DIY Capacitor Reference Box

 

 

This is a DIY capacitor reference box that can be used to verify and compare multimeters. It's not a metrology grade reference, but has capacitors with better tolerances than what is typically available in electronic circuits. It's still relatively cheap.

For better results, a calibrated instrument should be used to check capacitor values.

The reference box contains capacitors in the range from 10 pF to 10 mF. 

It should be said from the beginning, that capacitances below 100 pF is hard to use in a box like this, due to stray capacitances in wires, terminals and the box itself. This is discussed later below.

Capacitor types

There are many different types of capacitors with different dielectric and specifications. Not all of them have a stable capacitance with regards to temperature, age, voltage, DC bias voltage and other factors. The most stable ones are in general film capacitors, where polystyrene and polypropylene dielectric materials are considered best. Of smaller types, silver mica are highly regarded. Also MLCC (multilayer) Class 1 CG0 (NP0) ceramic capacitors have very good specifications, but aren't available in larger values of capacitance. With larger capacitance, more than hundreds of µF (microFarad), only electrolytic capacitors are practical, due to size and price. Unfortunately common aluminium electrolyte types are not that stable and aren't manufactured with great tolerances. The tantalum type is a little better, especially the solid ones (non-wet). Newer types of polymer tantalum have even better specifications, but are still more expensive.

Capacitor selection

The following capacitors were used:

10 pF 5% 500 V Silver Mica CDE CD15CD100DO3F

100 pF 1% 500 V Silver Mica CDE CD15FD101FO3F

1 nF 0.5% 350 V Silver Mica SGM3-A (Soviet made NOS)

10 nF 0.3% 350 V Silver Mica SGM3-B (Soviet made NOS)

100 nF 0.3% 350 V Silver Mica SSG-2 (Soviet made NOS)

1 µF 0.5% 250 V Polystyrene K71-7 (2 500 nF in parallel) (Soviet made NOS)

10 µF 2% 600 V Polypropylene Miflex MKP10H610G-C

100 µF 3% 250 V Polypropylene (Chinese)

1 mF 6.3 V Tantalum SMD Kyocera AVX (4 x 1 mF in parallel/series)

10 mF 6.3 V Tantalum SMD Kyocera AVX (12 x 1 mF in parallel)

 

The capacitors were sourced from Ebay. Silver mica and polystyrene were NOS (new old stock) and the polypropylene film capacitors were new.

 

10 pF / 100 pF Silver Mica

 

 

 

 

 

 

1 nF / 10 nF Silver Mica

 

100 nF Silver Mica

 

 

500 nF x 2 = 1 µF Polystyrene

 

10 µF Polypropylene

 

100 µF Polypropylene

 

1 mF and 10 mF position

The 1 mF and 10 mF capacitor positions present a problem, though. There are no good single capacitors with good specifications and tolerance larger than 100 µF, unless we talk about metrology spec custom built and very expensive capacitors. Electrolytic capacitors in this range are cheap, but have wide tolerances and isn't good for this usage. The only affordable (for a hobby project) capacitors in this range with quite good tolerances are tantalum SMD (surface mount) capacitors. They are available up to 1 mF relatively cheaply and can be constructed by combining multiple values. However, the 1 mF and 10 mF positions cannot then have a stated tolerance, unless we were to measure them with a calibrated multimeter, or trusted the original specifications of the capacitors.

A batch of 50 1 mF 6.3 V SMD tantalum capacitors were sourced from Ebay. Their measured tolerances weren't really up to spec, so one can assume that these are specimens that haven't met factory tolerances. That's why they are sold cheaply on Ebay. The capacitors were measured with a relatively good multimeter. The batch contained values between 680 and 900 µF.

For easier measurement, cut the SMD capacitor package strips in e.g. strips of 10. Either tape them down on a sheet of paper, or mark the packages with numbers. Now with sharp multimeter probes, you can push the probes through the plastic strips to make contact with the capacitor terminals, without removing the capacitors from the package. Write down the values in a spreadsheet and use a formula for parallel capacitors. 

To get 1000 µF, I selected three random values, subtracting their series value from 1000, to get a value similar to another random capacitor. This way I found 789, 800 and 834 µF in series to get 269 µF. That in parallel with a 731 µF capacitor made exactly 1000 µF.

For the 10 mF position, 12 tantalums were selected to in parallel become exactly 10 000 µF. If the capacitance is larger than in this sample, 11 or 10 capacitors could be enough. This can be done by sorting the values in ascending order in the spreadsheet. Select any 12 in consecutive order and show the sum. Move the selection until the sum is just above or below 10 000. Now exchange one or two values with a lower or higher value until you get close to 10 000.

The tantalum capacitors were soldered to small prototype circuit boards.

Putting together the box

A box with the size 133 x 109 x 58 mm was sourced from the Internet. An Elma audio grade selector switch was used that has very low contact resistance. Two gold plated speaker terminals were used as measurement terminals. A third terminal was mounted and connected to the chassis as a ground terminal for shielding purposes. 

 

A copper strip was connected to one terminal to serve as a common soldering point. Silver plated teflon wire and Stannol silver lead solder was used for good measure.

 

Care should be taken to keep wires short for the lowest values (10 pF and 100 pF) and to keep the wires away from chassis.

 

Capacitors were glued to chassis with a few drops of silicone.

 

There was a dimensional problem with the largest 100 µF polypropylene capacitor. As you can see in the below picture, it was solved by cutting a hole in the bottom of the box.

 

Stray capacitance

The box was then measured with a Fluke PM6306. The following results were noted:

At 10 kHz:

7.94 pF between terminals, switch in off position
8.75 pF between positive terminal and chassis, switch in off position
17.00 pF between positive terminal and chassis, switch in 10 pF position
67.75 pF between positive terminal and chassis, switch in 100 pF position
156.2 pF between negative terminal and chassis (changes a few pF with switch position)

Schematic showing the stray capacitance

The negative terminal has a relatively large copper strip (maybe not ideal) and one pole of all capacitors are permanently connected to this with their own wire. Possibly a lower stray capacitance could be achieved with a two pole switch and shorter wires. The smallest capacitors were placed as close to the switch as possible, soldered directly to its terminals.

Pretty much the same results were obtained with another, cheaper meter. This also tells us that residual capacitance of the box is around 8 pF.

A consequence of this is that the 10 pF position shows 17.6 pF. To adjust this, a Silver Mica capacitor of 24 pF was put in series with the 10 pF capacitor. Now the box shows about 10 pF in this position. Also the 100 pF position is a little high (109 pF) due to the residual capacitance. A 1800 pF silver mica was put in series to remedy this. 

 

The higher capacitance positions weren't affected notably by the residual capacitance and were really good. E.g. 1 nF showed 1.0093 nF, 10 nF 10.005 nF and 1 µF 1.0047 µF. Even the 10 µF capacitor was measured at 10.039 µF.