S9 - Still Checking SKEW ???

[Nord]

Still tracking down some info about S9's only please...

There is a group of us that are working to find out why we have a nagging issue with Data Skewing and being progressively worse.
Some members here have had their S9's for quite a long time; others are very new to ownership.

We believe that perhaps there are some parts/ assembly differences in the machines that may lead to differences in how they are affected by this problem.
We are now looking at Dates and Lot Numbers for clues to changes.
This is completely voluntary participation by S9 people to solve "our own problems through working together".

Please "pm" me direct with the Info...
Please DO NOT Post the answers or Numbers in this thread...

I will record, over the next couple of days, all the data and try to put in in order...
After the info used for this purpose, it will be deleted and I will notify here when that is done.
The Serial Numbers will not be shared or used for any purpose other than solving this "Skew Issue"
I will also Post here whether this theory has had any validity at all about skews...
If we see a pattern... I will share that info and we will try to track down what hardware differences there are
That will also signal that some machines need to be changed/ upgraded.
I will notify you if we find that you are in need of changes.
This could take a while... please be patient.

We are looking for Answers to 6 Questions... Please answer in that order...

Q 1. What approximate date did you purchase your S9 ??? ( MM / DD / YYYY )
Q 2. Have you had any skew problems that you know of ???
Q 3. Have you looked at your Flow and Detailed Data closely to compare if they are more than 2 seconds apart when starting ( on the graphs in ResScan 3.10 )
Q 4. What is the maximum amount of Skew that you have seen ??? ( in seconds )
Q 5. What is the Serial Number of your S9 ( located on a plate on the back of the S9 )
Q 6. What is the Lot Number of your S9 ( located on a plate on the back of the S9 )
( ie: Serial Number # 22101113094 Lot Number # 244381A )

BTW: all S9 machines seem to have the same Firmware version as shown in the LCD... SX 474 - 0601 (If by any chance you have something different - Please let me know )

As always any helpful suggestions are appreciated...
Nord

[gpk111]

No skew problem detected from my S9

Purchased - March 13, 2010
S/N - 22101-136861
Batch - 243341A
Firmware - SX474-0601

Thanks for your generous advice and counsel on this very helpful forum!

[Nord]

Thanks Gerry...

Nord

[Nord]

Still looking for more people to respond...

Nord

[Nord]

Not getting a "great response" from people about the data skew on the S9.

I have assumed that one or more of the following is true:

1. Very few people are actually getting a data skew from their S9.
2. Many people are getting data skew but it is minor.
3. The data skew may correct itself over time.
3. Some people do NOT recognize what the data skew is and DO have it.
4. Some people do NOT care whether its there or not.

For the people that did respond...
I will PM them with results of "how the serial numbers may affect their skew"...

Nord

[fiberfan]

Not getting a "great response" from people about the data skew on the S9.

I finally moved the machine today to get the data so check for a pm from me. Perhaps others aren't responding with serial number and lot number for similar reasons.

[coreyg]

I haven't seen the skew. One day, the detailed data was 1.5 seconds off but it did not get worse and the next day was right on.

I do plan on responding to your pm, but I just haven't gotten to it yet. Sorry.

-Corey

[DreamDiver]

I got more feedback from ResMed about Skew. My emphasis in blue.
I am a bit confused, so I thought I'd hand this over to our board members who are less brain-fogged.
I know either Nord or dave21 was able to find the area in the data that shows write times, but I'm not sure where that's located in the text part of the file.
Apparently, Sandisk SD cards have faster a housekeeping algorithm than many other manufacturers.

I would be grateful if someone could clarify what all of this means.

What it boils down to is this according to ResMed:
A number of us got crummy cards, but that number is very small. Everyone else got good cards, so they're not seeing skew. Nord, you will not likely see skew on your new machine because it came with one of the new cards. If you try one of your other cards, it will likely show skew because of the inefficient housekeeping of some brands.

One very cool thing came out of all of this: looking very closely at the .edf files has been a lot of fun.

There’s a couple things going on here, but you are confusing a few things. The SD card has a house keeping algorithm that only has certain periods of time that its ‘available for communication’ which has to align with our own data write periods. This varies by mfg.

This availability can be on the order of 250ms

http://www.ultimaserial.com/sandisk.html

Note that every manufacturer can employ their own house-keeping algorithm, that will ultimately determine how often these limits are reached. This is why we expect to see difference between SD card manufacturers, although they all follow the same standard.

This “access time limit” is much bigger than the numbers that the users are quoting here – the numbers they are quoting, relate to the speed of the card - which is much faster, and is only applicable for a single block write – when the card is “available for communication”.

We can look into the file write data and see when the exact time it took to open, read, and write the data and its clearly getting worse over time. There’s also a confusion on the board where some of the users are ripping the card out of the device before its closed the session (it doesn’t make sense to write data every single second)

[KatieW]

Thanks for posting this DreamDiver. So, did ResMed say what the people who got the crummy cards should do?

