20161116:

Just to be sure the kids understand, and to help one kid who was sick last time, let's redo the charge pump expermient. This time, with screenshots and more bugs.

Let's start with the oscillators : the classic 74HC04 provides 2 inverters, then we hook a resistor and a capacitor. By now, the kids should do this circuit without any difficulty :-)

What are their value ? I have set the goal to 10KHz. A previous workshop has shown that the frequency is about 1/2RC and we have C=100nF. Let's compute the resistance...

Some whiteboarding later, I come up with 500 Ohms. I can find 510 ohms in my stash of resistors so let's go. We get about 8KHz, that's -20% of the goal but pretty close, considering the other unknowns.

There's a problem though : of the 3 circuits that were assembled, 2 showed problems as they would not oscillate. The reason is not clear, bad capacitor ? Maybe a too low resistance ? (I don't have smaller capacitors at hand)

The behaviour is odd, as the scope shows a small burst of activity then the trace turns flat, mid-point between 0V and 3V.

So we focus on the only circuit that seems to work.

We connect a 1µF ceramic capacitor to the output and connect the other electrode to the scope probe.

The blue trace is initially a copy of the original yellow square wave but it quickly drifts (in about one second) toward -1.5V/+1/5V

This is explained by the capacitance of the scope probe: 1µF × 1M Ohms = 1second

Cue in discussion about low-pass, high-pass, time constants and what if we set the probe impedance to the 10× mode :-)

Then let's introduce a diode : current passes in one direction but not another. Tthe diode is connected to the +3V rail and the passing direction is outward the rail.

The trace has shifted upwards, the capacitor's electrode is at a higher mean voltage than the power supply.

Now, let's see with the diode connected to 0V:

The waveform has shifted below 0V, but there is still a part that reaches 0.7V, which is the "drop" of the diode. The output voltage now reaches -3V +.7= -2.3V (this was not apparent in the previous ss because of a bad setting of the 0V level)

Now, let's add the 2nd diode and a large capacitor. Oooops, it seems to overload the 74HC04 which stops working, so let's use only 1µF.

We measure the generated voltage : with a step-up configuration and 1µF, that's 4.74V. When adding a 150K resistor load in parallel with the filter capacitor, that's 4.54V, or a 200mV drop. From there we can calculate the impedance of the generator and estimate the drop with other loads (approx 1mV/Kohms, or 1M Ohms of impedance...)

It's not efficient, but advantageous to generate very high voltage when cascaded.

Discussions about diodes lead to other applications, in particular the antique keyboards, and diode matrix ROM (oh, wait, a common subject these days :-D)

From there, how does one creates a diode ROM ? It's a matter of decoding. What can it be used for ? Let's say, a 7-segments decoder :-)

How does one decode the ROM ? it's another matrix, but with AND gates this time. Very easy to design on Minecraft, once you know how to copy-paste an element :-D

Back to the proto boards and actual (N)AND gates. The simple N-MOS inverter is shown again, and a second transistor is added to prevent it from conducting current to 0V : that's how we make a NAND.

If the transistors are in parallel, it's a NOR.

Considerations about complementary MOS circuits are discussed : less power draw, faster, but at this time we're happy with a pull-up resistor. We measure the rise time (when the button is released) at about 400ns : that's slow but we don't need to run at Megahertz speeds...