v20.6(7):
1) added async logic for anytrigger mode, so the anytrigger signal will last for 10us(100MHz).
therefore, the other 3 quabos will not miss the anytrigger signal.
2) registered the 256 pixel signals, and used LUT to speed up the logic for the timing requirement.
v20.4(5):
1) added any trigger.
when one quabo gets a ph event, other three quabos will be triggered by the ANYTRIG pin.
You should set bit8 in CHANMASK_8 to 0 to enable this function.
2) two or three pixels trigger a ph event.
You need to set bit0 or bit1 in GOEMASK to 0 in config file.
The two bits are set to 2'b11 by default.
v20.2(3):
1) added two new commands:
a) read one ph frame out for calibration;
b) write calibration data back to fpga memory.
2) fixed a bug in firmware, so we can use img and ph mode at the same time.
v20.0(1):
(1) Fast PH readout is implemented. This part is moved from microblaze software to FPGA, so it's much faster now.
(2) 10-bit TAI time is added to IM and PH packets, which is from WR.
quabo0 sends the TAI time out, and quabo1/2/3 receive the TAI time.
(3) Added two new cmds:
a) add a new cmd for setting IP addresses of IM and PH packets.
after the cmd is sent to quabo, a reply with MAC addresses will be sent from quabo.
the default IP address of IM and PH packets is 192.168.1.100.
b) add a new cmd for setting IP address of HK packets.
the default IP address of HK packets is 192.168.1.100.
(4) Any host computer in the local network is able to send cmds to quabo, and quabo will reply to this host computer(if there is a reply).
(5) PH packets contain 16-bit, signed value, which could be negative.
(6) Packet version is added to the packet, which is used for checking PH packets.
a) if the packet version is 0, it's an old PH packet, and it contains 12-bit, unsigned value;
b) if the packet version is 1, it's a new PH packet, and it contains 16-bit, signed value.
Normally, we don't care about the packet version.
v11.8(9):
(1) The bug of periodic high flux measurments is fixed.
The reason is that frame_go and a state machine in im_mode_state_machine.v were reset by pps,
but the state machine shouldn't be reset by pps.
Therefore, a pps_rst port is added for fixing the bug.
(2) Led flasher selection bit is added to the new firmware.
The default led is the new one.
To change the led, control_quabo_v10.py should be used.
(3) v11.8 is for QFP quabo, and v11.9 is for BGA quabo.
The code of the two version are 99.999% the same, and the difference is shown in v11.7 comment.
v11.7:
Made the changes necessary to support the BGA PCB design. There are two parameters that must be set to compile
the design for either the QFP (old) or BGA (new) versions. In the gateware, the maroc_dc IP (in the Block Diagram)
has an attribute "PCBrev" which can be set to either 0 (QFP) or 1 (BGA). (This maps to the Verilog parameter
REMAP_FOR_BGA in maroc_dc.v). This sets the mapping of the 256 trigger signals.
In the software, in main.c, there is a #define REMAP_FOR_BGA. This should be commented out for the QFP version.
THESE TWO PARAMETER SETTINGS MUST AGREE. We'll need to do two compiles and bitfile generations whenever we make a change in the
future, so we can accomodate both PCB revisions. I suggest we name them with an even number for QFP and odd for BGA,
eg 12.0 and 12.1.
v11.6B:
(1) In v11.6A, SPIx1 is changed to SPIx4, but it can't boot up;
In v11.6B, SPIx4 is changed back to SPIx1.
(2) Since v11.5, some code is added to fix negative value bug, and here is the code:
for (i = 0; i < 256; i++) if (ph_baseline_array[i] & 0x1000) ph_baseline_array[i] = ph_baseline_array[i] | 0xf000;
v11.6A:
(1) fix a IM packets bug--the nanosec in IM packets is not synced to PPS, which has been fixed here;
(2) fix a clock setting bug--In maroc_dc core, axi clk is regarded as 62.5MHz, but it's 100MHz;
a parameter(ADC_CLK_DIV) is changed from 2 to 4 for fixing this bug.
v11.5:
the default status of shutter_command pin is 3.3V, instead of 0V
v11.4:
add pull-down resistors for the 256 input pins
v11.3:
(1) fix the timing offset issue;
We fixed the timing offset bug--when integration time < 50us, here is a timing offset.
The reason is that we can get a pixel data from axi4-fifo every 8 clk, so it will take 2048 clk at least to get all 256 pixel data.
I chagne the axi4-fifo to a standard fifo, which is much faster.
