[R-390] Carrier Level Meter Issues

[Roy Morgan]

Below is the whole of my Diatribe: capleakagetesting.txt. It tells about safe ways to test even very tiny leakages in caps.  Electrolytic cap testing is covered in reform.txt

> On Mar 9, 2022, at 2:23 AM, Jim Whartenby via R-390 <r-390 at mailman.qth.net> wrote:
> 
> When doing any kind of leakage testing, you have to think about unintended consequences.  
> If you plan to measure current by putting the VOM or VTVM in series with the DUT (IN CURRENT MODE) (Device Under Test) plus the power supply then if the DUT fails in a short, the power supply will force the maximum current that it is capable of supplying through the meter.  That will do wonders to the pointer of the analog meter movement or the circuitry of a DMM.  Not to mention heating up the DUT.
> You should limit the maximum current so as not to overstress either the DUT, the power supply or the test equipment.  …

Capleakagetesting.txt
From K1LKY

Capacitor leakage testing

This is my diatribe on testing capacitors for leakage.  Other diatribes are:

reform.txt  about reforming electrolytic capacitors
powercordsandbypassing.txt  about safe use of line cords and line bypassing
capleakagetesting.txt  about testing capacitors for leakage
variacs.txt  about the use and dangers of variacs
(coming soon: linebucking.txt  about reducing line voltages)

1) Find a B+ supply that will deliver a voltage as high or higher than the capacitors rated working value.  A variable supply is nice but not necessary.  Current capacity is not important - a few milliamperes is fine.

2) Get a VTVM or a DVM with high input impedance (10 megohms is common).

3) Set the voltmeter to measure volts on a range above the supply voltage.  Connect the common terminal of the supply to the common terminal of the voltmeter.

4) With the supply off for safety, connect the capacitor from the high side of the supply to the high side of the voltmeter.

5) Turn on the supply.

6) Observe the meter.

The meters input resistance causes it to operate as very sensitive microammeter.  A reading of 10 volts across 10 megohms indicates a current of one microampere.  One volt, one tenth microampere, or 100 picoamperes.

You can do the equivalent of this test for coupling caps by removing the tube from the circuit and carefully measuring the voltage at the grid of the following stage.

Example: with a supply voltage of 350 volts and voltmeter indication of 50 volts, the capacitor is conducting a current of 5 microamperes.  The capacitor has an impressed voltage of 300 volts (350 minus 50).  You can figure the capacitor leakage resistance by Ohm's law, or by proportions.  Figuring by proportions, it has 6 times the voltage as the voltmeter, so it has 6 times the resistance, or 60 megohms.

In my experience, it is common for old paper capacitors to indicate one quarter to three quarters of the supply voltage in this setup.  It is also common for modern film capacitors to indicate less than a few tenths of a volt.

Consider the case of an old paper .01 uF capacitor feeding the audio output tube in a receiver. The preceding stage operates at a plate voltage of 200 volts.  The old paper capacitor leaks about 100 microamperes.  The output tube grid resistor is 100 Kohms.  The voltage developed across the grid resistor from the leakage is 10 volts.  This 10 volts reduces the grid-cathode operating bias on the audio output tube from minus 14 volts to minus 4 volts.  In the case of a 6V6, or 6AQ5, that will increase the standing plate current from a normal 25 or 30 ma to about 80 or 100 ma.  The audio will sound terrible and the tube will last only a few hours instead of a few thousand hours.  Leakage in the blocking cap at the audio pre-amp stage is even more damaging to the sound since the stage operates at lower bias levels.  (Notes about BC-348's:  1) Some use the resistance of the filter choke in the negative B+ supply line to develop bias for the audio output tube.  Connecting a home brewed plate supply directly to ground causes the output tube to run with out bias, resulting in terrible audio and a quickly worn-out tube. 2) Many BC-348's use flat, black, rectangular bypass caps that are not mica but are paper. They are almost invariably leaky or shorted.)

Consider the case of a screen bypass capacitor in a receiver IF stage.  The B+ supply is 220 volts, normal screen current is 5 ma, screen resistor is 22K, and screen voltage is about 110 volts.  The tube operates with normal gain.  Now, if the screen bypass cap leaks 3 milliamps, the screen voltage will go down to something like 60 volts.  The tube will operate a lower gain, will not respond in the same way to AGC voltage, and will be more subject to overload and distortion on strong signals.  If many IF and RF stages are having similar screen bypass leakage problems, your radio will be quite dead.  I have a number of as-yet un-re-capped  receivers like this.  Recently an un-restored SX-101A produced a faint pop and it's gain dropped dramatically.  I suspect a shorted screen bypass cap.

