This instrument is not an "in-circuit" checker. An in-circuit checker does not provide the leakage test feature, only provides a good/no-good operation as the Heathkit models C-1, IT-22, but these models can provide a "clue" about some possible leakage (partial eye opening on the short test). Alternatively the Eico 955, which is a more sophisticated model and perhaps a bit more complicated to use due the added ability to measure capacity in-circuit (but within a limited range). The Eico does not provide any "clue" about some possible leakage due the low testing voltage (6 volt), a leaky capacitor will be considered good unless the leakage was very high (practically shorted).

Obviously, under normal use, at least one end of the capacitor to be tested must be disconnected, but the operation may be complementary to the use of the other instruments mentioned.

Operation is simple, and consists in to connect the test leads to "RES" or "CAP" binding posts (center is POSITIVE and is common to all testings).

• Resistance measurements are read directly on the outer scale when the switch is set at the "R" position. When the switch is set at "Rx100", it is merely necessary to add two zeros to the resistance measurement obtained.
• Capacitance measurements are reflected on 4 ranges, one inner scale marked with 3 equivalent marks and another scale labeled "extended range" located in the extreme inner portion of the dial calibration. On electrolytics the Power Factor control must be set ON, this control only works on the high and the extended range (last two capacity ranges).
• Leakage measurement is accomplished by selecting the voltage desired on LEAKAGE TEST and rotating the other switch to the LEAKAGE position. When this switch returns to NORMAL the capacitor tested is automatically discharged via the 2K resistor used for the "R" scale, this provides a quick and smooth discharge.

When a resistor or capacitor is connected in the correspondent binding posts, the presence of a matched value is indicated by an increase in the shadow angle of the indicator tube, and the presence of leakage is indicated by a decrease in the shadow angle. Calibration is also easy, it consists simply in to connect in its respective binding posts a calibrated resistor/capacitor that matches a center dial value and loose/fasten the knob setscrew positioning it in a correct dial reading.

It seems that the typical conclusion about the mandatory replacement of paper capacitors should be revised. While there are some paper capacitors that die on their own, even without being used, there are others that stand the test of time very well, these are the paper-in-oil type (PIO) capacitors. PIOs have an outward appearance similar to electrolytics, but they are not. These capacitors were relatively not well known because its use was focused on military gear. The reason was that PIOs support more voltage and capacity in smaller dimensions than wax-paper types. The surprise is that many of these old PIOs perform very well when tested, so many times there is no need to replace them. About 70 years ago, all paper capacitors were classified in the same "package", but nowadays it is necessary to make distinctions, and also relate these distinctions to the devices used for testing paper capacitors.

The C-3 does not distinguish between types of paper capacitors, and the C-3 manual does not mention PIOs in any way. Usually PIO capacitors weren't used as a default component, Heathkit did not use PIOs, the manual refers to all of them as "paper", but the evidence is that the dielectric type of paper in oil withstands aging much better than the dielectric type of waxed paper. Tests carried out with C-3 show that capacitors with capacitance less than 1 µF can be tested successfully in the paper position (largest capacitances are an open topic for investigation).

In few words, a capacitor shunted by an invisible resistor. The value of this resistor varies depending on the effective DC current amount. This variation is indirectly proportional: if DC increases, R decreases. A leaky capacitor affects time-constant value and circuit isolation. The importance of this affectance depends on the amount of leakage but, if the equivalent resistance is high, the affectance is low. That is: Leaky capacitors may work perfectly on one circuit and bad on another circuit depending on the relationship between leakage and features of involved circuit. At this point in time the original value of the capacitor is corrupted, mutated to a variable value that depends on voltage and circuit specs. This is easily detected by the capacitance tester circuit included in this instrument. When a paper capacitor misses the original value shows on a lot of cases a leakage clue; the capacitance shift is initially low, but when leakage increases capacitance is seriously affected, capacitor becomes corrupted.
When a capacitor with great leakage drain is measured with this instrument (bridge method) an erratic/undefined result is obtained, and when it is checked on other instrument by the frequency method its meter shows a surprisingly increased "capacity".

How to spot a leaky capacitor?

Checking operating voltages. Strange readings may detect the existence of + DC in places where it must not have (i.e. tube grids) or its absence in places where normally it must have (i.e. cathodes). A hot half-watt resistor in the + DC line may detect the existence of a leaky decoupling capacitor.
Using instruments like the IT-22 and interpreting the shadow gap in the eye tube.

When replacement is needed? (IMHO)

Mandatory:
- Tuned critical circuits (i.e. oscilators, measure bridges).
- Sensitive to bias circuits/components (i.e. power amps, HI-FI outputs, mechanical filters, transistors).
- High voltage circuits (i.e. electrolytics in power supplies, bias filters in oscilloscopes).

Recommended:
- Passive tuned circuits (i.e. IF's, filters, HI-FI compensating networks).

Optional: (if the leakage level is low/moderate).
- General low-profile circuitry (i.e. decoupling capacitors, pass capacitors if the coupled circuit is not critical).

Is it possible to evaluate if the leakage is so significant?

