Unable to perform IQ sampling normally

[honghuahh]

SW version :2.4.0
HW revision: nRF52833 development board, PCA20054 Direction Finding Antenna Board 1_0_0
I use the routine director_finding_central  and  director_finding_periopheral for IQ sampling, and can obtain IQ sampling data.
When I put the original
static const uint8_t ant_patterns[] = { 0x2, 0x0, 0x5, 0x6, 0x1, 0x4,
0xC, 0x9, 0xE, 0xD, 0x8, 0xA };
Change to
static const uint8_t ant_patterns[] = { 0xC, 0x9, 0xE, 0xD, 0x8, 0xA,
0x2, 0x0, 0x5, 0x6, 0x1, 0x4 };
The Bluetooth connection will disconnect, making it impossible to perform normal IQ sampling.
May I ask why this situation has occurred?

Below is the serial port output

[honghuahh]

I did some tests.

When static const uint8_t ant_patterns[] = {0xC, 0x9, 0xE, 0xD, 0x8, 0xA,
0x2, 0x0, 0x5, 0x6, 0x1, 0x4 };

The position of the two boards is shown below. The central can collect some IQ data, and the packet CRC ERROR often occurs. After a period of time, the connection will be disconnected

Then move the central position to the position shown in the figure, and the connection will immediately disconnect,

and IQ data cannot be collected

 

But when I used the original

static const uint8_t ant_patterns[] = { 0x2, 0x0, 0x5, 0x6, 0x1, 0x4,
0xC, 0x9, 0xE, 0xD, 0x8, 0xA };

central is in the top two positions, the connection and IQ sampling are normal, this phenomenon is very strange, why the connection is related to the order of antenna switching