change initial value of state machine

[Matt Clement]

Hello

I have a design that has a state machine with 39 states. I would like to
start the state machine at state 39 rather than the left most state in the
type list. I have seen this done online and in many tutorials however it
doesnt simulate correctly in Quartus II. Is this the proper way to define
the initial state to start on power up?

.....more vhdl.

ARCHITECTURE ONE OF HALF_CLONE IS
TYPE STATE_TYPE IS
(IDLE,S0,S1,S2,S3,S4,S5,S6,S7,S8,S9,S10,S11,S12,S13,S14,S15,S16,S17,S18,S19,S20,S21,S22,S23,S24,S25,S26,S27,S28,S29,S30,S31,S32,S33,S34,S35,S36,S37,S38,S39);
SIGNAL STATE1: STATE_TYPE;
SIGNAL STATE: STATE_TYPE := S39;
SIGNAL TEMP : STD_LOGIC_VECTOR(31 DOWNTO 0);

......more vhdl.

Thanks for any help.
Matt

 

[Matt Clement]

oops. guess I should actually count the # of states. Obviously the design
snippet shows more than 39 states.
question is still relevent.
Matt

 

[ghelbig]

The proper way to set the initial state is with a reset. Quartus _may_
be getting confused about initial and default states.

Like this:

process(RESET, CLOCK)
begin
if (RESET = '1') then
STATE <= S39;
elsif (CLOCK'event and CLOCK = '1') then
case STATE is
when IDLE =>
if somecondition then
STATE <= S0;

... more code here ...

 

[kclo4]

Matt Clement a écrit :

Hello

I have a design that has a state machine with 39 states. I would like to
start the state machine at state 39 rather than the left most state in the
type list. I have seen this done online and in many tutorials however it
doesnt simulate correctly in Quartus II. Is this the proper way to define
the initial state to start on power up?

....more vhdl.

ARCHITECTURE ONE OF HALF_CLONE IS
TYPE STATE_TYPE IS
(IDLE,S0,S1,S2,S3,S4,S5,S6,S7,S8,S9,S10,S11,S12,S13,S14,S15,S16,S17,S18,S19,S20,S21,S22,S23,S24,S25,S26,S27,S28,S29,S30,S31,S32,S33,S34,S35,S36,S37,S38,S39);
SIGNAL STATE1: STATE_TYPE;
SIGNAL STATE: STATE_TYPE := S39;
SIGNAL TEMP : STD_LOGIC_VECTOR(31 DOWNTO 0);

.....more vhdl.

Thanks for any help.
Matt

It will works ... in simulation but never on board because the first
initialisation :=S39 of a variable or a signal is just use by the sim
tool but not synthetizer.
You should define your initial state by a reset and I personnaly always
add a case others when I write a state machine

ex:

> ....more vhdl.
>
> ARCHITECTURE ONE OF HALF_CLONE IS
> TYPE STATE_TYPE IS
> (IDLE,S0,S1,S2,S3,S4,S5,S6,S7,S8,S9,S10,S11,S12,S13,S14,S15,S16,S17,S18,S19,S20,S21,S22,S23,S24,S25,S26,S27,S28,S29,S30,S31,S32,S33,S34,S35,S36,S37,S38,S39);
> SIGNAL STATE1: STATE_TYPE;
> SIGNAL STATE: STATE_TYPE := S39;
> SIGNAL TEMP : STD_LOGIC_VECTOR(31 DOWNTO 0);
>
> .....more vhdl.

process(clk,rst)
begin
if rst = '1'then
state <= S39;
elsif rising_edge(clk) then
case state is
when IDLE =>
....
when S0 =>
....
when S39 =>
....
when others =>
state <= S39;
end case;


end if;
end process;

like this you should avoid any trouble


alexis

 

[Matt Clement]

Here is my delima
The board is already in production so I am trying to create a firmware
"update" that will allow my board to start up in a state that is different
than the left most in the list. I cannot add additional hardware on board
without a redesign of the board itself. I was trying to correct the problem
purely in VHDL. I also cannot change the way the pins are mapped without
rendering the board useless. I figured if I was able to start it in an
unused state then maybe I could do my initialization in that state and never
return to that state once started. Is there any other options/approaches to
this constrained problem? I need to set a signal to ones on initial power
up of the board and then start the normal operations where the signal can be
manipulated.

Thanks
Matt

 

[Mike Treseler]

The board is already in production so I am trying to create a firmware
"update" that will allow my board to start up in a state that is different
than the left most in the list.

Add a register to the design that
counts up to N and stops.
Create an internal reset pulse by
decoding one of these counts.

If you are lucky, the device
will start the count at zero for you
without a reset.

Good luck.

-- Mike Treseler

 

[ghelbig]

You mentioned Quartus-II, so I'm guessing you're using "Brand-A" parts.

I'm also guessing that you don't have a system reset handy, which is
causing you greif.

Most "Brand-A" parts have a concept of a power-on reset. And IIRC, the
synthesizer selects the first enumerated value of the state in the
declaration as the POR value. I also seem to remember that the
initalizer is only honored for signals that do _not_ change.

In your case, it _might_ be as simple as reordering your states as
(S39, IDLE, S0, S1, etc.) Or more legibly, (POR, IDLE, S0, S1, ....
S39)

And I also think that you're dicovering all kinds of new adjectives to
describe a synchronous circuit that's built without a reset.

GH

 

[Matt Clement]

Indeed GH you are correct with the colorful phrases. I also have changed
,prior to your reply, the order to list S39 as the left most state and then
I do my reset in that state, but it requires me to wait until the first
clock cycle before it initializes. I would prefer to have it initialize
without waiting for a clock input. If I put a sequencial statement before
the process(FSM) will it only execute the first time it runs or each time?
It seems that maybe I can initialize them before the process and should only
return to the process on each event??

Thanks
Matt