Light display controller

1. The gist

The board has eight small LEDs in a row. You are building a circuit that lights one of them and moves the lit one along the row, one step per second, so a dot appears to travel across the display.

Two slide switches, LEFT and RIGHT, control it. Their four combinations:

1. LEFT on. A dot travels from the right end of the row toward the left.
2. RIGHT on. A dot travels from the left end toward the right.
3. Both on. Two dots, one going each way. They pass through each other in the middle.
4. Both off. The LEDs go dark. The dots stay where they were, and when you turn a switch back on they carry on from that position.

A third switch is the reset. It puts both dots back at their starting ends and clears the display.

That is the whole behaviour. The circuit is small. The work is in how you are required to design it.

3. The kit

4. What is happening inside the circuit

Three storage registers, eight bits each.

• DISPLAY holds what the LEDs currently show.
• LMASK holds the position of the left-travelling dot.
• RMASK holds the position of the right-travelling dot.

The masks carry the dots. LMASK starts as 00000001, a single 1 at one end, and each step it moves that 1 one place along. RMASK starts as 10000000 and moves its 1 the other way.

DISPLAY copies whichever mask is active. With one switch on it takes that mask, with both on it takes the two combined, and with neither on it takes all zeros.

Two details decide whether your circuit works.

The dot has to come back. Move a single 1 along an eight bit register eight times and it falls off the end, leaving 00000000. The display would then stay dark for ever. So the registers rotate: the bit that leaves one end reappears at the other. The handout does not say this outright, so it is a decision you make and then write up.

Switching off must not erase the position. When both switches are off, DISPLAY is loaded with zeros, and LMASK and RMASK are left alone. A register with an enable input holds its value whenever the enable is low, so holding costs no extra hardware. You simply do not tell those two registers to load.

5. What the reset switch does

The reset forces every flip flop in the design to a known value: DISPLAY to all zeros, LMASK to 00000001, RMASK to 10000000. It is the only thing that ever puts the dots back to their starting positions. Turning the two travel switches off stops the dots and leaves them where they are.

Your demo starts by flipping reset on and then off again.

6. The rules that cost marks

• VHDL only. Verilog is refused.
• Structural style only. You build the circuit out of small named parts wired together. Describing the behaviour and letting the software work out the hardware is refused.
• Build every part yourself. No ready made blocks from the Vivado catalogue or from the internet. Flip flops, adders, comparators, all of them yours.
• The whole design is synchronous and resettable. Clock and reset reach every block.
• Slow the display down with an enable, and leave the clock alone. The board's clock runs at 100 million cycles a second, far too fast to watch. You count 100 million cycles and emit a single pulse, then let the registers move only on that pulse. Feeding a slowed signal into the clock inputs themselves is a known way to produce a circuit that behaves differently in simulation and on the board.
• Demonstrate it working to the TA before the report is due.

7. How to approach it

Session one, together, on paper. Work through stages 1 to 3. Do not split this up, and do not open Vivado yet. Everything after it depends on these being right, and a TA is in the room to check them. Start the Vivado download on your partner's machine before you begin, so it runs while you work.

Then agree the list of signals between the two halves of the design. Write down the exact name and width of every wire that crosses from the control logic into the datapath, for example a two bit selector that chooses what DISPLAY loads. This list is the agreement that lets two people write code separately and have the pieces fit together afterwards.

Then split the building. One person takes the datapath: the three registers, the selector, the combining logic. The other takes the counter that produces the one per second pulse, plus the control logic. Each writes a small test for their own parts.

Session two, together. Wire the halves together, simulate the whole thing, then synthesise it and program the board.

Then the report. Whoever built a part writes that part of the report, and one of you reads the whole thing at the end so it sounds like one writer.

8. What goes in the report

From the handout's own list:

• Simulation results, with an explanation, for every VHDL file
• Comments through all the source
• A flowchart of your solution
• A block diagram of your solution
• The ASM charts, the datapath and the control logic
• Your solution described in your own words
• The problems you hit and what you did about them
• All source code appended, and the files submitted alongside the report

The problems section is worth planning for. The rotation decision in section 3 belongs there, and so does anything that behaved differently on the board than in simulation.