hz to 1 Hz. This lasts for 10 minutes with 572 events and no problems until we end it manually. 9420 We go back to pulsing LEDs so that 4 channel pairs would get read out - 2 continuation records - but with the slower LED pulser rate. This run lasts for 3.5 min (2.5 min of LED pulsing) - with 151 events - and hangs with equip 0 NET_SERVER errors. 9421 Now we create a system so that a maximum of 10 K would be read out per channel. We pulse LEDs so that 3 channel pairs would be read out and at the slower rate of 1 Hz. The camac list still asks for 5 us of readout - since the LEDs are on all this time we expect 2 K /channel or 12 K /event. The run lasts for 8.5 min (7.5 min of LED pulsing) - with 477 events - then crashes with an equip 42 error followed immediately by the equip 0 NET_SERVER error. The data file size is consistent with 12 K - sized WFD events. 9422 Now we request 1 ms of readout time in the camac list. We expect about 10 K per channel - 2 K due to the 5 us wide LED pulses plus about 8 K due to background radiactivity - 60 K per event. The run starts giving equip 42 errors after about 2 min but goes on - eventually crashing after 4.5 min with equip 0 NET_SERVER errors. 125 events are logged but from fil occupancy of 5%, it is clear that 60 K - sized events have been collected 9423 We go back to 5 us of readout time. We add ERP readout in camac list but provide no LED pulsing. The run hangs after about 1 min with equip 42 and equip 0 NET_SERVER errors. Looking at events offline, we see no TOHM events, one ERP event with triggers on non-instrumented WFD channels and finally there's an ERP event with triggers on 2 channels instrumented for WFD readout - but the data file ends after the id=1 record. 9424 With same settings as previous run. Lots of equip 42 errors - but strangely, no NET_SERVER errors. We notice that the ERP stop is unplugged from the WFD Stop Master and plug it in. The WFD data for events from this run looks incorrect. 9425 Same as previous - but ERP stop plugged in from beginning. We get 4 quick equip 42 errors followed by the run hanging with an equip 0 NET_SERVER error - there's 11 events in about 30 s. 9426 We remove ERP readout from camac list, and now there's only TOHM readout with no LED pulsing. The run goes for 20 min than hangs with an access violation - at this time Fabrizio was starting a PHRASE min-acq on uVAX 6 to give us LIP triggers and we were trying to get more TOHM events via "whoopee" triggers. 9427 Now we have TOHM as well as LIp triggers. The run hangs in 13 s (2 events) with NET_SERVER error. 9428 Same as previous but runs for about 11 min (159 events). No problems until Bob starts generating "Whoopee" TOHM triggers - when the acq crashes with a NET_SERVER error. 9429 We reload the default system and the PHRASE mini-acq is off. We get several quick equip 42 errors and the acq hangs. Looking at the data, the events still contain continuation records - so the default system was not loaded properly. 9430 Reboot system from console. Same settings as previous and there are no problems 9431-33 Reload new system. PHRASE is restarted on a mini-acq on uVAX6. We comment out TOHM readout and pulse LEDs to generate a profusion of LIPs. Somehow we get no LIP triggers but only run 9432 hangs with a NET_SERVER error. OBSERVATIONS: ------------ (1) A preliminary look at the events in Runs 9416-9422 indicates that the events are being split into records just the way we would like. With 4 channel pairs read out we have the first event record with the WFD header saying that there's 8 channels to be read out - the first continuation record contains the first 6 channels and the second one contains the next two. The run would hang quicker if we had 2 continuation records for each event than if we had just one - but the size of the record does not seem to be a factor. (2) We do not understand what causes the "VME/VSB write errors on equip 42" or "Error compacting equip 0 - NET_SERVER error" (3) Setting the readout window via camac list seems to work. The amount of data generated in 1 ms is consistent with background readioactivity estimates. (4) The TOHM readout works fine but the ERP readout doesn't. (5) The system seems to be always loaded correctly with the "edebug/load" command - if you get the "network object unknown at remote node" message. (if we got an "awaiting debug activity" message, we would just reload the system.) If you boot the system from the console - it looks like there's a 50 % probability of loading the default system. Using MC NCP, the load file for MVMAC1 is [macro.super.current.scint]supervme_mv1.sys if the exec node is VXMACA - the default system. If the exec node is VXMACB, the load file is [macroeln.wfd]super.sys - where we keep our test version. Booting from the console sometimes loads the system in [macroeln.wfd] - so after completing our tests, we copied the default version to this directory. - Ashutosh