S9 Skew investigation - your help needed!

[dave21]

To eliminate the PC factor, I've used just the ERASE option on the S9 to see if that makes a difference. Now that I've had skewing for a week where I haven't power cycled the machine, I'm now resetting to a baseline and also won't power cycle the machine and see how quickly the skew comes back. Then in a weeks time I'll ERASE the card in the S9 again but run another week power cycling at least once per day (just prior to mask session) and see how long it takes to skew.

This should hopefully give me a good analysis if anything prolongs the skew or reduces/nullifies the skew in any way.

Thanks
Dave

[DreamDiver]

DreamDiver,
In theory, the S9 side of the brick supplies the power required by the S9, regardless of slight voltage variations. Having said that, the thread about battery back up implies that a near perfect back up power source is required to supply power to the brick.

I know I tossed this theory into the pot, but I also know I'm not going to be the one who'll be able to figure out whether there is a way to measure if it's true. I only thought about it because I thought 'Hmmm... What if anything other than sixty cycles throws some internal time-stamping chip on the S9 into digital psychosis, causing the skew?' It was late. I guessing this is a dead lead anyway.

... By having 12 mask sessions in a day is basically only advancing the onset to the point at that you are accelerating the amount of time (effectively folding it like a time warp) to get to that figure where you see the skew but quicker.

I agree. The increase in skew is in some way proportional to the increase in number of mask events on a card since format - not 24-hour sessions.

To eliminate the PC factor, I've used just the ERASE option on the S9 to see if that makes a difference. Now that I've had skewing for a week where I haven't power cycled the machine, I'm now resetting to a baseline and also won't power cycle the machine and see how quickly the skew comes back. Then in a weeks time I'll ERASE the card in the S9 again but run another week power cycling at least once per day (just prior to mask session) and see how long it takes to skew.

This sounds like a great approach. I'd been wondering about formatting it on the S9 too. I'm curious as to whether it deletes all your compliance data on the machine. That's the only thing that has concerned me about trying it. I'm guessing compliance data is conserved on the S9, yes?

[dave21]

This sounds like a great approach. I'd been wondering about formatting it on the S9 too. I'm curious as to whether it deletes all your compliance data on the machine. That's the only thing that has concerned me about trying it. I'm guessing compliance data is conserved on the S9, yes?

It didn't delete the compliance data, e.g. it keeps the run time (521.2 hours), I've created a new patient profile and imported the card into this to check, it did wipe all the summary data (it didn't write the old set of summary data or detailed or hi-res data, that's all now been deleted from the machine).

Good thing is I have daily backups of my card, so either way I can reconstruct the card to bring back the summary data if I really need to, but I do also have it all imported into ResScan anyway.

Even after performing the ERASE on the card (and S9), Flow starts 1 second early but ends 1 second later than the other graphs for the first night. The same goes for the second night after ERASE.

My initial thoughts are does anyone have any graph line up with the start/end of mask session so that the flow starts and stops identically with other graphs? If so, then this could be the start of a skew, showing there's definitely something not right in the S9. If not, then maybe it's just normal use.

Thanks
Dave

[dave21]

My initial thoughts are does anyone have any graph line up with the start/end of mask session so that the flow starts and stops identically with other graphs? If so, then this could be the start of a skew, showing there's definitely something not right in the S9. If not, then maybe it's just normal use.

Answering my own question, I checked back and unfortunately when I started with my S9 on Feb 15th, prior to fully importing into ResScan, I didn't have ResScan until Mar 1st I did lose all my Hi-Res data from Feb 15th to Feb 28th so I can't check what the machine was like out of box comparing the Flow graph.

Looking at the flow data I had on the Mar 1st, I had 4 second skew where Flow was starting 4 seconds earlier at start of mask session and 1 second skew where Flow stopped 1 second early.

For me it would be good to understand what others saw when they first started using the S9 -or- after an ERASE, if you have the Hi-Res data, go back and check your first night and see whether the Flow started early/later than other graphs and finished early/later than other graphs.

From my understanding really the FLOW graph should match up exactly, even after an ERASE, bearing in mind all other graphs match up, so this would suggest something a foot (as Sherlock Holmes would say)

Thanks
Dave

[Nord]

Dave

Were you using your ScanDisk HDSD card all this time... ???

ResMed has told DreamDiver that the fault of the skew is based on some faulty SD cards by ResMed.

I have read the ScanDisk article about CRC files and being 2 different speeds of 100ms and 250ms...
I'm not sure what category my own ScanDisk 2 Gig card is... I know its class 2 but may still be the slower read/write timeout.
I know alterego61 has been using a class 4 without any skew issues.
Uncle Bob switched to a class 2 and he had no more skew...

I am going to post some other thoughts about the skew on that thread. I think there may be several other issues in play.

Nord

[alterego61]

Hi all,

I've been using a 16 GB class 4 card from day 2 (now on day 13). I noticed my first skew yesterday. I've been very busy at work and I haven't had a chance to look back and figure out when it started.

[Nord]

Hi all,

I've been using a 16 GB class 4 card from day 2 (now on day 13). I noticed my first skew yesterday. I've been very busy at work and I haven't had a chance to look back and figure out when it started.

Hi alterego

Look forward to hearing... although I was hoping that ResMed wasn't just giving an easy answer.
I guess they were... back to the drawing board.

Nord

[Laurie1041]

"Power-cycling" caught my eye. OK, don't laugh but what is "power-cycling"? Is this unplugging and plugging the S9 back in before use? What is the technical reason behind power-cycling every night?
Thanks! Laurie

[DreamDiver]

"Power-cycling" caught my eye. OK, don't laugh but what is "power-cycling"? Is this unplugging and plugging the S9 back in before use? What is the technical reason behind power-cycling every night?
Thanks! Laurie

You are correct - power-cycling refers to unplugging your machine, but the original reasons behind the practice were purely theoretical and now proven to be unfounded. Power-cycling was originally done after you turned off the machine as part of a forum-member-run experiment to see if turning off the machine between mask sessions helped reset skew. It didn't.

The current best practice for best data control:

• Wait at least two minutes before unplugging the machine after turning it off by the power button.
  Why? Because there is a two-minute 'write' cycle for the SD card. If you had any apneas within two minutes of turning off the machine, this will give the machine a chance to write the last bit of data to the SD card. Go for a bathroom break or do your exercises, then unplug the machine. Alternatively, if you don't care whether the flow data matches the AHI data for the last two minutes, don't worry about it.
• Upload your data to your computer at least once a week if you don't want to lose flow data. Remember to 'discard duplicate data' during the process if you don't want to lose your flow data. This is a complicated issue because 'duplicate data' isn't really duplicate data. With every mask session, the flow generator firmware assesses the date values of all sessions already on the card and deletes all hi-resolution data (in most cases, flow data) from the card if it is older than one week/eight days. So if you overwrite your existing data to your computer, you end up removing flow data from your computer if you select 'overwrite duplicate data' because the existing data on the card no longer contains flow data for records one week old or older. I suspect the same is true for medium-resolution data after one month, but haven't tested my theory because I want to keep my data. ResMed should consider making it so that ResScan 3.12 remembers this preference instead of asking every time.
• Reformat the card every two weeks to reset skew back to almost-zero. If you just leave the card in for a month or six, by the end of that time, the flow data will skew widely from the Apnea and Hypopnea data points in ResScan -- so much so that corroboration between flow and apneas is nearly impossible to corroborate because they won't even be on the same screen. If your doctor needs data, and you don't keep best practices because it's not important to you, please consider at least reformatting your card two weeks before your doctor needs the data so the data will be on its best behavior for the doctor.
• Backup your patient data as often as you deem appropriate if you don't do regular computer system backups.

A little more about skew...
The problem is in the firmware of the S9 flow generator and not within ResScan. If you have a machine with skew, there is nothing you can do about it. From what we have discovered, all American and European machines are affected by skew. There are apparently some machines of Australian origin with no skew.

We sort of gave up on this a while back due to friction amongst some forum members. This was originally an experiment for only S9 users who were interested, but many who weren't interested thought we were telling them how to run their machines. Quite the opposite. This set of Skew threads was intended only for those who wanted to help figure out if there was a way to fix the firmware bug by best practices. There wasn't. All we can do is reformat the sd card every two weeks to reset the skew to an acceptable level for visual corroboration of data in ResScan.

Also apparently, ResMed will not recall or fix the skew on existing machines because a readout bug does not (again apparently) put patients in danger with regards to receiving flow generator therapy. It's cheaper for them to just write this mistake off.

That said, major software bugs are not unusual in the world of flow generators. The Respironics M-Series Pro has its own set of software bugs. Owners lose one to two whole days of data every four weeks regardless of whether they reformat their smart card or not. It just disappears. Nobody knows why, but again since a firmware bug for data graphing doesn't put patients in direct harm, it's not considered worth fixing.

As users, we must decide how Monkish we are and do what makes us comfortable when dealing with software glitches. It would be nice if the makers would do a better job of vetting the machines before marketing them. They would do well to approach the squints on this forum and ask us to test them first.

[ignorant1]

An interesting thread.

As a prospective future S9 purchaser, I am curious to know if ResMed has acknowledged this problem, AND IF SO, does anyone know if newer machines will receive an upgraded firmware to resolve this problem?

[DreamDiver]

An interesting thread.

As a prospective future S9 purchaser, I am curious to know if ResMed has acknowledged this problem, AND IF SO, does anyone know if newer machines will receive an upgraded firmware to resolve this problem?

This is unlikely. According to one Resmed representative I corresponded with, when a machine has gotten governmental approval, it is for the entire machine with existing specifications. If they make any changes, the entire flow generator needs to go through the assessment process again - for multiple governments all over the world. It's a costly process. The spec has already been approved as is. They may as well concentrate their efforts on a totally new design and incorporate what they've learned in the new design.

The M-Series Pro's firmware was never fixed even though it's associated humidifier went through several iterations due to a leaky design. It would probably have been grossly more expensive to fix the flow-generator than to fix the humidifier.

Frankly, it would be wiser if there were an object-oriented component approval system so firmware could be swapped without having to redo the entire approval process. It wouldn't be hard to do -- just a little more tedious and perhaps more expensive up front the first time. The machine should be designed for firmware and component updates. Get the whole design/update system approved. The end result would be better machines and the capability of fixing errors found only after long-term real-world experience.

In both instances -- S9 and M-Series Pro -- the problem only shows up after long-term use and is not directly affecting patient therapy. A component system designed to trap real-world data and incorporate it into design flaw reduction on-the-fly would be a major advantage in flow-generator design.

[Nord]

An interesting thread.

As a prospective future S9 purchaser, I am curious to know if ResMed has acknowledged this problem, AND IF SO, does anyone know if newer machines will receive an upgraded firmware to resolve this problem?

ResMed does NOT acknowledge the problem in my experience... partially because their "customers" are not the end users and thus NOT an important part of the sales chain.

That said... the S9 has many positive features that outweigh the "skew" inconvenience in my opinion. The skew is "fixed" by reformatting the SD card at a certain point that relates to the amount of times powering "on". It is a minor inconvenience. I look very closely at my Flow data and I believe that there are some approximations in CPAP derived numbers and the information should be used to give you trends and issues as opposed to definitive information.

There has been some upgrades to original firmware that has NOT fixed the skew... but if they don't recognize an issue with it... how can they solve it.

I would not hold my breath for a hardware/firmware solution (that would be another Central Apnea )... base your future choice on features between machines knowing that each manufacturer has some issues with their products.

Nord

[DreamDiver]

... base your future choice on features between machines knowing that each manufacturer has some issues with their products.

Nord

I agree. Even with its quirks, I'd have the S9 over any other machine. It really does provide better therapy for me. Others find the System One works best for them. YMMV.

Actual therapy trumps quirky data.

[idamtnboy]

I agree. Even with its quirks, I'd have the S9 over any other machine. It really does provide better therapy for me. Others find the System One works best for them. YMMV.

Actual therapy trumps quirky data.

You worked on the data skew issue a lot last spring. What's your take on what I present in my thread today about S9 data being a Resscan problem, not a machine or card problem?

[cwied]

I agree. Even with its quirks, I'd have the S9 over any other machine. It really does provide better therapy for me. Others find the System One works best for them. YMMV.

Actual therapy trumps quirky data.

You worked on the data skew issue a lot last spring. What's your take on what I present in my thread today about S9 data being a Resscan problem, not a machine or card problem?

Just my 2c: I just installed Resscan on a new laptop and downloaded the data from my SD card. The skew was present, which I think implicates the firmware more than Resscan.