(2) fix the ossasional udp checksum issue;
I found a bug in my state machine, and fixed it;
v11.2:
(1)implement shutter control
Focus stage shares the 4 ports with shutter control, and it depends on the quabo position.
If it's quabo0, the 4 ports are used for focus stage; if it's quabo1, the 4 ports are used for shutter control;
v11.1:
(1) add mode 6--8-BIT IM mode;
(2) add mode 7--8-BIT IM mode simultaneous with 16 pulse height mode;
(3) time tick is changed to 1us;
(4) acq_int is changed to 16 bit, so the max integration time is 65536us = 65.536ms;
v11.0:
It's the same as v10.5F.
Because we have a lot of new features, we release v11.0.
v10.5F:
(1) fix a reset bug
The elapsed_time counter is 29bits, and it increases every 4ns, so I move 2 bits to left, making the resolution 1ns in the packets.
In verilog , it's
elapsed_time_reg <= {(elapsed_time<<2)};
I thought (elapsed_time<<2) is 31bits, but it looks it's still 29bits, and the upper 2bits are cut,
that's why we lost some upper bits.
I made the changes like this:
wire [31:0] elapsed_time_wire;
assign elapsed_time_wire = {3'b0,elapsed_time};
elapsed_time_reg <= {(elapsed_time_wire<<2)}
v10.5E:
(1) change the reset design of bin_counter.
The reset will start before im_mode_state_machine sending data
v10.5D:
(1) fix ip checksum bug
v10.5C:
(1) constraints added in placement.xdc;
create_pblock pblock_ctr_reset
add_cells_to_pblock [get_pblocks pblock_ctr_reset] [get_cells -quiet [list base_mb_i/elapsed_time_gen_0/inst/counter_reset_1_reg base_mb_i/elapsed_time_gen_0/inst/counter_reset_2_reg base_mb_i/elapsed_time_gen_0/inst/counter_reset_3_reg]]
resize_pblock [get_pblocks pblock_ctr_reset] -add {SLICE_X30Y39:SLICE_X31Y40}
(2) constraints changed in timing.xdc;
a) remove--#set_false_path -from [get_clocks -of_objects [get_pins base_mb_i/clk_wiz_0/inst/mmcm_adv_inst/CLKOUT5]] -to [get_clocks -of_objects [get_pins base_mb_i/clk_wiz_1/inst/mmcm_adv_inst/CLKOUT0]]
b) remove--#set_false_path -from [get_clocks -of_objects [get_pins base_mb_i/clk_wiz_1/inst/mmcm_adv_inst/CLKOUT0]] -to [get_clocks -of_objects [get_pins base_mb_i/clk_wiz_1/inst/mmcm_adv_inst/CLKOUT1]]
c) add--set_false_path -from [get_clocks -of_objects [get_pins base_mb_i/clk_wiz_1/inst/mmcm_adv_inst/CLKOUT0]] -to [get_pins {base_mb_i/maroc_dc_0/inst/USR_LOGIC/IM_MUX[*].MUX_L/dout_reg[*]/D}]
(3) change some code in bin_counter for removing the deadtime;
(4) the reset of im state machine is changed to ".frame_reset((!mode_enable[1])),".
v10.5B:
Remove all debugging signals, and add placement constraints in placement.xdc
v10.5A:
(1) We get IM packets(ACQ_MODE=2) from hardware instead of software, so we can get high speed packets;
(2) We can set ACQ_MODE to 0, so that no packets will be sent out except HK packets.
v10.5:
(1) finish high speed IM mode(ACQ_MODE = 3).
We can get PH packets and IM packets at the same time:
when it's PH packet, its ACQ_MODE = 1;
when it's IM packet, its ACQ_MODE = 3;
(2) I just finished function test,including UDP packets checksum, timing difference,
but I'm not sure whether the scientific data is correct;
We need more tests on it;
v10.4:
(1) replace set_focus function with rick's new code to solve issue of stage stops when it reacges tge upper optical switch;
(2) fix the system crashed issue(a fifo in eth_core_ctrl core was full...);
(3) fix the issue of losing connection to the last quabo on mobo;
The second eth port on the last quabo is open, so it can't finish the auto-negotiation. I configure the second eth port to fixed 1000Mbps, not auto-negotiation;
(4) fix mac address issue;
some mac addresses are illegal, and it's related to the first byte;
I set the first byte of mac address to 0x00 to solve the issue;
(5) fix the issue about stage didn't work with more than 2 quabo on mobo;
focus_limits_on(output) port is connected to bus, so I change is to three state, if it's quabo1/2/3;