| Symptom |
Reporter |
Root cause, or best guess |
Resolution |
|
| a |
Faulty string detected by DOR
Driver |
John Jacobsen |
a. Faulty
decoding by UART b. Faulty string from GPS clock c. Noise triggering UART |
Faulty UART decoding may be
another artifact of timing violation. To date, no faulty strings have been detected at the serial output of the GPS clock (4/8-4/22/05) |
| b |
19,999,999 counts paired with
20,000,001 counts in the next second (Seen on all cards simultaneously) |
Claire Pettersen |
Symptom induced by incorrect timing relationship between ↑ transitions of 10 MHz and ↑ transitions of the 1Hz (or 1pps) signal. | Change GPS clock cabling per recommendation |
| c |
Δt = 40,000,000 |
Claire Pettersen |
Skipped time cal from driver due
to suppression of malformed string (see a above) |
Software needs to deal with
occasional omitted tcal |
| c1 |
Δt = 40,000,000 or higher
multiple of 20M |
GTP |
a. PCI fault causing lost packet b. Driver error causing packet to be skipped c. Operating system fault causing lost packet d. Missing 1 pulse per second pulse |
Seen several times before
4/18/05. Periodic on 1-day scale Not seen since 4/18/05 with variant (logging) driver. |
| d |
Δt not a multiple of 20,000,000
due to insertion of a zero byte into the the DOR time string, or an 8 bit shift in the DOR time data. Least significant bits appear to be discarded (Seen in the same second on two of four DOM Hubs) In the 4 cases observed (by DC) this error the DOR timestamp appears to be shifted up one byte, and the new lower byte was '0x01'. Fix: ignore such stamps. |
Claire Pettersen Dima Chirkin |
Symptom induced by incorrect timing relationship between ↑ transitions of 10 MHz and ↑ transitions of the 1Hz (or 1pps) signal | Change GPS clock cabling per recommendation |
| e |
GPS Quality character not a " "
(Space) but *is* one of the other valid characters (usually anyway).
--DC |
Dima Chirkin |
Normal behavior: Observed in the bldg 50 test lab. Assumed to be caused by sub-optimal antenna placement resulting in tracking of an insufficient set of satellites. Symptom of reacquisition of satellites. | See pix... Mount GPS antenna clear of obstructions and shadows per recommendations in manual |
| f |
GPS clock changed from a good to a bad value.
|
Dima Chirkin |
DOR card GPS String Buffer not Flushed when starting a data run., or another symptom of incorrect timing relationship between ↑ transitions of 10 MHz and ↑ transitions of the 1Hz (or 1pps) signa | Software fix:: ignore the new value. |
| g |
GPS clock changed from a bad to a good value. | Dima Chirkin |
DOR card GPS String Buffer not Flushed when starting a data run, or another symptom of incorrect timing relationship between ↑ transitions of 10 MHz and ↑ transitions of the 1Hz (or 1pps) signa | Fix: if the new value is
repeated more than once, accept the new value. Calibration with the good GPS structure therefore depends on having at least 3 GPS last structures in a ring buffer before performing a calibration (which is done in fat-reader, but may not be possible with the current data collector's setup). I've (DC) just sent the corrected calibration class to Mark, where I *could* (I think) correct this type of error except when the bad GPS is recorded in the first tcal cycle. |
| h |
Observed valid (usable) GPS-DOR structures which referred to UTC times at least hour before the current run. This may need to be fixed if dor clock can be reset during the start of new runs. | Dima Chirkin |
Stale
DOR time structures not flushed before starting a new run. |
Future software fix? |
