Cap Leakage Tester Schematic

[RockinRocket]

I want to build a cap leakage tester and looking for something simple.

I have a small power transformer that will support 1 12ax7. I forget the specs but should do 250-300v easy.

I have a small enclosure box 6x8x2" or so.

Any ides?

[Prairie Dawg]

Haven't been around here in a long time.

You can build a direct reading one from the schematic in Alan Douglas' Tube testers and Classic Test Equipment which I did.

Or what's easier is to get yourself an insulation tester like the HVAC boys and girls use, they've got a little power supply that can produce 1kv. I lift one end and then test them and if they show less than infinite resistance (2000 megohms or higher) out they go.

[Tony Bones]

If you have a power supply, a resistor, and a DMM then you can measure leakage.

Apply some DC voltage across the cap. The leakage (if any) is DC current. How to measure it? Disconnect the power supply from one end of the cap and insert the resistor in series. Now all of the leakage current goes through the resistor. Measure the voltage drop across the resistor with the DMM; you're measuring leakage current.

It's safer to put the resistor on the ground side instead of the HV side, but it's up to you.

Choose the size of the resistor depending on how much current you plan to measure. 1uA will generate 1V across a 1M resistor.

[R.G.]

Tony is right about the method. However, be aware that anything with "micro" in the word gets difficult to measure with a hand meter. Getting any kind of accuracy means you have to know how your meter acts with tiny voltages.

Meters all have some internal resistance. In olden, analog meter days, this was often 20K to 50K ohms per volt on the voltage ranges, so on the 200 volt range you'd get 4M to 10M. That's pretty good until you start measuring voltage across a 1M resistor. In that case your meter itself is causing a 25% to 10% error because of the meter resistance in parallel with the resistor you're measuring across.

Modern digital meters are nominally better than analog ones, having a high and more constant input resistance, generally about 10M. Problem is, they often don't tell you what the input resistance is. It's usually only specified as greater than some number.

The better techs in the Golden Age solved this issue by testing their meters to discover what their internal input resistance was. That way, they could do the measurement and correct (by pencil and paper or ... slide rule...) to get a closer and more informed answer. Knowing this about your digital meter would give you a really good way to make this measurement.

The advice on replacing leaky caps in the old Jack Darr amp repair books was to hook your (analog) meter in series with the cap and a power supply and measure the voltage. The criteria for "good" was to measure no voltage higher than zero on any range.

[RockinRocket]

Great info

This is something any builder should test before building the amp

[trobbins]

Testing for leaky coupling caps is always a good initial test to apply to a new or just restored valve amplifier. This is typically done by removing all valves, and using a variac to raise all the stage B+ levels up to their max capability for capacitor voltage rating (supply decoupling and signal coupling capacitor ratings). Connecting a digital voltmeter across each grid-leak should measure effectively zero volts, and similarly across each B+ supply chain dropper resistor or screen decoupling feed resistor.

Digital megohmmeters can come up cheaply on ebay, and are a very useful test tool for other parts as well as capacitors - especially for transformer insulation, and even if you have some high value resistors that a typical DMM falls short on.

Sometimes vintage boat anchors come along on ebay that have specialist abilities for insulation resistance testing. I recently restored a Megohmmeter that covers the range 1Meg to 1Tera, assuming your DUT can cope with 1kV.

[stiltamp]

Here the schematic of a leakage current tester,
unfortunately only in German language.

CapacitorLeakageCurrentTesterGermanLanguage.pdf

[Tony Bones]

RG makes a good point about being conscious of a meter's input resistance; it appears in parallel with the sense resisor. But, All the 'modern' DMM's that I've looked at have an input resistance of 10Meg. That includes some cheap meters that I bought at Harbor Freight when they were 1/2 price ($2.98 at the time; I bought ten.) As long as your current sense resistor isn't more than 1Meg, you can reasonably ignore the influence of the meter in this application. If you need to measure microamps to three digits, then that's a different problem. We're just looking for leaky caps here!

[tubeswell]

Besides measuring for VDC on the ‘AC lead’ of each coupling cap, you will usually get other contextual clues that a cap is leaky, such as a scratchy vol or tone pot, or dramatically reduced output signal (eg from a leaky cap throwing off the bias voltage in a following preamp stage), and in the case of a leaky coupling cap to a power tube grid, signal distortion (from PI output imbalance) or red-plating from loss of bias.

[stiltamp]

This is my translation of the German article form above, english is not my mother tongue, so please be indulgent to my mistakes.

The test voltage is set by the ratio of the voltage divider left.
Put in the capacitor you want check, after that actuates the switch.
Now the capacitor will be charged.
In order to reduce the time taken to charge the capacitor
a momentary switch can connected in parallel to the 2.2 MegOhms resistor.
As soon as the capacitor is charged, open the switch.
Now starts the discharging (depending on the insolation).
The test jacks, the wires and the componets arangement requiers a a good insulation,
preventing current leakage.
The 4.7nF capacitor in parallel to the test capacitor must be a high quality type.
This capacitor is needed for test capacitors less than 1nF,
because a new charging current impulse would be to small to be displayed on the magic eye.
The 10 Kiloohms potentiometer being so set as to the light bars of the magig eye are about half closed.
(This corresponds to an infinite insulation resistance value)
After that, the capacitor to check is connected to the test jacks.
As soon as the switch is actuated the capacitor woll be charged.
A poor insulation resistance increases the grid voltage in positive direction,
this leads to a reduction of the light bar lenght.
If the lightbares completely disappears, the testes capacitor is unusable.
The highest test voltage (200 Volts) gives the highest display sensitivity.
A insulation resistance value of 500 MegOhms corrosponds to a decline of about 4 Millimeters on the light bar.
If measurement of insulation resistance values less than 500 MegOhms required,
the 10 Kiloohms potentiometer must be turned up.

LeakageTester.png

[martin manning]

Nett, danke!

Pretty simple circuit, with (bonus) great retro vibe. Just needs a 6.3VAC/250VDC supply, I think I would use a momentary contact SPDT toggle switch for the one that applies voltage to the DUT, and connect the other (normally closed) terminal through a resistor to ground. That way the cap and test terminals are quickly discharged and held at ground potential to avoid unexpected surprises. An on-on-on toggle would work for the voltage selector to keep it simple and compact.

[stiltamp]

Thanks for your modding hints.