I have a SanDisk 1.0 GB card, it is the back-up for my camera, and has never been used. Should I try that instead?

[DreamDiver]

Thanks for posting this DreamDiver. So, did ResMed say what the people who got the crummy cards should do?

I have a SanDisk 1.0 GB card, it is the back-up for my camera, and has never been used. Should I try that instead?

I would try the SanDisk card, if I had one. Why not?
I know that SanDisk has not always had the reputation it has now. Depending on the age of your card, your mileage may vary.
I'll ask.

Meanwhile, if there is anyone out there who can explain the 250ms housekeeping window, I'd be grateful.

[DreamOn]

There’s also a confusion on the board where some of the users are ripping the card out of the device before its closed the session (it doesn’t make sense to write data every single second)

What is being said here? "Session" is defined in ResMed literature as being a 24-hour period, from noon to noon. Are they saying that the card should not be removed from the S9 machine until after noon? Or are they referring to "session" as the period of time when the machine is actually being used, with the blower on (perhaps including a bit of time after the machine is turned off to allow the machine to write all the data to the card)?

I don't think people are "ripping the card out of the device" while the machine is still running, so I'm not sure what they're saying. Can we get the suggested procedure clarified by ResMed, as the verbage is very unclear.

[Nord]

Meanwhile, if there is anyone out there who can explain the 250ms housekeeping window, I'd be grateful.

Hi DD

The "housekeeping issue" is basically a time out the media uses to check the read/ write details and make sure the the "checksum" is correct for the details that it wants to write.
That check or house keeping is done at 100ms or 250ms.
If the check is done and is correct... it will write the data.
If the check says incorrect... it will report that and not write the data.
In other words... there will be a small amount of data missing and causing a skew...

Nord

[DreamDiver]

There’s also a confusion on the board where some of the users are ripping the card out of the device before its closed the session (it doesn’t make sense to write data every single second)

What is being said here? "Session" is defined in ResMed literature as being a 24-hour period, from noon to noon. Are they saying that the card should not be removed from the S9 machine until after noon? Or are they referring to "session" as the period of time when the machine is actually being used, with the blower on (perhaps including a bit of time after the machine is turned off to allow the machine to write all the data to the card)?

I don't think people are "ripping the card out of the device" while the machine is still running, so I'm not sure what they're saying. Can we get the suggested procedure clarified by ResMed, as the verbage is very unclear.

I may not have all the information perfectly correct but what it amounts to is this: after pressing the on/off button and taking your mask off in the morning, wait a few minutes before taking the card out. Here's why:

The data is written to the sd card under three separate circumstances:
once a minute, perhaps to add detailed data and to update compliance data,
once every two minutes, perhaps to add hi-res data,
once immediately for each apnea or hypopnea.

Exactly which data is written at what period is not as important as just knowing that the data is written in such a manner.
In the example below, if you turn off the machine at the designated time and then remove the card before the last two minute session one minute session have finished writing, you'll see the example apnea show up in a place where there is perhaps pressure data but no flow data. It's as if it's hanging out there without out supporting flow data as proof because we didn't wait long enough for the last bit of data to write to the card.

[kennethryan]

Nord,

Same as the other poster above, I am planning to respond to your PM once I've been using the machine a couple weeks. It's only been about 9 days so far. (And no skew yet).

ken

[Nord]

There’s also a confusion on the board where some of the users are ripping the card out of the device before its closed the session (it doesn’t make sense to write data every single second)

What is being said here? "Session" is defined in ResMed literature as being a 24-hour period, from noon to noon. Are they saying that the card should not be removed from the S9 machine until after noon? Or are they referring to "session" as the period of time when the machine is actually being used, with the blower on (perhaps including a bit of time after the machine is turned off to allow the machine to write all the data to the card)?

I don't think people are "ripping the card out of the device" while the machine is still running, so I'm not sure what they're saying. Can we get the suggested procedure clarified by ResMed, as the verbage is very unclear.

Hi DreamOn

I believe, ResMed is saying that the card is being removed from the S9 too soon after it has been shut down. I have been thinking about the fact that Flow is written once per minute to the card. I have looked at many, many Flow Data files that were timestamped and 99.9% of the Flow Data files are all exactly minutes long. There are NONE that are some hours, some minutes and some seconds long.

I don't know about everybody else but I have tried to get the S9 to start and end on different numbers so that the data would be more exact. No chance.
The S9 rounds off data either at beginning, end or both ends... so that there is less data by as much as nearly 2 minutes.
That wouldn't necessarily skew the data... as long as the data in the Other Detailed graphs also keep in step with the FLOW and either drop data or start after the Flow has started.

Its a little manoeuvering but the S9 does it each minute and does it pretty well except when something else interferes:

The SD card and its speed for housekeeping (CRC)
We remove the SD card before it gets to write its final details... we should leave the card in at the end of each mask session for at least several minutes (just to be sure)

Nord