0
\$\begingroup\$

clock signal generator

I have an old PCB that uses this clock generator (means I'm absolutely sticking to the schema). The IC is standard 7400, resistors are the same, but the capacitors I'm using have 470 uF (500's are hard to come by). The clock drives a traffic lights simulator, hence the period is 1~2 seconds.

When I increase the the variable resistor to about 75%, the clock stops. O-scope shows me both the capacitors are discharging more slowly, and the voltage on the gate inputs gets stuck in the undefined region.

That brings me to these questions:

  1. Why this keeps happening?
  2. What's the relationship (equation) between the R and C in each "branch"? Should I increase the C, or lower the R, or what exactly?
  3. how are the capacitor & resistor values calculated = Timing / voltage / current requirements, considering the IC used?
  4. Anything else needs to be mentioned? (School me please)
\$\endgroup\$
9
  • 2
    \$\begingroup\$ good news: nowhen in history have capacitors ever been as accurate as they are today. Chances are the 500 µF capacitor in your circuit diagram had a \$\pm\$30% tolerance – meaning your 470 µF is more likely to be acceptably close to what this circuit was designed to operate with. \$\endgroup\$ Nov 4, 2018 at 19:32
  • 2
    \$\begingroup\$ Are you certain it is \$ 7400 \$ and not perhaps \$ 74LS00 \$ ? \$\endgroup\$
    – glen_geek
    Nov 4, 2018 at 19:34
  • \$\begingroup\$ @glen_geek I am, unless the markings are wrong. But what difference would using another family have? \$\endgroup\$ Nov 4, 2018 at 19:45
  • \$\begingroup\$ @PavolHanzel if the NANDs are faster or have tighter hysteresis than those of the ancient days of lore when people drew the schematic you have, then this circuit simply won't work. Your circuit only works because old logic ICs were "bad". \$\endgroup\$ Nov 4, 2018 at 19:46
  • 3
    \$\begingroup\$ Tell you what. Just as an experiment, try tieing one input of each gate to +5 instead of to the other. This will reduce the input current requirements by a factor of two, and might just give you the result you want. \$\endgroup\$ Nov 4, 2018 at 22:00

1 Answer 1

1
\$\begingroup\$

What you see here is a very straight-forward oscillator:

your NAND gates are just wired up to be inverters.

If you just take one inverter and connect its input to its output, it should theoretically start to oscillate infinitely fast.

If you think about it, an inverter is just an amplifier with an exact phase shift of 180° and a high amplification.

In practice, its speed is limited, so that won't work reliable.

So, what you do instead is have two inverters in series, giving you a phase shift of 360° \$\pm\$something; since you can't tell 360° phase from zero phase – that's basically stable (if nothing excites them).

Then, you add phase shift – ideally, something that has a phase shift of 180° at but a single frequency.

In this case, two high pass filters were built. High-pass filters have a frequency-dependent phase shift.

Your high-pass filters are RC filters:

  • C1ıR13
  • C2ı(R14+P)

due to P being adjustable, you can tune the frequency at which the oscillation stabilizes.

Lord knows the actual phase of these inverters – 1970's logic wasn't really fast or precise in analog terms. So, a theoretical analysis of this circuit will not yield useful data.

Since, with good chance, you don't have access to the same batch of suboptimal logic ICs that were used in the original design, but probably to logic ICs that are far better and faster, as soon as you don't sufficiently dampen that C2 filter, things stop to work.

Long story short, this only works with bad hardware of the past. Don't stick to the original schema; build your own oscillator. It's way easy these days (or really, since the invention of the NE555, which was exactly meant to solve this kind of problem. You are literally the first, and very likely the last person I meet for whom using a 555 would be a technical upgrade).

\$\endgroup\$
9
  • 1
    \$\begingroup\$ Instead of the NE555, a single 74HC14 inverter with one resistor between input and output and one capacitor between input and ground would also do the trick. Because that one has a hysteresis. \$\endgroup\$
    – Janka
    Nov 4, 2018 at 20:39
  • \$\begingroup\$ well, but you then would still be fully dependent on the current abilities of a logic IC (feeding the low-pass RC filter that you describe, whose characteristics depend on the voltage being stable...); that feels unfortunate, seeing that you actually want a stable oscillator rather than a proof of concept that you can implement some oscillator with but an inverter. \$\endgroup\$ Nov 4, 2018 at 20:51
  • 3
    \$\begingroup\$ Whenever I read 500µF, I know stability of the time constant is a non-issue. \$\endgroup\$
    – Janka
    Nov 4, 2018 at 21:08
  • \$\begingroup\$ oh, well. gotta point there. \$\endgroup\$ Nov 4, 2018 at 21:14
  • 1
    \$\begingroup\$ I would prefer the 74HCU04 or the 74HCU00 parts for an oscillator. Unbuffered. The propagation and transition times are balanced. \$\endgroup\$
    – jonk
    Nov 4, 2018 at 21:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.