Why is this clock signal connected to a capacitor to gnd?

I am trying to understand the following circuit:

My problem is to understand why the CLK signal is connected to the capacitor (C7). The bottom side of C7 is connected with a resistor to GND. This means that the "plate" will (after some delay) have the same potential as GND. The upper "plate" will oscillate with the CLK signal. The logic is connected to the bottom side, so I don't understand how the NAND Gate can ever get some other input than LOW on the bottom pin. This means that the NAND gate will never output LOW and thus the RAM never stores anything, which doesn't make sense.

The only effect the capacitor could have in my understanding is to delay and flatten changes in the clock signal, but I don't understand the use of this.

Also I don't understand the meaning of the resistor. It pushes the loading time of the capacitor, but after the bottom side is on GND potential there will be no current after that.

I hope someone can help me understand this.

• This is a large schematic and is hard to read due to the scaling. Could you perhaps add an arrow or circle to show what capacitor you mean? Apr 3, 2019 at 20:13
• Can you show where the CLK signal originates? Apr 3, 2019 at 20:14
• @Hearth... Click on the schematic for a larger version.
– BobT
Apr 3, 2019 at 20:17
• Since the clock is oscillating, why do you believe the bottom plate of capacitor is at steady state? Apr 3, 2019 at 20:17
• Meh, I think it's an OK schematic. I've seen much worse. It's not "badly drawn" just because one part is suboptimal.
– pipe
Apr 4, 2019 at 8:30

C7 and R58 form a high pass filter, also known as a differentiator.

The purpose of using a differentiator in this spot is to cause a short pulse on the rising and falling edge of the clock signal.

This diagram shows the effect of a differentiator on a square wave (which your clock will be.)

As you can see, it makes short pulses on the edges of the square wave.

I'm not sure why that circuit needs the short pulses instead of the square wave, though. Too many ICs I'd have to look up to figure out what is going on.

• I think it's an RX edge latch Apr 3, 2019 at 22:48
• I also ask myself why this is needed. The multiplexer also allows a button push which won't be much faster then the unmodified clock signal. Apr 4, 2019 at 21:42

The circuit overall is a 16-byte memory bank for a homebrew computer of some sort, with manual programming capability via the switches and lights. Probably the most complicated 16 bytes of memory you'll ever see!

The R-C combination being asked about is used only when the CPU is running — i.e., when PROG is not asserted. It is probably being used to shorten the high time of the write-enable pulses going to the memory chips in order to meet hold-time requirements when the CPU is writing to memory. This only works if the clock high time is significantly longer than the R-C time constant.

10 nF × 1 kΩ = 10 µs

So presumably the clock is something less than 50 kHz.

Actually, the 74189 is not a slow part — the minimum write pulse width is a few tens of ns — so the R-C time constant could be much shorter, by a couple of orders of magnitude.

That is a high pass RC filter with a pole at 15.9kHz, which is unusual for a clock signal because it will attenuate the signal somewhat. The purpose is possibly to keep the clock more than 15-20kHz, to keep the memory running above that speed as a lower clock speed would not transition.

As others have noted, it's a differentiator that limits the time that WE is asserted.

The outputs from the 74189s go into high impedance with WE is asserted, so it could be to make sure that the LEDs are not excessively dimmed, or to ensure that the output BUS_* lines are stable before the other edge of the clock.

The latter option might be used of the output bus is read on the other clock edge.

For your information, if the resistor is not there then the NAND gate will never gets a logic high in PIN 1 and always remains low and the resistor also increase the rise time and fall time so the circuit will not even act as a low pass filter, so we cannot decode the intention of the designer.

• I just wanted to point out that this question is quite old and has an answer that is accepted and upvoted already. It is probably better to focus on newer questions that don't already have accepted answers. But thank you for trying to answer a part of the question that was not addressed by others. Feb 27, 2020 at 8:06