meta data for this page
  •  

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
Last revisionBoth sides next revision
app:using_analogtest [2022/03/02 14:55] – [What do all these graphs mean?] flanapp:using_analogtest [2022/03/02 15:09] – [How exactly does this work?] wordsmithing; added some explanation flan
Line 17: Line 17:
 === Noise === === Noise ===
  
-The Noise test in the lower left gives us a baseline for the density of noise that is generated by the drive when in "full panic" mode. Having a high noise level here **doesn't** mean that a drive is bad or will have trouble reading a disk. It is more the representation of the worst-case scenario of a drive trying to read a disk but it cannot find any data. Many drives have an automatic gain control function that tries to increase the sensitivity of the read head in order to lock in on the flux transitions on the disk. Unfortunately, the sensitivity can be turned up so much that the drive starts making up data that doesn't actually exist. The graph shows the number of "fake" transitions that the drive sends per rotation.+The Noise test in the lower left gives us a baseline for the density of noise that is generated by the drive when in "full panic" mode. Having a high noise level here **doesn't** mean that a drive is bad or will have trouble reading a disk. It is more the representation of the worst-case scenario of a drive trying to read a disk when it cannot find any data. Many drives have an automatic gain control function that tries to increase the sensitivity of the read head in order to lock in on the flux transitions on the disk. Unfortunately, the sensitivity can be turned up so much that the drive starts making up data that doesn't actually exist. The graph shows the number of "fake" transitions that the drive sends per rotation on a track that has been wiped of data.
  
 === Window Stability === === Window Stability ===
Line 70: Line 70:
 The primary analog test uses a test track that is comprised of a whole bunch of smaller tests. A test has a header, gap, payload, and settle field. The header contains information about the specific test (like the gap size). Following that is the gap (no flux area) whose duration ranges from 1µs to 48µs in 250ns increments. Then there is a payload which is a special bit sequence that is used to be able to detect bit slip, desynchronization, and other conditions. And then the settle which is a bit pattern that lets the analog amplifier/gain control/whatever cool down before the next test. The primary analog test uses a test track that is comprised of a whole bunch of smaller tests. A test has a header, gap, payload, and settle field. The header contains information about the specific test (like the gap size). Following that is the gap (no flux area) whose duration ranges from 1µs to 48µs in 250ns increments. Then there is a payload which is a special bit sequence that is used to be able to detect bit slip, desynchronization, and other conditions. And then the settle which is a bit pattern that lets the analog amplifier/gain control/whatever cool down before the next test.
  
-What it is checking for is the integrity of the gap and payload. If the gap is clean (no bits injected) and the payload is as well, then that is a success and it adds to the gray lines in the Stability graph. In the case of a failure, the payload is checked for integrity and if the payload is intact, then we check the gap. If the gap has a spurious transition (injected bit), then the time offset from the last flux transition of the header to the first spurious transition is recorded onto the First Injection Timing graph at the bottom. +To create the test track, the track is wiped (WRREQ on to engage head erase coil and writing no data), then the tests are generated and written. 
 + 
 +The test track is then read 50-ish times. What the test is checking for is the integrity of the gap and payload. If the gap is clean (no bits injected) and the payload is as well, then that is a success and it adds to the gray lines in the Stability graph. In the case of a failure, the payload is checked for integrity and if the payload is intact, then we check the gap. If the gap has a spurious transition (injected bit), then the time offset from the last flux transition of the header to the first spurious transition is recorded onto the First Injection Timing graph at the bottom.