S9 flow skew bug... Anyone else seeing this?

You're right, there will be more wear and tear on the card, but I don't think it's going to cause much of a higher failure rate. The only difference would be the write protect switch moving across the two contact points you can see on the circuit image of the card below as it slides in that single channel.



At worst case the card might stay write protected if you wear out the circuit board contacts.

KatieW wrote:

Nord wrote: I believe the only corruptions people are getting, come from ejecting the card early or improperly.

Nord, by early or improperly, do you mean before closing down ResScan? Or do you mean not using Windows Explorer to "eject", as AirBreather explains in this thread:

viewtopic.php?f=1&t=50788&p=467268#p467268

I have to tell you, I really appreciate the work that you, DreamDriver, Dave, and others are doing. It's beyond my capability, and I would be lost without it.

Hi Katie

Yes. ResScan 3.10 should be closed (exit) from Windows before you eject the SD card. That will allow ResScan to stop using the SD card and messing up files that are being used by the program.

Then eject the card from windows. Ejecting "removeable" media is second nature in Apple OS and is very easy to find. In Windows you have to go to "My Computer" which will give you a desk top view of all the media options. Each computer will be different such as internal or external card reader, CD's, hard drives etc. It will likely be labeled "F" Drive on most and if when you formatted you gave the Volume a name that was easy, such as "ResMed" or "Card" or something else... then it will be easy to identify. Right Click on the drive and you will have some options... select "eject"... this will eject only the media (SD card) and the Reader (hardware) will still be available to use for other things. That will tell Windows and other programs (virus checkers etc) to stop using the SD card and it will close properly w/o corrupting any files.

If you want to be very protective... then you may want to "lock" the SD card but generally that introduces some other issues. I did that several times but I find it unnecessary as I do not believe that ResScan or Windows is creating the "skew" issues... just the "corruption" issues.

When I put my card in... before opening ResScan 3.10... I create a folder on my desktop and copy all the files to that folder. That is essentially the same as "locking" the card. But since I'm already backing it up anyway... that saves me several steps of locking and unlocking.

If you follow media hygiene pattern... there should be NO corruption. You can then set a schedule for how long you want to keep your back-ups.

1. Open Windows
2. Load SD card
3. Back up Folder by date created for SD contents
4. Open ResScan 3.10
5. DL card contents
6. Exit ResScan 3.10
7. Eject SD card
8. Exit Windows

That should cure any corruption problems for SD cards...
Now... the skewed contents of the SD card is a different "kettle of fish"...

Nord

Nord wrote:... If you follow media hygiene pattern... there should be NO corruption. You can then set a schedule for how long you want to keep your back-ups.

1. Open Windows
2. Load SD card
3. Back up Folder by date created for SD contents
4. Open ResScan 3.10
5. DL card contents
6. Exit ResScan 3.10
7. Eject SD card
8. Exit Windows

That should cure any corruption problems for SD cards...
Now... the skewed contents of the SD card is a different "kettle of fish"...

Nord

I can attest that this does not keep the contents from skewing. I used the above hygiene this morning and yesterday, but got a four-second skew on my data. So, today, I'm going to use the 3-second power down technique between mask events. Let's see if that helps.

Nord wrote: That should cure any corruption problems for SD cards...
Now... the skewed contents of the SD card is a different "kettle of fish"...Nord

Well, now I have to go back and re-read both threads (which I have now printed for reference)...I didn't understand that these are 2 different problems.

DreamDiver wrote:I can attest that this does not keep the contents from skewing. I used the above hygiene this morning and yesterday, but got a four-second skew on my data. So, today, I'm going to use the 3-second power down technique between mask events. Let's see if that helps.

I'm pretty sure the problem is emanating from the S9 and not from card corruption in a PC or through ResScan, just need to get enough days worth of data with the various tests to prove this fact.

Power cycling the S9 before use might make a difference, still not 100% sure but will be trying this.

DreamDiver wrote:

Nord wrote:... If you follow media hygiene pattern... there should be NO corruption. You can then set a schedule for how long you want to keep your back-ups.

