Page 3. Femto Capacitance Meter
Printed Circuit Boards.

There are two PC boards attached to the front panel - a large one above the switches, and a small one below them.

With a 4,000 count LCD (using 3701) there is an additional large board for the display and driving chips (4543s). The XOR oscillator switch is also on that board. The board is stacked on the same screws as the other large board, with 5/16" (8mm) spacers between them. In this case only, the display is on the component side of the board. It is socketed into molex pins.

The board is shown in Figure 13. The decimal point wires go through the board near the rotary switch. They need to be color coded 30 gage kynars.

Quite a few jumper wires will be needed, as usual with complex digital boards. If you use dry transfers, you might get all of the bus lines on; but if a marking pen is used, jumper wires will probably be needed for some of them. There are a few short jumper wires under the display; and those wires have to stay on the copper side of the board; so they are soldered on last.

Eleven wires link to the other board through a header; but there is no header on the display board. The wires are attached directly to the board near the corner. The wires then extend 17cm to a male header. I use machined pin sockets for male and female headers, because they save space and are readily available. When soldering wires into the top of the socket, a low temperature iron is used along with liquid rosin for fast soldering, so as to not melt the plastic. However, there should be enough space available for other types of headers.

The wires to the header include 4 chip selects (to 1 pins of 4543s), 4 bits of binary code (A, B, C, & D), The overflow signal to blanking pins (7 pins), +5V and Ground.
Figure 14. shows the main board when using a 3701. The comparator should have a bypass capacitor, 1-10F, between pins 4 and 8.

If two range setting resistors are used in series, the op-amp can be slid back further. The resistors stand vertical to save space. Both leads are attached to the board for stability. The long lead is in front for attaching the wire to the rotary switch.

The LED display is on the copper side of the board. Therefore, the pins are attached by wires only, as if a perf board were being used. A socket is not used.

Seven resistors are located above the pins. They stand vertical. Thirty gage wires go from resistor leads to pins. Being so short, they can cross without insulation.

Since the LED produces heavy digital currents, a capacitor of 1,000 F must be located near the A/D, between pins 1 and 25, even if the voltage is regulated with an LM340. Otherwise, digital noise causes the A/D to produce errors.

On the LED board, the 3701 is moved close to the edge to save space; and therefore, most of the digital connections require jumper wires, which are about as easy to use as drawing lines anyway.
Figure 15. shows the small board and switches. The wires going over the boards are 30 gage kynar. They are soldered to wire loops which have both ends in the board, so soldering from the top does not cause them to unsolder. Tinned bus is preferable for loops, but not essential. Small quick-connect pins might be located. if experimenting with components is desired, use loops for soldering them on from the top. This procedure is suggested for the small capacitor with the reference capacitor, because you may decide later to improve the accuracy based upon measurements of reference capacitors.

Where supply wires jump onto the board, a couple of 10F capacitors or larger are used as a matter of principle.

The rotary switch can have resistors coming off the pins, if space saving is needed.
Power Supply.  

Figure 16. shows the power supply. For an LCD meter, it would not have to be that rugged.

For an LED meter, the 5V regulator needs to be heat sank to the bottom of the box. For that, an aluminum bracket is made - bent twice creating a U shape. The chip then sits above the edge of the board.

The box is of course grounded at some point. The ac power line has a 10M resistor going to ground for dissipating static charge.

A shielded transformer is preferred.
Insertion Pins.

The insertion pins used on the prototypes were adapter pins designed for breadboards. They are no longer available, but a new type is produced by Mill-Max and sold by Digi-Key (near IC sockets in catalog). A suitable size is 0.025-0.037", which is Fig. 16, Digi-Key part number ED5009. The very large capacitors will not insert, but they can be held in place while measuring. Holding large caps does not alter their measurement significantly.

The next size up can be used. It is Fig. 24, ED5013. Small leads will be loose, but they can be bowed inward slightly, which locks them in place.  (return to text)

Construction Hints.
Figure 17. shows the use of a 1.985 reference voltage for increasing the resolution of calibration measurements.


To measure capacitor leakage, as shown in Figure 18., apply a voltage, then allow it to sit 30-60 seconds and remeasure the voltage using a FET op-amp as a buffer. To measure the reverse current of a diode, put it in parallel with the capacitor.


To measure the offset of a comparator, increase the voltage on one of the inputs through a resistor divider, while watching for the change of state. Note the voltage at which change of state occurs. If the polarity is incorrect, reverse inputs.


Solid wire solder if far better than core type. Core type puts rosin everyplace except where it belongs, causes carbon to coat the iron, and can destroy sensitive parts before getting a connection. Liquid rosin can be put directly on the connection, which speeds soldering and keeps it off the iron. Then the solder is picked up on the tip of the iron, which frees one hand.

But better than rosin is citric acid disolved in distilled water. It won't create a burnt carbon coating, and it washes off easy with water. It will conduct a small amount of current if not removed.

The copper on the board should be coated with solder after etching, which speeds soldering as well as preserves the surface. When the copper surface is cleaned with acetone, citric acid will coat it more uniformly.

Apply liquid citric acid with a toothpick, except on a large surface, where cotton swabs work good. Use citric acid saturated, which means there should be some undisolved cristals on the bottom. Citric acid is available at pharmacies.


For an LED display, a lense is needed to show contrast. Plastic windshield tinting will stick on with water in place of a lens.


When making the box, do not drill the holes in the tabs until the bottom is drilled. Marking through the bottom will align them properly for sheet metal screws.


One reason why the Femto Capacitance Meter is needed is because small capacitors could not previously be measured adequately. For example, fractional picofarad capacitors might be manufactured with high precision, but they would all be off by about 20-50%, because the manufacturer could not measure them any closer than that. On top of that, markings are often nonexistent or impossible to read on small capacitors. So they were a real problem.

Test leads could be used for everything above 100 pF, but I didn't include them, because they are a real nuisance. They sweep things around on a workbench, and they tangle when stored.

Adapter pins (for breadboards) are no longer available for insertion of the test capacitors. But there are a variety of styles of pins available. Buy different types and test them. Another alternative would be to use tinned can steel to construct insertion pins. Roll it wide at the top for large leads and narrow at the bottom. Then solder it to the copper side of the board. It can be adjusted after attaching by pinching with side cutters or prying open. If small leads do not make good contact, they can be bowed inward slightly to create a suitable stress force.

I designed for a lot of external adjustments. Most of them could probably be eliminated with no more than 1% error. But they are desirable for high precision and certainty. If for example, a lot of ripple were placed on the peak detector storage capacitor, a pull down switch would not be needed. However, the pull down (called set) is so fast and convenient that I see no reason to exclude it.

Zero adjust does not measure on the negative side; all negative values read zero. Therefore, zero is adjusted to 1, which can be ignored or subtracted from the read.

Some persons would want a capacitance meter to look like the ones in the stores - battery powered, pocket size, rectangular, test leads, measuring inductance. Those characteristics are not conducive to the precision that is sometimes preferred, and they are not ideal characteristics. A meter needs a sloping front face for easy reading. Infact, I made an aluminum bracket for a voltmeter for sloping it. And why battery power for something that is not carried around. Capacitors cannot be measured in-circuit, so a capacitance meter is not carried around.

Creating a digital display is quite a bit of work. If you need the meter but don't have lot of time, the thing to do would be to use a voltmeter for readout. But the sloping box would still be desirable for controls.

to page one