Hi, this is the pulse train producer I used in my master thesis. I needed to solve a variable modulo problem in VHDL for this. By introducing an estimator I was able to do it.

Pulse train producer
--DESCRIPTION
--This entity takes three parameters once it detects a edge on edge_in
--afterwards, it will produce a pulse train in accordance with these parameters
--the ratio between the high speed clock and the low speed clock is 8
--tested with a 800MHz high speed clock
-- DELAY:
-- cycles before it starts producing the pulse train, 0-2**8-1
-- as this entity most likely will run with a faster clockj then the rest, I added
-- this parameter so you can sync stuff up beween the state machines clock cycles.
-- eg:
-- ___|^|_|^|_|^|____ (ch1)
--________|^|________ (ch2)
-- cannot be done without the delay parameter as |^| has a lenfth of about 1.25ns (800MHz clock) while minimal reaction time
-- for the state machine is 100ns (running at 10Mhz right now)
-- LENGTH:
-- width of the pulses to be produced, 1-2**8
-- eg:
-- ___|^^|____ (ch1) -> length = 2
-- ___|^|_____ (ch2) -> length = 1
-- COUNT:
-- number of pulses to be procuded, 1-2**8
-- This is needed if you want to have a pulse train.
-- Pulses will always be symmetric (so _|^|_|^|_ or _|^^^|___|^^^|___ but never _|^|___|^|___)
-- Reminder:
-- all values must be below their max values, either 0-2**8-1 or 1-2**8
LIBRARY IEEE;
USE IEEE.STD_LOGIC_1164.ALL;
USE IEEE.NUMERIC_STD.ALL;
LIBRARY UNISIM;
USE UNISIM.VCOMPONENTS.ALL;
ENTITY pulser_generic IS
PORT(
clk_in: IN STD_LOGIC;
clk_high_speed_in: IN STD_LOGIC;
edge_in: IN STD_LOGIC;
rst_in: IN STD_LOGIC;
length_in: IN Integer RANGE 1 TO 2**8;
count_in: IN Integer RANGE 1 TO 2**8;
delay_in: IN Integer RANGE 0 TO 2**8-1;
pulse_out : OUT STD_LOGIC;
finished_out: OUT STD_LOGIC
);
END pulser_generic;
ARCHITECTURE main OF pulser_generic IS
SIGNAL edge_in_old: STD_LOGIC;
SIGNAL serdes_input: STD_LOGIC_VECTOR(7 DOWNTO 0);
SIGNAL estimator: INTEGER RANGE 0 TO 2**8-1;
SIGNAL estimator_increase: INTEGER RANGE 1 TO 4;
SIGNAL chunk_counter: INTEGER RANGE -2**5 TO 2**14;
SIGNAL length_act: INTEGER RANGE 1 TO 2**8;
SIGNAL count_act: INTEGER RANGE 1 TO 2**8;
SIGNAL delay_act: INTEGER RANGE 0 TO 7;
SIGNAL running: STD_LOGIC;
BEGIN
OSERDES2_inst: OSERDESE2
GENERIC MAP(
TBYTE_CTL => "FALSE",
TBYTE_SRC => "FALSE",
DATA_RATE_OQ => "SDR",
DATA_RATE_TQ => "SDR",
DATA_WIDTH => 8,
INIT_OQ => '0',
INIT_TQ => '0',
SERDES_MODE => "MASTER",
SRVAL_OQ => '0',
SRVAL_TQ => '0',
TRISTATE_WIDTH => 1
)
PORT MAP(
OFB => open,
OQ => pulse_out,
SHIFTOUT1 => open,
SHIFTOUT2 => open,
TBYTEOUT => open,
TFB => open,
TQ => open,
CLK => clk_high_speed_in,
CLKDIV => clk_in,
D1 => serdes_input(0),
D2 => serdes_input(1),
D3 => serdes_input(2),
D4 => serdes_input(3),
D5 => serdes_input(4),
D6 => serdes_input(5),
D7 => serdes_input(6),
D8 => serdes_input(7),
OCE => '1',
RST => rst_in,
SHIFTIN1 => '0',
SHIFTIN2 => '0',
T1 => '0',
T2 => '0',
T3 => '0',
T4 => '0',
TBYTEIN => '0',
TCE => '0'
);
finished_out <= NOT running;
PROCESS(clk_in)
BEGIN
IF RISING_EDGE(clk_in) THEN
edge_in_old <= edge_in;
serdes_input <= ('0','0','0','0','0','0','0','0');
IF rst_in = '1' THEN
running <= '0';
ELSIF running = '1' AND chunk_counter * 8 >= 2 * length_act * count_act + delay_act THEN
running <= '0';
ELSIF edge_in_old = '0' AND edge_in = '1' AND running = '0' THEN
running <= '1';
length_act <= length_in;
count_act <= count_in;
delay_act <= delay_in MOD 8;
chunk_counter <= -delay_act / 8;
estimator <= 0;
CASE length_in IS
WHEN 1 => estimator_increase <= 4;
WHEN 2|3 => estimator_increase <= 2;
WHEN OTHERS => estimator_increase <= 1;
END CASE;
ELSE
IF estimator * 2 * length_act + delay_act < 8 * chunk_counter THEN
estimator <= estimator + estimator_increase;
END IF;
chunk_counter <= chunk_counter + 1;
FOR i IN 0 TO 7 LOOP
IF chunk_counter * 8 + i < 2*length_act*count_act + delay_act AND (
--we should only have 4 here but somewhere is a "one-off"-error and I just cannot find it right now
--the problem is that the estimator is being counted up one to early but I do not see why this is.
--test with delay_act = 7 and delay_act = 0 to see this error
((2*estimator - 2)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator - 2 + 1)*length_act + delay_act) OR
((2*estimator + 0)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator + 0 + 1)*length_act + delay_act) OR
((2*estimator + 2)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator + 2 + 1)*length_act + delay_act) OR
((2*estimator + 4)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator + 4 + 1)*length_act + delay_act) OR
((2*estimator + 6)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator + 6 + 1)*length_act + delay_act)) THEN
serdes_input(i) <= running;
END IF;
--alternative version but did not generate optimized code
--variable tmp: std_logic;
--tmp:= '0';
--FOR j IN -1 TO 3 LOOP
-- tmp:= tmp and '1' IF ((2*estimator + 2*j)*length_act + delay_act <= chunk_counter * 8 + i AND chunk_counter * 8 + i < (2*estimator + 2*j + 1)*length_act + delay_act) ELSE '0';
--END LOOP;
--IF tmp = '1' THEN
-- serdes_input(i) <= running;
--END if;
END LOOP;
END IF;
END IF;
END PROCESS;
END main;