1. Open Windows
2. Load SD card
3. Back up Folder by date created for SD contents
4. Open ResScan 3.10
5. DL card contents
6. Exit ResScan 3.10
7. Eject SD card
8. Exit Windows

That should cure any corruption problems for SD cards...
Now... the skewed contents of the SD card is a different "kettle of fish"...

Nord

I can attest that this does not keep the contents from skewing. I used the above hygiene this morning and yesterday, but got a four-second skew on my data. So, today, I'm going to use the 3-second power down technique between mask events. Let's see if that helps.

Hi all... hope this helps (and not confuses too much)

BTW: If you have different evidence or differing opinion... please chime in. I am investigating but I only have my own evidence to report. So....

I believe that "skewed contents" relates to the S9. We CANNOT "lock" protect the card data before installing on the S9. It "writes" and sometimes "over-writes" to the SD card at least each session and perhaps EACH TIME WE REMOVE THE SD CARD.

Since many people have progressive skewing of results and many people are removing and replacing the card many times (including me)... I'm thinking it does a "check files" and "over-write" of Summary data each time the card is replaced. I think it does a write or over-write of Hi Rez and Detailed Data during each session that there is any data available including summary.

Yesterday I had a card in the S9 and removed it at 6:30 AM after a night's sleep. I placed it in ResScan and read the Detailed Data for the night. I took another formatted and blank SD card and placed in the S9 at 10AM and had a nap. When I awoke at 11AM... I took the SD card from the S9 and read the data in ResScan. It only had Detailed Data from the "nap" and summary data from all other dates. The Detailed data from the earlier sleep during the same session date was missing. The S9 either knew that it had already been written or didn't have the same files as the "original" SD card.

Today I waited till 11AM to remove the SD card and it had several sleep times of Detailed Data recorded. I'm guessing that it writes Detailed Data after each sleep and it does not overwrite the Detailed data and maybe only writes it once... but it does overwrite the Summary Data which includes some of the Detailed Data and it continually updates that info which means it is checking all the files on the card and writing to at least some of them. That may account for the skewing of data because it is changing the other info and the Detailed Hi Rez data remains in the same place.

When I look at the Datalog folder for each date... there is a different amount of files for different dates... not just the size of the file. It may be adding files to tell the program how to modify it's data... not just over-writing with new data. That would account for the greater amount of files.

Sorry about the technical jibberish...

But... If we shut off the S9 and re-start it... it unloads memory and probably re-writes all the summary files according to the LATEST info which would correct all the summary data already collected.

In simple terms...

!. If we start a session and have a sleep... the S9 records the data on the SD... then take the SD and plug it into ResScan... we get stats for say 2 hours.
2. Then during the same session we put the SD back in the S9 and have another sleep... then the S9 records the data on the same SD... plug it into ResScan... we now have new stats for say 5 hours. The Summary Data and Stats have changed and somehow the S9 or ResScan has to make some changes or it will be skewed.
3. It is left to the S9 to record all data and ResScan simply shows the data that it receives.
4. Memory in the S9 is limited to 365 days of Summary but only 1 session of Detailed and Hi Rez data.
5. SD card is used to store 30 days of Detailed data and 7 days of Hi Rez and transfer to ResScan. (plus, of course all the Summary)

Therefore best Hygiene for SD card and S9 to help minimize skewing data is:
1. Power down S9 to "clear memory" when not being used (after cool down if needed)
2. Only remove SD card ONCE per day max and ideally ONCE per week so that the S9 can make corrections to data stats the minimum amount of times.
3. Must collect Hi Rez data AT LEAST every 7 days or it will be lost.
4. Maybe best to use the same SD card as much as possible so that corrections for Summary, Detailed and Compliance info is minimized as all the data is on the SD.

I know its rambling but thats how my mind works these days...

Nord

I locked the card before inserting it in my card reader and still had a flow skew today. I use the icon in the notification area to stop the reader/card before disconnecting the reader and removing the card.

