• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Please educate me: Tube testing, the life test, and Grid Leakage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Been testing a bunch of Vintage EL84's and 7189's lately. Many of these I have used over the years and they show healthy emmissions but fail the life test. As I understand it, the life test lowers the filament voltage to see if the emissions remain constant. Tester is a Solid State Sencore TC162 Mighty mite V11. Oddly, I even have NOS 7189's from reputable sellers that are failing the life test. Didn't have a tester when I bought them 20 plus years ago. Not sure how much faith I should hold in the life test. These tubes sound great and are unobtainable now so I'd really like to use them. Yes, the Sencore is a "toy", but do I see people with real Hickoks, etc. screening tubes with a life test? Or does their "real" testing make it unnecessary. Are these going to fail any day and take out an amp? Some will drop so slowly into the "bad" area on the meter that I wonder if I should not be holding the life test switch on for so long. It is a spring loaded momentary contact switch. I'm no expert at this, can someone advise me on the proper way to use and interperate these tests.
On a related note, I don't understand what grid leakage is. A couple of these NOS Rca 7189's have healthy emissions, don't waiver a bit through the life test, but show slight or more grid leakage. Tester manual sayes they can be used in a pinch, but keep an eye on them and replace when possible. This advice is aimed at TV repairmen though. Remember, these are audio output tubes, so I don't want to jeopardise the amps with tube failure.
So, should I be rejecting anything that fails the life test or shows grid leakage? One tube goes to the bad zone during grid leakage test and then climbs back up to good. What about this one?
Thanks for any education you guys may offer on this subject, Chris
 
I use the TC142 which doesn't use the Life button. I just drop the voltage switch 1 notch and that usually tells you when you have a strong tube but in todays world of higher line voltages the filaments get higher than normal voltages. So if your tubes are failing the life test now, they are certainly weaker than the tester is even telling you. That said, the supply to your amp is going to be higher too, so you may get good emmissions for a while even from a weak tube.

The grid leak test is a different matter. If you get much of a bump off the bottom in the grid leak test you could have a tube that will get seriously leaky in the circuit under full power and heat. The best test for that is to get a testing socket and measure any DC on the grid with a K biased tube running in the amp. A "NOS" tube can sometimes show grid leak and this will diminish as the tube is burned in for the first few cycles, so in that case it's best to run a new tube that may show a grid leak, in the amp with grid voltage monitoring.
 
As I understand it, the life test lowers the filament voltage to see if the emissions remain constant.

Not constant, but a drop of "x" amount or percent - the amount of drop depends on the tester (my B&K 747B has a "life test" switch that lowers the heater voltage by 10%). The theory is that a tube approaching the point in its life where the cathode can no longer maintain adequate emission will have more drop off when the heater voltage is reduced. It's a very subjective test, and even after all these years and 1000's of tests I'm not sure I can give you a hard and fast rule of what it good and what isn't.

Oddly, I even have NOS 7189's from reputable sellers that are failing the life test. Didn't have a tester when I bought them 20 plus years ago.

After 20 years in the box there is a good chance they are experiencing some cathode poisoning - contamination deposited on the cathode from other internal parts of the tube outgassing, etc. Almost always those tubes will be fine after running for a few hours - or you can use the Morgan Jones "bake the tube" trick, either of which will remove the deposits and return the tube to full health. DON'T TOSS THEM OUT!!

Not sure how much faith I should hold in the life test. These tubes sound great and are unobtainable now so I'd really like to use them.

See my comments above...

Are these going to fail any day and take out an amp?

If the only test they fail is the life test it is HIGHLY unlikely they will fail catastrophically.

On a related note, I don't understand what grid leakage is.

A bit oversimplified - but it is current flowing from the control grid to other elements in the tube that shouldn't be flowing.

A couple of these NOS Rca 7189's have healthy emissions, don't waiver a bit through the life test, but show slight or more grid leakage.

Again, running the tubes for a while will often reduce or eliminate any excess grid current. Another thing to watch out for is deposits between the tube pins - this can cause the tube tester to show grid leakage, but it isn't real. Clean the base and pins well and it'll be gone.

