Tutorial?: Test for coupling cap leak in circuit

[jordan86]

I did a quick search and could not find this covered. I've seen some different practices on this before though and wanted to see if I could get some simple and definitive steps.  I have an 1965 AB763 in almost original condition including all caps. I intend to replace electrolytics but would like to test the coupling caps for leakage, as it has some low output issues. They are the blue molded ones so I'd like to do as little desoldering/removal as possible.
All the dc voltages seem to check out pretty close the schematic spec. I've not checked the grids though.

So a few questions:

1) Can you test for dcv on the downstream side (grid side) of the coupling cap in circuit? Neg probe to chassis ground.
2) Or do you have to pull that grid side leg out of circuit to do that?
3) Should the power tubes be removed during this process?
4) How many millivolts are generally "acceptable" (1mv?), or is ZERO the only option?

[WimWalther]

1) Can you test for dcv on the downstream side (grid side) of the coupling cap in circuit? Neg probe to chassis ground.

Absolutely, this is the most common method. You don't need to move or disconnect anything, at least initially.

Just measure the voltages on the power tube grids and see how they compare to one another. As far as what you should expect to see, it depends on the bias scheme.

Self bias amps should have a few mV of DC. Those with fixed grid bias should be about 1V lower (less negative) than the bias supply itself.
[/quote]

[shooter]

like to test the coupling caps for leakage

in-circuit, working "normal", you should find VDC on the "left" side, NOT the "right". (mV are expectable, more than a volt, iffy)
it gets a little sketchy if the coupling cap is connected to a grid, your meter will "add" stuff that really isn't there.

[mresistor]

what about testing for DCV on the pots.

[sluckey]

You must understand the operation of the downstream circuit if you want to check for cap leakage in circuit. I can think of three common circuits that will always have voltage on the downstream side of a coupling cap,,, cathode follower grid, LTP PI grids, fixed bias output tube grids. There's only one way to accurately check for cap leakage in these three examples. You must disconnect the downstream lead of such a coupling cap. Another not so common example is a grid leak biased tube. They operate with the grid at a negative voltage and the cathode is usually connected to ground. In fact, any tube that has it's cathode connected to ground will have a negative voltage on the grid.

And then there's the crazy way the 5E3 volume control is wired! If the volume control is set to zero and you're trying to check for cap leakage, the downstream side of leaky coupling cap will be at ground and falsely show no leakage. But turn the vol pot up and there is the leaky voltage.

These are just common examples of circuits that you must disconnect the downstream side of the cap to properly check for leakage. I'm sure there are other examples. If you didn't know some or all of this, then I advise you to ***ALWAYS*** disconnect the downstream lead of a suspected leaky cap. Then you will ***KNOW*** for sure.

what about testing for DCV on the pots.

Good point. But here again, you must know how to analyze a circuit. Look at the INTENSITY pot in a Princeton Reverb. The pot has a negative voltage on all three pot lugs. You would have to disconnect the downstream side of that .1µF coupling cap to test for leakage.

Sometimes it's perfectly normal for voltage to be present on both sides of a coupling cap.

[WimWalther]

@jordan86

Everything others mention is useful, valid information. My comments specifically address the final output tube coupling caps as your questions mention removal of the power tubes, and so I assumed you were asking about that section specifically.

As sluckey mentioned, there are a number of circuits where you will find moderate to high DC voltage on both ends of a cap. In these cases, you either need to know exactly what to expect or one end of the cap must be disconnected to make a measurement.

[tubeswell]

See Fig. 4-4 on p57 of CH4 of the Jack Darr book

[jordan86]

Working on this today. I’ve pulled three legs on different couplings caps. The two going to the PT grids and one off of normal channel stage 2 (V1B) that goes to the preamp mixing resistors.

I am measuring mV but my meter is bouncing around quite a bit. Is that normal? I read on GeoFX about putting a 1M resistor to ground from the lifted leg and measuring across the resistor. I’ve done that and the meter still bounces around. Jumps from like -20mv to +8,15, or even 30mV.

Side note: The amp I’m working on is a 1965 Deluxe Reverb. I replaced all the electrolytics. Voltages and resistors seem to be close within spec.

The amp stays cleanish all the way to 10 and sounds quieter than my Princeton reverb. Normal channel is also quieter than the Vibrato

[jordan86]

Also measuring about 28v on the grids of the PI but I presume that is normal.

[shooter]

IF you're measuring VDC.... MINUS volts shouldn't really be there, 0 - 30mV, I'd call acceptable


but;
you can see minus VDC IF your ground reference is "above" ground
a sketchy meter
you're actually reading VAC and the meter is trying to "make it VDC"

[tubeswell]

If you are clipping the VDC meter leads on before firing up the amp (as you really should), then you will see voltage on the ‘floating’ lead of the cap for a second or so while the other cap plate charges up to the quiescent high voltage (and that is normal because the cap’s floating plate is ‘seeing’ a changing electrical charge on the other plate as it changes voltage as it charges up) but after that, once the ‘live’ plate settles to its quiescent voltage, you shouldn’t read* any voltage on the floating plate of a good cap because induction only occurs across the dielectric barrier when there is a change in current (*if you’re switched to the ‘hundreds of volts’ setting on your meter).

[jordan86]

Thanks, everyone. I really suspected I had some leaky coupling caps but this process has confirmed they are indeed fine and good. In the process of soldering my PI coupling caps back in, I did another triple check on things and discovered the actual culprit to my low volume issue. Poor ground on the Phase Inverter section. Ground wire was just barely resting on top of the pool of solder on the grounding plate. Crazy that it's been like that for 60 years. I nudged it ever so slightly and it broke completely free. Got it soldered back in well and she rips now!