replacing the seven segment decoder/driver with a pal+latch

I am designing a circuit that will use the seven segment LED display for output.

I dont have the decoder driver chip and dont want to buy it.

I do have lots of 22v10' and 573/373' (I hate the latters pin out).

Can I decode on the 22v10 ... feed its output to a 573/373 and then to the seven segment LED display?

Would fan out be an issue?

I suppose I should just test it ... when I get home, but I have already started the PCB and nearly done ...

Absolutely!

YES YES I totally forgot (banging head against wall).

I am stuck in 16L* land. Originally I was going to do the decode driver in a 16L8 with 573/373 ...

Forgot to "resync" my thinking when I pickeup the 22v10's.

Good point thanks. I love this newsgroup :)

I assume this is a one-off project.

What's the 573/373 for? The 22V10 contains flip-flops.

Lattice GAL22V10 outputs can sink 16mA. That's enough for an LED segment.

I implemented this 'font' in my clock project (Moto 68HC908). Its actually quite readable.

My only comment about 22V10s as decoder/drivers is if you use 1 for each digit, they're kind of power hungry.

GG

Well, the newsgroup loves smart questions like this. As a matter of fact, you've got me thinking of doing a BCD-7-segment in some kind of PLD. I've also seen this pattern (or something very similar) for Sexydecimal ;-) _ _ _ _ _ _ _ _ _ _ _ _ | | | _| _| |_| |_ |_ | |_| |_| _| |_ | _| |_ |_ |_| | |_ _| | _| |_| | |_| _| |_| |_| |_ |_| |_ |

Cheers! Rich

Below is to be preferred: _ _ _ _ _ _ _ _ _ _ _ _ | | | _| _| |_| |_ |_ | |_| |_| |_| |_ | _| |_ |_ |_| | |_ _| | _| |_| | |_| _| | | |_| |_ |_| |_ |

Best regards, Spehro Pefhany

On Wed, 07 Dec 2005 22:55:58 -0500, Spehro Pefhany wrote: ...

OK, you're right, that is better than mine. :-)

(but I _did_ kind of make it up on the fly. :-) )

Thanks! Rich

...

...

Speaking of LUTs, what ever happened to the 8223, 32X8 PROM? They were handier than a rat for a universal 5-input, 8-output thing. Yes, that's thirty-two by eight - thirty-two eight-bit bytes. Maybe these days they're just using SOIC OTP 27512s or something?

Thanks, Rich

Too small to bother with. Even 64Kx16 is too small. Use a PLD of some sort.

Here's some simple VHDL code for a 7-segment decoder as above, with segments inverted to drive a common-anode display (written by David Warren-Smith). Note that VHDL isn't case-sensitive, so D==d, S==s, SEG7==seg7:

LIBRARY ieee; USE ieee.std_logic_1164.all;

-- Title "7 segment display driver circuit";

-- File: seg7.vhd

ENTITY seg7 IS PORT (D : IN STD_LOGIC_VECTOR (3 DOWNTO 0); -- BCD input S : OUT STD_LOGIC_VECTOR (6 DOWNTO 0)); -- 7 segment outputs END seg7;

ARCHITECTURE display OF SEG7 IS BEGIN s

In article , Spehro Pefhany wrote: [....]

I have found that it is better to make the names of ENTITYs and ARCHITECTUREs more obvious such as:

ENTITY SevenSegmentConnect IS and ARCHITECTURE SevenSegmentDisplay OF SevenSegmentConnect IS

It doesn't matter in a simple case like this but when you have a hundred or so of each, remembering what matches up with who can be troublesome.

The equivalent function, but I've found that a process using case/when often produces better logic:

LIBRARY ieee; USE ieee.std_logic_1164.all;

-- Title "7 segment display driver circuit";

-- File: seg7.vhd

ENTITY seg7 IS PORT (d : IN STD_LOGIC_VECTOR (3 DOWNTO 0); -- BCD input d : OUT STD_LOGIC_VECTOR (6 DOWNTO 0)); -- 7 segment END seg7;

ARCHITECTURE display OF SEG7 IS BEGIN

seg7: PROCESS (d) BEGIN CASE d IS WHEN "0000" => s s s s s s s s s s s s s s s s NULL; END CASE; END PROCESS; END display;

Synthesizers are template driven and I've seen if/elsif/elsif/... and when/else/when/else/... produce some terribly serial logic. At least with Synplify the case/when template is often a lot cleaner.

I find your method to be awkward when trying to follow or scope a signal through many levels of herarchy. The names get to be unmanageable.

Humm... I tried both ways, and the Xilinx tools created 4 slices/7 LUTs/3 levels of logic both ways... automatically extracting a 16 x 7 ROM in both cases. For what that's worth.

Thanks for the tip.

Best regards, Spehro Pefhany

In article , Keith wrote: [...]

I don't see how. How does what I've done effect the signal names? I assume you mean that the signal names get unmanageable.

In the tools I've used ( small set I admit ) the names in reports are built up with instance names not the type of the entity so it is the instance name that you want to keep shortish.