As commented above, importance of leakage depends of the relationship between shunt-resistance and features of associated circuit. The Condenser Checker allow us "to calculate" this shunt-resistance by testing a capacitor under several voltages. This instrument is very sensitive to leakage current (too sensitive in my opinion). If the test on nominal voltage range is OK (totally open shadow), capacitor is EXCELLENT. But... what about a 400 VPP capacitor that shows totally open shadow in the 25 volt range but it goes closing increasingly up to 350 volt range? Is it usable? To determine that, I consider a "very acceptable capacitor" the one that has an equivalent leakage greater than 10 Mohm, and a "questionable but still usable capacitor", the one that has an equivalent leakage greater than a 5.6 Mohm (measured at its NOMINAL VOLTAGE, at lower voltage the leakage is significantly lower). These considerations applies to non-electrolytics only.

1. Connect a 10 Mohm resistor in series with an milliamperimeter (tester on mA range) in the CAP binding posts, setting the voltage range to working voltage. Power Factor OFF. Note the reading.
2. Repeat the procedure with a 5.6 Mohm resistor.
3. Repeat the procedure with the capacitor being tested; compare the readings.

Usually a capacitor with low to moderate leakage will work correctly associated to low impedance components or in "low-profile" circuitry. Broadly, paper capacitors greater than 0.1 µF must be replaced both in high or low voltage circuits because large capacity paper capacitors are more sensitive to leakage precisely due their higher capacity (it has more waxed paper area susceptible to corruption). From 0.01 to 0.1 µF it depends on voltage applied, and those within the 0.001 to 0.10 µF range are perfectly usable unless some exceptions. In case of not considering these thoughts a massive recapping should be performed, this may be a lot of work, since all European equipment manufactured until early 60's and most American equipment manufactured until late 60's would need attention (there are some exceptions, e.g. Hammarlund). This also creates another problem: the "excessive-new" component along with the old one, a kind of "frankenstein circuit", which it is clearly evidenced when solid-state circuitry is added to tube circuitry.

An example of these considerations: the C-3 itself.

- The 2 µF bridge capacitor (reference value for the bridge): Leakage in this capacitor will cause bad readings. Same applies to the "20 Kpf capacitor" (2 x 10 Kpf). Replacement of both will be mandatory.

- The 0.25 µF decoupling capacitor included in the leakage test circuit: This condenser can accept low leakage. A leaky capacitor in this place will cause lower testing voltage. Capacity value may be affected but it is not critical.

- The 0.01 µF coupling capacitor to 1629 grid: This condenser can accept moderate leakage. A leaky capacitor in this place will cause lower speed in eye opening. It works at low voltage; the leakage at low voltages is generally low. Capacity value may be affected but it is not critical.

Preparing a stored electrolytic capacitor to be used.

The procedure is: Power Factor ON and checking leakage by steps starting from 25 volt up to nominal voltage of the capacitor and watching the eye tube opening. If the eye opens totally, jump to next scale. Stop when the eye starts to close totally. And now the most boring part... maintaining the leakage knob turned ON for a while, watching the eye opening again... (in order to avoid this, the return spring would have to be taken off, but it is not reccommendable for checking under normal conditions). After awhile the eye increases its opening... jump to next voltage range... and repeat the same procedure until the eye is reasonably open at its nominal voltage.

This procedure cannot recover the original value of a electrolytic capacitor partially or totally dry, as it is not possible to recover, only concerns regarding leakage. Therefore, it's always necessary to verify capacity before the leak check. Fortunately, the C-3 allows that, and the ability of capacity measurement built into the device is very useful. A dry electrolytic capacitor will show an erratic behavior, no matched value on its own range, and no logical balance on the rest of ranges.

Level of closure/aperture of the eye tube Mod:

The 1M resistor between pins 3 & 4 (plates) determines the limits of closure/aperture of the eye tube. If the value of this resistor is too low the eye will not open fully and it could strongly overlap when closed. An increased value makes possible to adjust an adequate closure with moderate overlapping or simply get the eye just closed when nothing is being measured. A 3M3 resistor would be a good candidate.

"25" Switch Position Mod:

The "25 volt" position for the leakage test is, in fact, a "40/45 volt position" (Power Factor OFF/ON). This is a general factory issue of this instrument, but... which is wrong? ... the labeled "25 V" mark or the internal divider design? (45 volt seems best scaled to the next 150 volt position than 25 volt considering voltages involved in tube circuitry and their components). In order to create a true 25 volt position the last two resistors on the low end of the voltage divider must be changed: 10k by 5K6, and 22K by 27K. If the test voltages for the rest of the positions needs to be increased: 6K8 and 33K. Modifying the circuit the 25V range will be implemented correctly, but if you consider 45 volt more useful than 25 volt (like me), delete and change "2" to "4" in the writing.

Voltage Balance Mod:

The 47 K resistor and the 22K-22K-22K-22K-10K resistors in series are the main voltage divider: 47 K resistor applies for the 1629 tube (about 150V) and the "98 K resistor" applies for the leakage voltage selector (about 450V). Aged components will cause modification of this balance. If it happens, the 1629 will run too hot, or the leakage voltage selector will be overcharged. Replacing this 47 K resistor makes possible to restore original balance: If R increases, the voltage of the 1629 tube increases, and the voltage of the voltage selector decreases; conversely if R decreases.

Tech Notes: