Sample CPAP Data / STR.edf parsing.

[AJolly]

Is there a collection of sample ORIGINAL CPAP data available anywhere for testing, especially for validation of settings parsing?
10 .edf's, in particular official str.edf's from bipap and ASV/ASVAuto (aircurv) machines. Extra points for iVAPS , AirCurv 10 ST-A and Stellar devices.

When switching fw modes, various tools (Oscar, multi-glasgow indexer) parse STR.edf incorrectly, and I'd like to run down why:

1. Figure out if its the fw causing the problem or the tools.
2. Either see if we can fix the fw to create more "compliant" logs, and/or update the tools so they can read the settings correctly.

[luckyabsoluter]

I have some firmware. Please, contact me https://signal.me/#eu/3olOca_-8RIt9xJUzRONgYToCZMBTTVXkhTqyPv8JSUbEuBp5HZExLXeM-LG3Qh2

Code	Rev	Name
37043	0305	AirCurve™ 10 ASV
37051	0302	AirCurve 10 VAuto
37051	0401	AirCurve 10 VAuto
37031	0305	AirSense™ 10 AutoSet™
37028	0306	AirSense™ 10 AutoSet™
37028	0401	AirSense™ 10 AutoSet™
37031	0305	AirSense™ 10 AutoSet™
37032	0305	AirSense™ 10 AutoSet™
37033	0305	AirSense™ 10 AutoSet™
37090	0401	AirSense™ 10 AutoSet™
37101	0401	AirSense™ 10 AutoSet™
37105	0401	AirSense 10 AutoSet For Her
37117	0401	AirSense 10 Elite

[m-kozlowski]

mkozlowski@mgmt:/storage/cpap-files/resmed-firmware# pr -mt   <(strings firmware-aircurve-37051-SX567-0401.bin | grep ^S\\.) <(strings firmware-airsense-37101-SX567-0401.bin | grep ^S\\.) |less
S.RampEnable                        S.RampEnable
S.RampTime                          S.RampTime
S.C.StartPress                      S.C.StartPress
S.C.Press                           S.C.Press
S.EPR.ClinEnable                    S.EPR.ClinEnable
S.EPR.EPREnable                     S.EPR.EPREnable
S.EPR.Level                         S.EPR.Level
S.EPR.EPRType                       S.EPR.EPRType
S.BL.StartPress                     S.AS.Comfort
S.BL.IPAP                           S.AS.StartPress
S.BL.EPAP                           S.AS.MaxPress
S.EasyBreathe                       S.AS.MinPress
S.VA.StartPress                     S.SmartStart
S.VA.MaxIPAP                        S.PtAccess
S.VA.MinEPAP                        S.ABFilter
S.VA.PS                             S.Mask
S.RiseEnable                        S.Tube
S.RiseTime                          S.ClimateControl
S.Cycle                             S.HumEnable
S.Trigger                           S.HumLevel
S.TiMax                             S.TempEnable
S.TiMin                             S.Temp
S.SmartStart
S.PtAccess
S.ABFilter
S.LeakAlert
S.Mask
S.Tube
S.ClimateControl
S.HumEnable
S.HumLevel
S.TempEnable
S.Temp

¯\(ツ)/¯

[m-kozlowski]

Seems doable, although there are quite a few interdependencies between structures from different globals[] sections that need to be sorted out. Keeping Airbreak's current "dump whatever firmware the user has > patch > flash > profit!" approach will be… challenging ;)

Everything ultimately comes down to changes in the CCX block - the 240 KB provisioning image. The firmware code itself is shared across all "flavors". Mapping var -> signal appears to be common and complete among firmwares too.
And there's roughly 200 KB of unused space in CCX, so there's plenty of room for record "merging."

[m-kozlowski]

I have some pretty promising results :)

There is still a lot to clean up: several structures required relocation, and I want to automate this in a safe and sensible way. There is also plenty left to test, for example whether the counters are properly put in the correct signals across different therapy modes - I will definitely need some help here, but so far it looks... really good.

STR.edf has grown to 115 signals.

With the required changes (type/variable/sample/stringptr mapping definitions plus the strings themselves), about ~1 KB of data was added to the CCX block.

	AutoSet	VAuto	ASV	Merge
STR	80	97	82	115
BRP	3	4	3	4
PLD	10	14	11	14
SAD	3	3	3	3

[m-kozlowski]

edf_ccx_merge.py
If anyone feels like testing it, I would appreciate the feedback.