Tester manual sayes they can be used in a pinch, but keep an eye on them and replace when possible. This advice is aimed at TV repairmen though. Remember, these are audio output tubes, so I don't want to jeopardise the amps with tube failure.

A LOT depends on the circuit. Some tubes with a small amount of grid leakage will perform very well in some circuits and will fail in others.

So, should I be rejecting anything that fails the life test

NO! Failing the life test (such as it is) just means the tube has some wear and tear on it. It is not an indication of impending catastrophic failure.

or shows grid leakage?

Maybe. See what I wrote above.

One tube goes to the bad zone during grid leakage test and then climbs back up to good. What about this one?

Run it or bake it a la Jones - then re-test. I won't be surprised if it passes with flying colors.

As SY said, trying it in circuit is by far the best test - but sometimes you don't have gear around that uses the tube you want to test!

My advice is this:

1. Get the best tester you can afford, a "mutual conductance" tester is far better in many ways than an emission tester.

2. Read the manual carefull and completely.

3. Test as many tubes as you can to gain experience. I'll never forget when I threw away a quad of Amperex EL84s - they showed all shorts on my just purchased TV-7B, my 1st tester. Well - it turns out that's normal with the pinout on Euro EL84s! Don't fall victim like I did!!

I hope this was some help anyway...
 
Thanks to all for your comments so far. In particular, thanks Jim Mc. for concisely commenting on all of all of my questions. You have put my mind at ease a bit about still using these great old RCA's. Such a shame to have them sit in a drawer. I've been using Russian stuff while these just sit there. I will not be afraid now to run the grid leaking tubes, but will pull them out for retesting after a few cycles.
 
I just got back from googling the Morgan Jones thing. Haden't heard of this before. Now I am wondering if I should be doing this with the Grid Leakage tubes. Would leaving them in the tester under emissions test have a similar effect? I have no idea what the temp of the tube is while under testing, or operating. I'm guessing though that they might get hotter in the Fisher 30A's than in the tester.
Any thoughts?
 
I just got back from googling the Morgan Jones thing. Haden't heard of this before. Now I am wondering if I should be doing this with the Grid Leakage tubes. Would leaving them in the tester under emissions test have a similar effect? I have no idea what the temp of the tube is while under testing, or operating. I'm guessing though that they might get hotter in the Fisher 30A's than in the tester.
Any thoughts?

You could leave them on the tester, or plug them into an amp (if you have one for that particular tube) and let the filaments heat up for awhile. The basic idea is that heating the tube activates the getter which will then eat some of the stray gas molecules flying around; it's these guys that can cause excessive grid current.

My Hickok tester gets really warm when I test tubes for an extended period of time (it's internals are completely sealed up), so it may or may not be a good idea to use your Sencor tester. I suppose that you could just wire up a tube socket for either 6.3V or 12.6V and bake them that way. If you are contemplating putting them in the oven, make sure the wife is out running errands and go easy on the temp dial as it's easy to melt/distort some of the valve bases.
 
I would like to understand tube testing - so was going to raise this very topic.
Thanks to everyone who has added useful info.

It seems as a tube is used, its transconductance lowers - is this correct ?

However, I can only find one graph of how gm decays with use - which is in this article on page 173.
http://www.alcatel-lucent.com/bstj/vol36-1957/articles/bstj36-1-163.pdf
It seems gm does not decay linearly with use.


Also, in a signal tube - its pretty clear that each triode should be tested individually.
Matching triodes is probably quite important for differential pairs and phase splitter circuits.


To test a tube, building a circuit to convert a scope into a curve tracer looks a bit much.

However, here is an idea that suddenly dawned on me after looking at a datasheet
Build the class A reference circuit as shown in the datasheet - apply a signal generator - and look at the output on a scope.
Then check the gain of a new reference tube versus the tube that is being tested.
http://www.radiostation.ru/tubes/6SN7.pdf

Would this work ?
.
 
Last edited:
For a tube - here is the only gm decay curve I can find - I've posted it only to show the shape of the curve.

For a 6sn7, the life expectancy would be more in the order of 10,000 hours.
 

