Testing and finding the error in my design (THINK it's in th

[Preben Holm]

Hi

I'm building the digital part of an oscilloscope (as already written in
earlier posts).
For this I have a trigger, a memory management unit and a presample unit
(basically a ringbuffer).

I've tested the ringbuffer by setting a slow varying signal
(square-pulse) on the A/D converter and then writing out the output to 8
diodes. Changing the pulse to fx. triangular I see the signal change on
the diodes a lot later (can't simulate the block memory in modelsim).
This thing really works (doing a lot of stuff)

But when putting this part together with the mmu-unit as this:
-----
memoryunit : mmut
port map ( clk => clk,
reset => reset,
prescale => prescale,
trig => trig,
read => read,
trigged => trigged,
holdoff_select => holdoff_select,
saved => saved,
rclk => rclk,
radr => radr,
din => sample_delayed,
dout => dout );


presampleunit : presampler
port map ( clk => clk,
reset => reset,
prescale => prescale,
pretrigger => pretrigger_select,
din => din,
dout => sample_delayed);
-----

I get some real strange trigger-positions that aren't the same for each
trigger, and a little mis-placed signal somewhere in the signal (but
else the sampled signal is just fine actually) - like


/\ /\ /\
/ \ / \ / \
/ \ / / \
/ \ / / / \
/ \ / / \ / \
/ \ / \ / \
/ \/ \/ \



It's strange... My first idea was that the signal wasn't saved correctly
and just kept saving the signal over and over the same memory while
reading a lot faster the little part of the signal was a new part that
was being written, but I guess that this would look more like this:

/\ /\
/ \ / \
/ \ / \
/ \ / / \
/ \ / \ / \
/ \ / \ / \ /
/ \/ \/ \/


So this error is probably not the correct one.

So I decided to try deactivating the ringbuffer (presampler module)
so my code looked like this:
-----
memoryunit : mmut
port map ( clk => clk,
reset => reset,
prescale => prescale,
trig => trig,
read => read,
trigged => trigged,
holdoff_select => holdoff_select,
saved => saved,
rclk => rclk,
radr => radr,
din => din, ------ THIS IS THE EDITING LINE
dout => dout );


presampleunit : presampler
port map ( clk => clk,
reset => reset,
prescale => prescale,
pretrigger => pretrigger_select,
din => din,
dout => sample_delayed);
-----

So now it works fine - but no prescaler (the trigger is very fine cut on
the right level, so this is very nice)

I get the trigger the correct place and no strange signals in the
transferred signal:

/\ /\ /\
/ \ / \ / \
/ \ / \ / \
/ \ / \ / \
/ \ / \ / \
/ \ / \ / \
/ \/ \/ \

So now I ask, where is the error, and where should I be looking.
I will publish all of my project code to you wise guys if this is
necessary, but here is the code of the presampler/ringbuffer:

-----
entity presampler is
Port ( clk : in std_logic;
reset : in std_logic;
prescale : in std_logic;
pretrigger : in std_logic_vector(3 downto 0);
din : in std_logic_vector(7 downto 0);
dout : out std_logic_vector(7 downto 0));
end presampler;


architecture Behavioral of presampler is
signal addra : std_logic_vector(10 downto 0);
signal addrb : std_logic_vector(10 downto 0);
signal pretrigger_i : std_logic_vector(11 downto 0);

component counter11bit
Port ( clk, ce, reset : in std_logic;
preset : in std_logic_vector(11 downto 0);
c : out std_logic;
q : out std_logic_vector(10 downto 0));
end component;

component fifo_mem
port ( ADDRA : in std_logic_vector (10 downto 0);
ADDRB : in std_logic_vector (10 downto 0);
CLKA : in std_logic;
CLKB : in std_logic;
DIA : in std_logic_vector (7 downto 0);
ENA : in std_logic;
ENB : in std_logic;
SSRA : in std_logic;
SSRB : in std_logic;
WEA : in std_logic;
DOB : out std_logic_vector (7 downto 0));
end component;

begin

-- every time there is a 'sample' signal, the counters should count
-- and save it into the "fifo".

presamplevalue : process(clk)
begin
if rising_edge(clk) then
case pretrigger is
when "0000" => pretrigger_i <= "000000001010"; -- 10
when "0001" => pretrigger_i <= "000000010100"; -- 20
when "0010" => pretrigger_i <= "000000011110"; -- 30
when "0011" => pretrigger_i <= "000000101000"; -- 40
when "0100" => pretrigger_i <= "000000110010"; -- 50
when "0101" => pretrigger_i <= "000000111100"; -- 60
when "0110" => pretrigger_i <= "000001000110"; -- 70
when "0111" => pretrigger_i <= "000001010000"; -- 80
when "1000" => pretrigger_i <= "000001011010"; -- 90
when "1001" => pretrigger_i <= "000001100100"; -- 100
when "1010" => pretrigger_i <= "000011001000"; -- 200
when "1011" => pretrigger_i <= "000100101100"; -- 300
when "1100" => pretrigger_i <= "000110010000"; -- 400
when "1101" => pretrigger_i <= "000111110100"; -- 500
when "1110" => pretrigger_i <= "001001011000"; -- 600
when "1111" => pretrigger_i <= "001010111100"; -- 700
when others => pretrigger_i <= "000000001010"; -- 10
end case;
end if;
end process;


-- Counters for addressing memory
addressB : counter11bit
port map ( clk => clk,
ce => prescale,
reset => reset,
preset => "000000000000",
c => open,
q => addrb);

addra <= addrb + pretrigger_i;


fifomemory : fifo_mem
port map ( ADDRA => addra,
ADDRB => addrb,
CLKA => clk,
CLKB => clk,
DIA => din,
ENA => '1',
ENB => '1',
SSRA => reset,
SSRB => reset,
WEA => prescale,
DOB => dout);

end Behavioral;
-----

The counter module is just a simple counter for addressing the memory:

-----
process(clk, reset)
begin
if rising_edge(clk) then
if (reset = '1') then
count <= preset;
else
if (ce = '1') then
count <= count + 1;
if (count(11) = '1') then -- this is the old count-value
count <= preset;
end if;
end if;
end if;
end if;
end process;

q <= count(10 downto 0);
c <= count(11);
-----


The code of the fifo_mem is quite long (but actually it's just a
ramb_s9_s9) with some ports left open!



Hope you guys/girls out there can help me!
It's really beginning to anoy me now


And like always: Thanks for helping me.

Preben Holm

 

[Jim George]

the diodes a lot later (can't simulate the block memory in modelsim).

Why not? If you are getting library errors, one common mistake is to do
an IP update of ISE without updating the Modelsim precompiled libraries
(if you're using MXE) or recompiling the libraries from source (if
you're using PE or SE). This does not cause problems when simulating
synthesized code, only when your code has instantiated components from
Unisim or Simprim.

I couldn't find anything wrong with your code, but I didn't do an
in-depth search.

 

[Preben Holm]

Jim George wrote:

the diodes a lot later (can't simulate the block memory in modelsim).


Why not? If you are getting library errors, one common mistake is to do
an IP update of ISE without updating the Modelsim precompiled libraries
(if you're using MXE) or recompiling the libraries from source (if
you're using PE or SE). This does not cause problems when simulating
synthesized code, only when your code has instantiated components from
Unisim or Simprim.

I couldn't find anything wrong with your code, but I didn't do an
in-depth search.

I may have found out, when making the PROM-file the tool doesn't
actually overwrite the file!!!
I have it working now, after removing the old PROM-file!
I wrote the prom all times in the last couple of times since I found it
annoying not being able to reset the circuit.


Thanks for looking through my code anyway!