I think it's the S9 that's doing the skewing when it records the next session.

Here's a look at my SD card showing the high res data for 16th March then showing it 8 days later minus the two high resolution files.


1. Skewing doesn't happen over time, e.g. 8 days later the data for 8 days prior doesn't get skewed
2. If you check the date stamps you'll see this (the 1hr time difference on the files are due to switching to BST
3. Skewing occurs only on the latest day and it seems to be a way of the S9 not being able to write everything in a timely fashion

e.g. I think there might be different caches for summary vs detailed vs high resolution data, and each isn't being cleared/reset at the same time frame causing some kind of skew in the latest data. Once skewed it continues to skew even further.

So now let's do a further investigation on the skew. For last night 30th March - I created a folder called 2010-MAR-31 as I backed up the card this morning. I know after checking ResScan that the data is not skewed in anyway. I can also verify that because for the last two day I've had a completed wiped new card image (now have two days worth of data) and I've been write protecting the card when inserting it into the PC although I don't think the write protect is changing anything here as it's the S9 that's skewing the data. I've then taken a look at the high resolution data file and highlighted in yellow the correct items I see.

The file name 20100330_225640_BRP.edf correlates with the time stamp within the file 30.03.10 - 22.56.40768 (left part of the image). So this says that the Hi Resolution data started recording at 22:56 and 40 seconds.



Now let's look at a minor skewed night...

The file name 2010313_221104_BRP.edf does not correlate with the time stamp within the file 13.03.10 - 22.11.17768. So this tells me that there is a skew going on.



I'm not 100% sure if this is going in the right direction other than when I check the last two nights worth of clean runs on an SD card where I have power cycled the machine each night, the times match up in the xxxBRP.edf file exactly. For any previous nights where I was seeing a skew, the times do not match up. So it seems to me that at some point in the session the EDF file becomes out of sync time wise. Whether it's at the start of the session or at the end of the session or part way through a session I'm not sure. My guess is maybe at the start or end of the session.

Unfortunately we can't read by hex all of the EDF file, so only the header is in ASCII that you see above. Ideally we need to be able to decode the file back into meaningful data to see what's going on to see if other graph times are skewing.

fiberfan wrote:I locked the card before inserting it in my card reader and still had a flow skew today. I use the icon in the notification area to stop the reader/card before disconnecting the reader and removing the card.

Hi FF

Unfortunately locking the card will only protect the Data that is saved on the card by the S9. It will stop any corruption by windows or ResScan if the card is ejected before either program is finished. As stated in this thread... the SKEWING of data is a different thing and relates more to how we get the card from the S9.

Is this the first skew that you noticed or could you have missed it before today?
When do you remove the card from the S9 and how many times is it removed in a 12 noon to 12 noon session?
How many sessions are on your card now?
Do you ever unplug or completely "power down" your S9?

We're trying to sort this out. I tried talking with Tech Support at ResMed today but that was no help... go figure

Nord

dave21 wrote: I'm not 100% sure if this is going in the right direction other than when I check the last two nights worth of clean runs on an SD card where I have power cycled the machine each night, the times match up in the xxxBRP.edf file exactly. For any previous nights where I was seeing a skew, the times do not match up. So it seems to me that at some point in the session the EDF file becomes out of sync time wise. Whether it's at the start of the session or at the end of the session or part way through a session I'm not sure. My guess is maybe at the start or end of the session.

Unfortunately we can't read by hex all of the EDF file, so only the header is in ASCII that you see above. Ideally we need to be able to decode the file back into meaningful data to see what's going on to see if other graph times are skewing.

Thanks Dave...

I believe that it is the right direction... the skewed data originates in the S9 for some reason, memory, split memory or whatever else it might be.

I was going to ask a whole bunch of questions to get to the bottom... but maybe we don't have too.

If the skew is cured by "hard boot" of the S9, then so be it. We can probably live with that until there is a more permanent solution from ResMed. I don't think I will hold my breath even if its a self-induced Apnea... When I asked their Tech Support today... they seemed helpful but not knowledgeable - can you say outsourced. The most he could say was that he though the card should be left in the machine all the time but he said he understood the need to back up.

Since you are now 2 days into an 8 day test of the hard boot theory... we'll soon know whether it helps. If that doesn't work well enough... then all the other work-throughs become more difficult. Let's hope.

When you hard boot, do you leave it off for long ?
Do you boot it just before you use it ?
Is it just once per day that you boot ?
Is the SD card in the machine before you turn it off?
Was your SD clean when you started this test?

Looking forward cause looking back doesn't help...

Nord

Nord wrote:When you hard boot, do you leave it off for long ?

It's been off for about 8 hours as I've been traveling

Nord wrote:Do you boot it just before you use it ?

So far I've been powering it on about 5 mins prior to use

Nord wrote:Is it just once per day that you boot ?

Yes

Nord wrote:Is the SD card in the machine before you turn it off?

Yes

Nord wrote:Was your SD clean when you started this test?

Yes, formatted in PC using Windows Vista, format=Fat32, default byte size allocation.

Really hoping that it doesn't skew. I'm still not 100% confident a power cycle is going to fix it, if there's a bug in the firmware then it's quite possible it's still going to occur, but a power cycle might kick it into touch if something is running away in memory between last use and next use and clearing the VRAM. The thing I don't understand is the skewing almost appears to happen randomly with people, some it takes a few days and some it takes a few weeks. Some it's there but hardly noticeable on the detailed graphs.

dave21 wrote: Really hoping that it doesn't skew. I'm still not 100% confident a power cycle is going to fix it, if there's a bug in the firmware then it's quite possible it's still going to occur, but a power cycle might kick it into touch if something is running away in memory between last use and next use and clearing the VRAM. The thing I don't understand is the skewing almost appears to happen randomly with people, some it takes a few days and some it takes a few weeks. Some it's there but hardly noticeable on the detailed graphs.

So am I... If its not the RAM then... I'm still thinking about that one

I'm thinking that the reason that it happens randomly is because everyone is following different practises and loading a different amounts of sleep for each session. Then everyone may not notice the skew as soon as others.

For example I switched machines so I don't have as long; I also started hard booting at different times just to try it out. We also have a pretty small sample that are following this thread... so maybe all those variables accounts for the differences

Nord

Nord wrote:I think it was Bill Bolton that stated the Detailed data was stored in S9 memory for up to 30 days and Hi Rez for 7 days based on his experience with the S8... but I'm not finding that. When I take a blank SD card and place in the S9... I only get Summary plus the current session for Detailed/ Hi Rez.

The S9 ONLY STORES detailed and high-flow data on a SD card - it is not stored on the S9 itself. The S9 itself only stores the summary data. The S8 only used the data card to transfer data from the S8 to Resscan and would write the previous 180 daya summary data and 5 days detailed data to the card every time it was inserted into the S8. The S9 is totally different.

bigk wrote:

Nord wrote:I think it was Bill Bolton that stated the Detailed data was stored in S9 memory for up to 30 days and Hi Rez for 7 days based on his experience with the S8... but I'm not finding that. When I take a blank SD card and place in the S9... I only get Summary plus the current session for Detailed/ Hi Rez.

The S9 ONLY STORES detailed and high-flow data on a SD card - it is not stored on the S9 itself. The S9 itself only stores the summary data. The S8 only used the data card to transfer data from the S8 to Resscan and would write the previous 180 daya summary data and 5 days detailed data to the card every time it was inserted into the S8. The S9 is totally different.

Yes I know that...

I was stating that Bill Bolton chimed in a while ago and said "different" when I had claimed that. He was offering info about the S9 because he thought it should be the same as the S8 and it isn't. Thanks for re-stating that...

If you're interested and since you just got your S9 and are using different cards every 7 days... perhaps you would be interested in keeping that pattern so that we can see if any skewing of data occurs. Are you interested ???

Nord

I'm planning on D/L every day and only using one card. Only time I will ever use 2 cards is if I am away from home for more than 7 days.