Use of FPGAs to teach Digital Circuits I

[FrankBuss]

You won't get the benefit of probing the digital waveforms as they propagate through the circuit using an oscilloscope because your circuit is totally confined within the FPGA (also FPGAs don't actually build circuits using logic gates).

With Altera chips, you can use SignalTap, which can sample internal signals like a logic analyzer. Of course it is not the same as an oscilloscope, but very helpful for debugging more complex circuits and maybe for learning, too.

[MrB]

Our first Digital Circuits class was all theory and not the best IMHO. We learned, for example, what a multiplexer did, but not why you would like to use one, which is the most important part for a lot of students (first why, then how).

Now we're getting VHDL with a Xilinx board and boy do I like it. After 3 lessons we were building VGA drivers with ease. I think the reason most college's prefer FPGA's over breadboard logic is the speed of which new concepts can be introduced, without spending a lot of time wrapping wires. Being able to program FPGAs is also a valuable skill.

[zapta]

If you want to do logic experimentation using 74HCT or 74LS ICs, by all means, get yourself a few and experiment.

You can also do that with an FPGA. Write models for a few 74XX ICs and then connect them together to get whatever behavior you need. This will drive the point that 74XX's are just small arbitrary chunks of predefined logic.

[langwadt]

If you want to do logic experimentation using 74HCT or 74LS ICs, by all means, get yourself a few and experiment.

You can also do that with an FPGA. Write models for a few 74XX ICs and then connect them together to get whatever behavior you need. This will drive the point that 74XX's are just small arbitrary chunks of predefined logic.

FPGAs are a terrible match for 74xx. FPGAs are for clean synchronous logic with flipflops on preferably a single common clock, 74xx is a mess that is pretty much the opposite of that

[FrankBuss]

FPGAs are a terrible match for 74xx. FPGAs are for clean synchronous logic with flipflops on preferably a single common clock, 74xx is a mess that is pretty much the opposite of that

Right, that's how they are usually used. But you can still implement any 74xx chip (except the ones with analog functions, and with schmitt trigger inputs depends if you can enable this on the FPGA). Could be fun to implement a few of them in one FPGA devkit, with all inputs and outputs connected to pins, then wire it externally. This has the advantage that you can measure the intermediate signals with a real scope, too, and probably silly enough to be featured as a Hackaday article

[Dongulus]

You won't get the benefit of probing the digital waveforms as they propagate through the circuit using an oscilloscope because your circuit is totally confined within the FPGA (also FPGAs don't actually build circuits using logic gates).

With Altera chips, you can use SignalTap, which can sample internal signals like a logic analyzer. Of course it is not the same as an oscilloscope, but very helpful for debugging more complex circuits and maybe for learning, too.

The OP will probably learn to verify logic using simulation as opposed to SignalTap. In either case those tools only give a functional view but they don't really communicate propagation delays nor rise and fall times.

[tomizett]

I tend to agree with the contributors who've suggested that you might miss some of the tangability and real-life pitfalls that come from working with traditional logic ICs. Surely FPGA and discreet logic design are different but complementary skills.

My main question, though, is what will you learn using an FPGA that you couldn't learn purely in simulation? If the answer is "not much", then would the university not be better off letting each student work individually - rather than have to share 3 to a board - and be able to take their work home with them at the end of the day?

To add a little perspective to my comments, I generally prefer to design only with ICs that are older than me...

[FrankBuss]

My main question, though, is what will you learn using an FPGA that you couldn't learn purely in simulation? If the answer is "not much", then would the university not be better off letting each student work individually - rather than have to share 3 to a board - and be able to take their work home with them at the end of the day?

There are a lot of things you can only learn from the real hardware. For example if you have a clocked design and don't buffer some input push buttons, then anything can happen, like a statemachine entering illegal states. Very difficult to write a full simulation which simulates all this problems, and usually people use logic level simulations anyway for complex designs, because gate level simulations are sooo slow. You would never see such a problem in a logic level simulation. But for learning basic digital circuits, a simulator might be all you need, if you don't care about the details and real world problems (could be teached in a later course).

[LabSpokane]

To add a little perspective to my comments, I generally prefer to design only with ICs that are older than me...

Is your advice coming from a hobby development or a commercial development perspective?  The OP is paying a boatload of money for a EE degree. Even if he/she goes onto grad school, the FPGA work is a marketable skill. Where is the market for new designs using solely decades old components?  I'm genuinely curious what sector of electronics this would be.

[mathias]

FPGA's will help you develop your "theoretical" digital skills much faster, but your practical skills will suffer. If the course is well designed, the time gained by not using breadboards should be dedicated to practical problems.

People here have already mentioned quite a lot of them (groundbounce, race conditions, etc.). Though you might never encounter these anymore (e.g. your synthesizer takes care of race conditions, setup-and-hold, ...), you need to know that things differ from theory to practice. Else, you end up with engineers who design circuits, the circuit gets up prototyped, and eventually never works. This is because some issue has popped up, which went undiscovered in the design phase because one didn't study the design critically enough.

You could put 1000 monkeys writing HDL/spice, and at least one will write the circuits you want. But that doesn't make him an engineer.

[drakke]

One, solderless breadboards should be banned from all labs. 

Tack Solder the connections and the grimace that occurs with traditional labs goes away.

Can you explain how and what you use? I'm a beginner.

I just bought a breadboard to build circuits and I can understand if you think the connections may not be completely solid all the time.

[tggzzz]

One, solderless breadboards should be banned from all labs. 
Tack Solder the connections and the grimace that occurs with traditional labs goes away.

Can you explain how and what you use? I'm a beginner.
I just bought a breadboard to build circuits and I can understand if you think the connections may not be completely solid all the time.

That's the primary reason. A component with a thick lead can open the spring so that when you put a component with a thinner lead in it makes intermittent or a hicher resistance connection.

There are other reasons that are not a problem with audio circuits but can be a problem with digital circuits, since they have high-speed (= high frequency) edges. The capacitance between springs adds extra loading, and long wires add inductance, and the two together can resonate. Inductance and resonance are a particular problem with decoupling capacitors, but can also foul up oscilloscope measurements. (For your reference in the future, see https://entertaininghacks.wordpress.com/2015/04/23/scope-probe-accessory-improves-signal-fidelity/ )