Other Parts Discussed in Thread: MMWCAS-RF-EVM
Hello TI Team,
I have modified the default Cascade_Capture.lua script provided by TI to run in a loop for automated data capture. The goal is to scan an aperture for imaging application. I am intermittently facing a TDA ARM Failed error where the TDA capture card fails to respond to the start command. The specific error in the log is: [RadarAPI]: Frame Periodicity Response from Capture Card timed out!!
--[[
A. FRAMING & CAPTURE
1. Triggering Slave (3, 2, 1) sequentially in a hardware triggered mode.
2. Triggering Master in a software triggered mode.
B. TRANSFERRING FILES
1. The data is stored in file(s) with max cap placed at 2 GB.
2. The files can be retrieved from the SSD (/mnt/ssd folder) using WinSCP.
Note: Update lines 18 to 49 as needed before using this script.
--
Note: "capture_time" is a timeout for this script alone to exit - it does not control the
actual duration of capture. The actual capture duration depends on the configured frame time
and number of frames.
--]]
capture_time = 2000 -- ms
inter_loop_time = 7000 -- ms
num_loops = 98
--[[
Note: Change the following three parameters as desired:
1. n_files_allocation: is the number of files to preallocate on the SSD.
This improves capture reliability by not having frame drops while switching files.
The tradeoff is that each file is a fixed 2047 MB even if a smaller number of frames are captured.
Pre-allocate as many files as needed based on (size_per_frame * number_of_frames) to be captured.
2. data_packaging: select whether to use 16-bit ADC data as is, or drop 4 lsbits and save 4*12-bit numbers in a packed form
This allows a higher frame rate to be achieved, at the expense of some post-processing to unpack the data later.
(Matlab should still be able to unpack the data using the '*ubit12' argument to fread instead of 'uint16')
The default is no-packing, for simplicity
3. capture_directory: is the filename under which captures are stored on the SSD
and is also the directory to which files will be transferred back to the host
The captures are copied to the PostProc folder within mmWave Studio
Note: If this script is called multiple times without changing the directory name, then all
captured files will be in the same directory with filename suffixes incremented automatically.
It may be hard to know which captured files correspond to which run of the script.
Note: It is strongly recommended to change this directory name between captures.
--]]
n_files_allocation = 0
data_packaging = 0 -- 0: 16-bit, 1: 12-bit
base_directory = "new_horizontal"
num_frames_to_capture = 0 -- 0: default case; Any positive value - number of frames to capture
framing_type = 1 -- 0: infinite, 1: finite
stop_frame_mode = 0 -- 0: Frame boundary, 2: Sub-frame boundary,
-- 3: Burst boundary, 4: HW/Sub-frame triggered
----------------------------------DATA CAPTURE-------------------------------------------
-- Function to start/stop frame
function Framing_Control(Device_ID, En1_Dis0)
local status = 0
if (En1_Dis0 == 1) then
status = ar1.StartFrame_mult(dev_list[Device_ID]) --Start Trigger Frame
if (status == 0) then
WriteToLog("Device "..Device_ID.." : Start Frame Successful\n", "green")
else
WriteToLog("Device "..Device_ID.." : Start Frame Failed\n", "red")
return -5
end
else
status = ar1.StopFrame_mult(dev_list[Device_ID], stop_frame_mode) --Stop Trigger Frame
if (status == 0) then
WriteToLog("Device "..Device_ID.." : Stop Frame Successful\n", "green")
else
WriteToLog("Device "..Device_ID.." : Stop Frame Failed\n", "red")
return -5
end
end
return status
end
-- Loop counter
loop_index = 1
while (loop_index <= num_loops)
do
capture_directory = base_directory .. "_" .. tostring(loop_index)
WriteToLog("Starting loop " .. loop_index .. " -> Capture Directory: " .. capture_directory .. "\n", "purple")
-- TDA ARM
WriteToLog("Starting TDA ARM...\n", "blue")
status = ar1.TDACaptureCard_StartRecord_mult(1, n_files_allocation, data_packaging, capture_directory, num_frames_to_capture)
if (status == 0) then
WriteToLog("TDA ARM Successful\n", "green")
else
WriteToLog("TDA ARM Failed\n", "red")
return -5
end
-- RSTD.Sleep(1000)
-- Triggering the data capture
WriteToLog("Starting Frame Trigger sequence...\n", "blue")
if (RadarDevice[4]==1)then
Framing_Control(4,1)
end
if (RadarDevice[3]==1)then
Framing_Control(3,1)
end
if (RadarDevice[2]==1)then
Framing_Control(2,1)
end
Framing_Control(1,1)
WriteToLog("Capturing AWR device data to the TDA SSD...\n", "blue")
if (framing_type == 0) then
-- Stop capturing
WriteToLog("Starting Frame Stop sequence...\n", "blue")
if (RadarDevice[4]==1)then
Framing_Control(4,0)
end
if (RadarDevice[3]==1)then
Framing_Control(3,0)
end
if (RadarDevice[2]==1)then
Framing_Control(2,0)
end
Framing_Control(1,0)
end
WriteToLog("Capture sequence for loop " .. loop_index .. " completed...\n", "blue")
loop_index = loop_index + 1
RSTD.Sleep(inter_loop_time)
endThis setup previously worked perfectly for 400+ continuous loops with a much shorter inter-loop delay. However, now I am seeing this timeout occur randomly.
-
Sometimes it fails at Loop 2.
-
Other times it runs successfully until Loop 30 and then fails.
-
Increasing the
inter_loop_timedelay gradually from 2s to 7s has not solved the issue.
If the script worked previously without the Stop command, what conditions (SSD state, buffer status, etc.) would cause the TDA to suddenly start timing out on the ARM command?
Any insights on debugging this intermittent TDA hang would be appreciated.
Thanks in advance,
Sakshi Verma