You can measure screen bypass and grid coupling capacitors in circuit by pulling out one or more tubes and measuring voltages on either side of the cap.  Take into account the voltmeter input resistance and any resistance to ground on the non-B+ side of the cap, such as the grid resistor.  You can do this withOUT removing any modules from the chassis in the R-390A.  Count your tube pin numbers in the correct direction (counter clockwise) when working from the top of the chassis.  A little drawing to keep nearby can help you in this.

Note:  Many older radios were measured with 1000 ohms-per-volt meters and the reported normal tube voltages reflect this.  Most affected are screen voltage and  voltages in high resistance circuits.  Notable examples are the TV-7 tube tester and most pre-war receivers.  Your TV-7 will *not* be calibrated correctly if you do it with a modern 10-megohm input resistance meter.  Just add a resistor in parallel with the meter appropriate to the scale you are using. (...Full scale volts times the "ohms per volt" of the meter they used.)

From: Henry van Cleef <vancleef at netcom.com>  
Subject: Re: Reforming, Chapter CCXXVI 
To: Old Tube Radios <boatanchors at theporch.com> 
Date: Wed, 7 Jun 2000 22:22:37 -0600 (MDT) 
Cc: boatanchors at theporch.com  

Don, I don't think it makes much difference whether you do it a 
section at a time or multiple sections all together. Indeed, I think 
we sometimes may make too much of a process out of reforming old caps.
Consider the Tek 530/540 scope. These are chock full of Mallory FP's 
(don't think Tek used any other in these), and have a time delay 
relay. Turn it on, and if it isn't smoking after 45 seconds, the 
relay clicks in, and everything gets hit with volts from some very 
large (amps continuous) power supplies. Yet they always come back to 
life. I swapped notes with Stan Griffiths on this a while back, and 
both of us have only had to replace a very few that had lost their 
capacitance. 

(note: the MIL SPEC on capacitor testing advises to use a reforming current of about 5 ma.  Bill Carns of the Collins community tells of carefully reforming many many electrolytic caps with a max of 5 MA reforming current and very high success rate.
Further note, that MIL SPEC has some seriously dangerous assumptions - to not do what it says with paralleled capacitors.)

Generally, when I light up an unknown scope for the first time, I have 
VTVM's on all of the power supply voltages (-150, 100, 225, 350, 500) 
and watch them as the relay clicks in. Generally the meters jump to 
life right where I would expect them, and most of the time, if they 
don't, it's poor contact with one of the regulator control tubes. 
I do limit the first run to a couple of minutes after the relay kicks 
in, and do a couple of more short (5-10 minute) runs while checking to 
see if the CRT lights up and if the horizontal stuff is working. 
That's forming with a vengeance. 

I'd hang some voltmeters on those Aerovoxes and hit them with working 
voltage applied to the other end of some 100 ma. current limiting 
resistors, and watch them come up. Turn them on and off a few times 
and let them cool if they are slow to come up. After toasting with 
low leakage for half an hour, check them for capacitance and low 
series resistance. Generally, when a cap of this type is tired, it 
will discharge to ground through 1K, then the voltage (measured on a 
typical 11 megohm VTVM) will come back up to 10% or more of the 
charging voltage. That's the sign of a sick puppy. I've got a 1944 
box with a bunch of 3-section Aerovoxes (20-20/450, 40/50, I think) 
that are all bad (series resistance). FP's generally come right up, 
even the 1941 jobs, unless they're leaking "coolant" (obvious) or have 
dried out. 

The ones that would short were the wets from the 1930's. My theory on 
this is that the electrolyte attacked the plates and the metal ions in 
solution made them conduct, with no film interface. The drys are not 
"dry," but the electrolyte is held in a blotting paper. All of the 
electrolytes I know of are boric acid based, but I think the 
manufacturers got smart about putting corrosion inhibitors in these 
(always proprietary) solutions by 1940. 

The little caps that get hot and go bang, like firecrackers, are 
tantalums, and I think their electrolyte is a nitric acid based 
solution. Very nasty. I recall having lots of trouble with popping 
tantalums in the '60's, but most of this was traced to installation 
with reversed polarity. 
Hank

John Poulton measured a number of new old stock silver mica caps and reports his results at:
http://jptronics.org/Collins/silvered_micas/
“My conclusions. NOS silver micas retain their leakage resistance performance exceptionally well over time. NOS silver micas also are very stable over time for their rated capacitance values. I would not hesitate to replace a silver mica cap with a vintage silver mica cap, but I would not do so without confirming leakage resistance and capacitance values first. “


Roy


Roy Morgan
K1LKY since 1958
k1lky68 at gmail.com
Western Mass