YH550 (BreathCare ECO Auto‐CPAP, A B Models)
The file consists of two main sections:
- File header
- Log records
The file header contains fixed length fields and is 51 bytes in length. It includes machine settings and a summary of the logs. Most numeric fields are unsigned bytes.
Dates are formatted with a single byte for year, month, day, hour, minute and seconds.
Years are 2 digit (25=2025 etc.)
Decimal values are represented as the value multiplied by 10
File is little endian
| Offset | Length | Data |
|---|---|---|
| 0 | 6 | Log start date/time |
| 6 | 6 | Log end date/time |
| 12 | 1 | Mode: 0x00=CPAP, 0x01=APAP |
| 13 | 1 | Ramp up time (minutes, integer) |
| 14 | 1 | Initial pressure (Decimal) |
| 15 | 1 | Minimum pressure (Decimal) |
| 16 | 1 | Maximum pressure (Decimal) |
| 17 | 1 | unknown |
| 18 | 1 | Humidity setting (integer) |
| 19 | 7 | unknown |
| 26 | 1 | Average leak volume (Decimal) |
| 27 | 1 | unknown |
| 28 | 1 | Average pressure (Decimal) |
| 29 | 1 | unknown |
| 30 | 16 | Product serial number |
| 46 | 2 | Record count |
| 48 | 2 | unknown (possibly related to record count) |
| 50 | 1 | Header end marker (0xF9) |
Each record represents a minute of data containing the pressure level and leak level as well as event flags, most of which are currently not known. Each record is 10 bytes in length.
According to the documentation, the file could contain SPo2 and pulse data. Which could be the fields with no data as my unit does not measure either.
| Offset | Length | Data |
|---|---|---|
| 0 | 1 | Pressure cmH2O (Decimal) |
| 1 | 1 | unknown (No data) |
| 2 | 1 | unknown (No data) |
| 3 | 1 | oai (Values seen are 0, 1) |
| 4 | 1 | hi (Values seen are 0, 1, 2) |
| 5 | 1 | cai (Values seen are 0, 1, 2) |
| 6 | 1 | unknown (Values seen are 0, 1) |
| 7 | 1 | unknown (No data) |
| 8 | 1 | unknown (No data) |
| 9 | 1 | Leak volume Lpm (Decimal) |
NOTE: The user manuals for the Yuwell units are included, there are no copyright terms in either documents.
From what I understand, some Yuwell CPAP/APAP machines do provide the following:
- SPo2 data
- Pulse data
It could be possible that this grammar includes this.
The binary grammar coverage currently stands at ~81%
There are several bytes in this dataset which currently have no identified data:
With data
- Bytes 49 and 50 of the header record, which appear to be closely related to the record count in bytes 47 and 48, usually off by one
- Byte 7 of the log record, values seen are
[0, 1]indicating some kind of event although I've not observed matching events in the Breathcare station software
Without data
- Header bytes 18, 20, 28 and 30
- Log record bytes 2, 3, 8 and 9
I have only observed zeroes in these bytes.