Data "Skew" on S9 Series

[Breathe Jimbo]

I saw this mentioned for the first time in the video "Best Practices for the Resmed S9 Series SD Card" on another cpap forum:

http://www.apneaboard.com/ResMed/resmed-SD-card.html

Is the skew a hardware or software problem?

Has ResMed acknowledged the problem or said it will fix the problem in a future ResScan release?

I had 7 months worth of data on my SD card before I learned about this!

[DreamDiver]

I saw this mentioned for the first time in the video "Best Practices for the Resmed S9 Series SD Card" on another cpap forum:

http://www.apneaboard.com/ResMed/resmed-SD-card.html

Is the skew a hardware or software problem?

Has ResMed acknowledged the problem or said it will fix the problem in a future ResScan release?

I had 7 months worth of data on my SD card before I learned about this!

Darn. They didn't even ask permission to use the video.
There is a firmware bug in the S9. It cannot be remedied using software. We went through a lot of drama last year about this time to learn this. Google "S9 skew site:cpaptalk.com" to find see what we learned. Until Resmed says otherwise or members discover otherwise, there have been no S9's released yet without the skew bug. There may not be any skew in the new S9 VPAP Adapt, VPAP S or VPAP ST, but nobody has yet reported one way or another. It's not really a big deal since it's about data and not about therapy. I will be creating a future release of the video that will better explain what the concern is and how to deal with it.

[Breathe Jimbo]

Interesting discussions in the prior threads.

Is the bottom line that we should reformat the SD card every two weeks in order to minimize the skew?

Was it ultimately concluded that unplugging the machine each morning does not help with the skew?

[DreamDiver]

Interesting discussions in the prior threads.

Is the bottom line that we should reformat the SD card every two weeks in order to minimize the skew?

Was it ultimately concluded that unplugging the machine each morning does not help with the skew?

Jimbo,
It doesn't really matter from the standpoint of therapy. If you're happy with how you feel in the morning, and you aren't going to be giving your doctor the data in the next two weeks, you don't have to do anything. Wake up and relax, or go to work or whatever. However, if you like to keep your data obsessively-compulsively clean or you're visiting your sleep doc in a couple weeks, it's nice for the graphs to line up when they view your data. Unplugging doesn't work.

[BernieRay]

...
Darn. They didn't even ask permission to use the video.
...There is a firmware bug in the S9...

It looks like they pulled the video and the links.

While I believe the problem is file fragmentation due to the file format of SD cards and not an assumed firmware bug, it really doesn't matter because the only means that we have to avoid the problem is the same regardless of the root cause. I delete all files on the card once a week (personally easier than remembering every other week) to deal with the skew issue.

But, I also disconnect the power each week at the same time. The reason for this is to avoid another issue I had where the data from the previous night was somehow getting included in the next night's data. Disconnecting power resolved that problem, so I keep doing it just in case.

[Breathe Jimbo]

Any reason not to reformat the SD card after each download into ResScan, before reinserting it into the S9?

[Otter]

While I believe the problem is file fragmentation due to the file format of SD cards and not an assumed firmware bug, it really doesn't matter because the only means that we have to avoid the problem is the same regardless of the root cause.

Seems to me that it's a firmware bug either way. If the firmware can't cope with the file system, that's a bug.

I delete all files on the card once a week (personally easier than remembering every other week) to deal with the skew issue.

Does simply deleting all files instead of formatting the card take care of it?

[BernieRay]

While I believe the problem is file fragmentation due to the file format of SD cards and not an assumed firmware bug, it really doesn't matter because the only means that we have to avoid the problem is the same regardless of the root cause.

Seems to me that it's a firmware bug either way. If the firmware can't cope with the file system, that's a bug.

I delete all files on the card once a week (personally easier than remembering every other week) to deal with the skew issue.

Does simply deleting all files instead of formatting the card take care of it?

Yes - that is what I've been doing for almost 6 months and I've had no data skew as a result.

[Otter]

Yes - that is what I've been doing for almost 6 months and I've had no data skew as a result.

Thanks.

[Breathe Jimbo]

The so-called "data skew bug" appears to be absent in the current alpha of SleepyHeadQT. Mark was very busy overnight. When I loaded my S9 Elite data this morning (the first time!), from an SD card that has not been reformatted in about a month, my events from last night (albeit only 3) were PERFECTLY aligned with my flow rate.

If you want to try the current alpha, you will need to install QT 4.7.3 (I did so using MacPorts) and build from source. We could use some testers who have more events, snores, etc.

[idamtnboy]

Interesting discussions in the prior threads.

Is the bottom line that we should reformat the SD card every two weeks in order to minimize the skew?

Was it ultimately concluded that unplugging the machine each morning does not help with the skew?

Jimbo,
It doesn't really matter from the standpoint of therapy. If you're happy with how you feel in the morning, and you aren't going to be giving your doctor the data in the next two weeks, you don't have to do anything. Wake up and relax, or go to work or whatever. However, if you like to keep your data obsessively-compulsively clean or you're visiting your sleep doc in a couple weeks, it's nice for the graphs to line up when they view your data. Unplugging doesn't work.

I agree 100%. I had 7 months of data on my card before I changed machines (replacement for one that died) and there was little data skew on the later data. I really kind of think if one doesn't worry about it it kind of goes away somewhat after several weeks of use.

I just took a quick look at some of my data. In the later data one night had both 0 and 15 secs skew, one before and the other after getting up in the night. A couple of nights later the skew was nearly one minute. At six months use skew is about 0, so it bounces around. I saw that same variability in data after one month's use of the card. That leads me to believe that the card condition and use has very little solely to do with causing the skew.

[idamtnboy]

The so-called "data skew bug" appears to be absent in the current alpha of SleepyHeadQT. Mark was very busy overnight. When I loaded my S9 Elite data this morning (the first time!), from an SD card that has not been reformatted in about a month, my events from last night (albeit only 3) were PERFECTLY aligned with my flow rate.

I don't know how it can be a SleepyHead solution, unless the record start times are adjusted by comparing the five separate data files. When I looked at the raw data on the card quite closely back in November the skew was right in the time stamps inside the data files. That comes solely from the S9 firmware. However, the time stamps do appear to be influenced by the card condition on cards that have been used for a few days only.

[Otter]

The so-called "data skew bug" appears to be absent in the current alpha of SleepyHeadQT. Mark was very busy overnight. When I loaded my S9 Elite data this morning (the first time!), from an SD card that has not been reformatted in about a month, my events from last night (albeit only 3) were PERFECTLY aligned with my flow rate.

I don't know how it can be a SleepyHead solution, unless the record start times are adjusted by comparing the five separate data files. When I looked at the raw data on the card quite closely back in November the skew was right in the time stamps inside the data files. That comes solely from the S9 firmware. However, the time stamps do appear to be influenced by the card condition on cards that have been used for a few days only.

Jimbo, did you compare between ResScan and Sleepyhead, and see skew in ResScan where there is none in Sleepyhead?

I'm sure the skew is in the data, but maybe Mark found a way to line the events up anyway. That would be something given the amount of time that Resmed has been (not) working on this problem.

[Breathe Jimbo]

Jimbo, did you compare between ResScan and Sleepyhead, and see skew in ResScan where there is none in Sleepyhead?

That would mean rebooting to windoze <barf>.

Mark found different time stamps. I don't have a handle on the technical details.

[Otter]

Mark found different time stamps. I don't have a handle on the technical details.

Cool. This means he solved in a few hours a problem that has been sitting on Resmed's desk for what, two years? So much for Resmed's coders.