Raw data unfiltered/filtered is in floating point for recent transmitter firmware

[tynbendad]

I have an anubis transmitter and cannot read raw data with xdrip+, instead I get "Raw: -0.13" in the bg data table.

I looked at the sensor rx messages using xdrip-js and it looks to me like my transmitter uses floating point for the unfiltered, filtered fields. here is my decoding of the messages I looked at when given controlled input from resistors:

it looks like xdrip+ is using an integer calculation to come up with its unfiltered result (SensorRxMessage.java only reads in integer values, and they are used in https://github.com/NightscoutFoundation/xDrip/blob/master/app/src/main/java/com/eveningoutpost/dexdrip/G5Model/RawScaling.java) - from my calculations this will not give the same results, e.g., for 400mg/dl the unfiltered will read approx 230mg/dl - this would be fine for calibration if the result was linear but since the exponent is not decoded the integer calculation currently in xdrip+ will not be linear.

i still don't understand why my raw was -0.13, but i believe there is at least one xdrip+ issue here.

my anubis transmitter is at firmware version 2.27.2.103, and my xdrip+ is at version 2022.04.06 (variant 3).

[tynbendad]

btw, I believe for the floating point data 35 is a better multiplier than 34 - using 35 the results for the same resistances are a closer match to the values I see on a 80xxxx transmitter (which uses 34.)

[tolot27]

@joshbohmer Any ideas?

[joshbohmer]

I think he's correct, I can confirm floats are used in places. I would need to spend time to confirm this one specifically but my initial impression is this would be a significant improvement in RAW accuracy over the current implementation

[tynbendad]

you can do the math on the values i posted in the first image - at 40MOhms the unfiltered should read about 500mg/dl (this is what my 80xxxx transmitter reads). interpreted as a fp value and multiplied by 35 it matches, while interpreted as xdrip+ currently does you get around 300mg/dl.

[LoriLori]

@tynbendad what amount of MOhms could I connect to transmitter contacts that I would get value of 100mg/dl?

[tynbendad]

@LoriLori 200MOhms should give you a raw of about 102

[Navid200]

What would be the reason for using raw values with a G6?
What would that enable you to do that you cannot already do using the native values coming from the transmitter?

[tynbendad]

You can get raw values even when you aren't getting native values.
You can correlate the raw values to the native values when you are getting them.

[Navid200]

Can you share an example that you have verified/tested?

For example, when the G6 transmitter issues a confused calibration error and stops producing readings for a while, does it keep sending raw values during that time?

[tynbendad]

this is uploaded by logger/xdrip-js, not xdrip+ but it shows what you are asking. raw is always sent by the tx, even for temporary session errors and sensor stopped/warmup...

[tolot27]

raw is always sent by the tx, even for temporary session errors and sensor stopped/warmup...

Does it mean that you can get preliminary raw values during warmup period? That would be an interesting feature.

[tynbendad]

yes, absolutely. we've been using raw for warmup, etc. since g5, others have used it since g4. on g6 it only works on 80xxxx,81xxxx, anubis, and maybe a scant few other tx versions.

[Navid200]

#85

[tynbendad]

#2094

[tynbendad]

this discussion can be closed, @jamorham has made the change in c5eb162