IQ sample data format

[Tesla_wei]

Hi,

I have got some problem of understanding the IQ sampling data.

As below is the official data format

0: I_Q (default)	Q	31:16	12 bits signed, sign extended to 16 bits
	I	15:0

Does it mean for each IQ sample, I have 16bits Q then 16bits I?

And in each 16bits, the first 4 bits should be either 0xF or 0x0, because the first 4 bits are the sign bit extension, is it correct?

Then rest of the 12bits are the value converted from ADC, does it matter if it is int or float?

I got some IQ sample data, but the format is different from my expectation, I put some of them below, it just represents the first 4 IQ samples (or two sets),

F4FF, EBFF, FAFF, 0C00,

As you can see, the second sample starts with 0xE, not 0xF or 0x0, I have bunch of similar data not start with 0xF or 0x0,

Would you confirm if my understanding is correct?

Thanks

Pengyu

[Dmitry]

Hi,

Tesla_wei said:

Does it mean for each IQ sample, I have 16bits Q then 16bits I?

yes.

Tesla_wei said:

And in each 16bits, the first 4 bits should be either 0xF or 0x0, because the first 4 bits are the sign bit extension, is it correct?

yes.

Tesla_wei said:

Then rest of the 12bits are the value converted from ADC, does it matter if it is int or float?

it's a signed integer value.

Tesla_wei said:

I got some IQ sample data, but the format is different from my expectation, I put some of them below, it just represents the first 4 IQ samples (or two sets),

F4FF, EBFF, FAFF, 0C00,

The values are in LSB byte order - if you swap bytes, you'll get right results. The real range when tested on my 52811 was about -300..+300 most of the time.

[Tesla_wei]

Thanks, that make sense to me.

Would you please share a segment of your IQ sample data for reference?