Autoranging digital capacitance meter

In 1999 I was thinking of building a capacitance tester. Looking through a pile of electronics magazines, I stumbled across a design in the 1981 Popular Electronics' electronics experimenter's handbook. It uses TTL IC's, of which I had a lot lying around. The meter contains 22 integrated circuits, among which:

• 74LS00 and 7400 as clock oscillator and program cycle logic.
• 7490 counters.
• NE555 for the measurement circuit and cycle generator
• 74LS74 as measurement cycle state machine and clock divider
• 74LS251 as clock speed selector
• 7442 decoders on the display board.

The actual meter circuit is a 555 free-running RC oscillator. It charges and discharges the capacitor under test. The measuring cycle control circuit waits for the charge cycle to start, counts the clock cycles during charge, holds the display for a second, resets the counters and starts again.

The counter has 6 decimal counters but only displays the 3 most significant non-zero digits. This results in a 4-decade autoranging function. The design has a clever trick to achieve this. It starts counting with the clock signal directly connected to the display counter. When the counter overflows, a decade counter is inserted to divide the clock signal, the display is set to “100” and the counting continues 10 times slower. This can be repeated until a division factor of 1000 is achieved. This sounds complicated but it requires less circuits and wiring than the most obvious way, to use 4 multiplexer IC's to select between 4 counter IC's. Because even the NE555 timer has limits, the autoranging is limited to 1:1000. Therefore, the meter has a range select switch, that changes the value of the charging resistor in the NE555 timer circuit. This results in two main ranges:

• nF, full scale ranging from 999 pF to 999nF.
• μF, full scale from 0.999 μF to 999 μF.

The display has an extended range to 4000 or even 8000. This is useful in the μF range, to measure large electrolytics with reduced accuracy. This is achieved by the addition of two overrange LED's, that serve as a binary coded extra half digit. Because the counter needs seconds to count to 4000, you can even watch the overflow LEDs go on and off and count the number of times the overflow occurs. This way, you can measure up to 5000, 6000, 7000 μF and further.

The original design used a 1.4 MHz RC-oscillator for its counter clock. Instead, I used a crystal oscillator with a cheap 4.43MHz color TV crystal and a divide-by-three circuit using two D-flipflops from a 74LS74, resulting in a 1.48 MHz clock signal. That spared me the fuss to select a suitable inductor and results in a better long-term stability.

I wanted to build this meter using components I had lying around. For the display, I used a 3-digit counter board with 7-segment LED displays, SN7490 counters and SN7442 decoders. For the autoranging and control circuits I used TTL and LS-TTL IC's I had at hand. For the monostables I used F9601 IC's. To drive the decimal points, I used a 7444 excess-3-Gray decoder instead of a 7400. This IC came with a package of surplus TTL IC's I bought in a sale at a local Tandy shop long ago. I couldn't think of any other use of that IC. It was quite fun to tweak the design to accomodate these odd IC's.

All components apart from the display board were soldered to a piece of perfboard.

I used an aluminium instrument case that had housed a digital clock and a power supply. It has some useless holes in the front and back panels, which I tried to camouflage a bit. The result is a bit odd-looking instrument. But functional.

Repair

After 25 years of good service, late 2025, while testing the capacitors of the Bush TR82C, the meter suddenly started giving a zero read-out. First I thought the measurement circuit could have been damaged.

I looked up my notes from 25 years before and a copy of the original article. I had to find out where to start. Checking the measurement circuit with my scope, I saw it was cycling at a high frequency when no capacitor was connected to the input. The frequency was even higher on the μF range (because of the lower resistance in the circuit). I checked the clock signals and got the impression that the clock signal was not coming through. I checked the cycles of the control circuit and got the impression that one of the 74LS74's was stuck. But using the single sweep function of the scope, I saw that the measurement cycle was complete, but very short. I also noticed that the reset input of one of the D-flipflops was floating somewhere between at 1.2V, which is odd. In the original design, this input was left floating and it was marked as a test point. But there was no test described that used it. And while it usually is ok to leave a TTL input floating, leaving it to rise to logical "1" by itself, this is not best practice as it makes the input sensitive to interference. So it should be floating at 3-4 V, but it wasn't. I decided to tie the reset input to +5V (the little red wire) and test the meter again.

It was working fine now. I recallibrated it and took the occasion to make some pictures and write this page.