search

LORD-MicroStrain / MSCL

MicroStrain Communication Library
https://www.microstrain.com/software/mscl
MIT License
76 stars 57 forks source link

Timestamp issue #390

Open 8c1h9q7hwz0cl opened 4 months ago

8c1h9q7hwz0cl commented 4 months ago

Hello, we are facing some inconsistent timestamp issues. We configured the IMU to run at 200Hz. In the ideal situation, the time difference (dt) between two consecutive IMU measurements should be 5ms. However, we observed that the dt varies between 3 and 7 ms. image

gmurrayrv9a commented 4 months ago

We are using MSCL and getting the timestamp as follows: for(mscl::MipDataPacket packet : packets) { //get the data in the packet mscl::MipDataPoints data = packet.data();

        mscl::MipDataPoint dataPoint;
        mscl::Timestamp packet_ts = packet.collectedTimestamp();

Then we use “packet_ts.nanoseconds()”

Looking at MSCL code we see that time stamps are not associated with the received accelerometer or gyroscope packets but rather stand alone in a 'internal timestamp' packet. There is code (in MSCL) to calculate delta from previous timestamp, add this to PC time and then attach this value to accelerometer/gyroscope data as it comes in. Sometimes multiple accelerometer packets are received before a time stamp packet causing multiple accelerometer data to have the same timestamp.

gmurrayrv9a commented 4 months ago

We have found a resolution that works for us. We enabled CLASS_SYSTEM channel 0xA0D5 to get TICKS and calculate our own timestamp using system time and TICKS.