Attachments

  • img_0871_2.jpg
    img_0871_2.jpg
    234.6 KB · Views: 339
.
However, here is an idea that suddenly dawned on me after looking at a datasheet
Build the class A reference circuit as shown in the datasheet - apply a signal generator - and look at the output on a scope.
Then check the gain of a new reference tube versus the tube that is being tested.
http://www.radiostation.ru/tubes/6SN7.pdf

Would this work ?
.

Yes. Your idea is better. In fact testing the tube in the intended circuit is the best way to test a tube. In fact even if yo own a "real" tester this way is still the best.

But even with the above you may still reject some tubes that would work in some circuits. Tubes inside a local feedback loop or cathode followers can work even if they test poorly in the class A amplifier. In fact I've swapped 12AX7 for 12AT7 and don't see any difference in performance in cases there the tube is on a local NFB loop. That said, most of the time the tube, or at least one section of it is NOT in such a circuit.

As for balance in applications like phase splitters. If you care a lot, design in a "balance" adjustment.
 
Yes, testing the tube in the intended circuit is the best way to test a tube. In fact even if you own a "real" tester this way is still the best.

But even with the above you may still reject some tubes that would work in some circuits. Tubes inside a local feedback loop or cathode followers can work even if they test poorly in the class A amplifier. In fact I've swapped 12AX7 for 12AT7 and don't see any difference in performance in cases there the tube is on a local NFB loop. That said, most of the time the tube, or at least one section of it is NOT in such a circuit.

Thanks for the guidance.
For a 6sn7, here is the class A resistance coupled amp as shown in the data sheet.

I'll need to give things a bit more thought before asking anymore questions.
 

Attachments

  • img_0873_2.jpg
    img_0873_2.jpg
    251.1 KB · Views: 305
Last edited:
For a tube - here is the only gm decay curve I can find - I've posted it only to show the shape of the curve.

For a 6sn7, the life expectancy would be more in the order of 10,000 hours.

The practical problem with tube testing for life expectancy is this:

Lets agree the when you are at 80% the tube is "on it's way out". But the 80% figure starts from whatever the tube measured when it was new. Do you know what it measured new? Tubes can vary by 20% out of the box. So you might have a brand new tube that measures 80%

So the only way to use this curve is to start a measurement program where you test each tube when new and then again every few hundred hours and look at the SHAPE of the curve and use relative numbers

In case I'm not clear. Lets say you want to measure a block of ice to see what percent of the ice has melted. So it weighs 3 pounds. Now you tell me what fraction of the ice is gone. Hard to do without knowing the starting point

One more thing. Your class-A test circuit is very sensitive test. But some circuits in amps are not. For example unity gain cathode follower might show zero change as the tube ages until one day it just stops working. This is why tube-swapping sometimes fixes a problem. A nearly dead tube would work fine in that spot. So what is the definition of "lifetime"? How many hours it will work in a class A common cathode amp or how many hours it will work in your specific use case?

Your best bet if you are worried about tubes is you buy a new tube and verify it works. Then keep that as your "standard" and if ever you suspect a tube of that type has failed swap it out for your standard and see if that fixes it.

These preamp tubes will last for years and tears, decades maybe.

If you want to make them last even longer then control the heater voltage. I typically place a big sand stone power resister in series with the heaters to bring the voltage down to about 5% under 6.3 volts, or about 6.0 volts. Doing this dramatically lengthens the life. the specs say you can go even lower but I leave room for low AC mains voltage. May tests show the tubes work fine at 5.5 volts. THey tend to be a little bit over engineered. the 6.0V setting is conservative and works. I use something like a 0.1 to 0.68 ohm resister. It does not take much to drop 0.3 volts
 
If you want to make them last even longer then control the heater voltage. I typically place a big sand stone power resister in series with the heaters to bring the voltage down to about 5% under 6.3 volts, or about 6.0 volts. Doing this dramatically lengthens the life. the specs say you can go even lower but I leave room for low AC mains voltage. May tests show the tubes work fine at 5.5 volts. THey tend to be a little bit over engineered. the 6.0V setting is conservative and works. I use something like a 0.1 to 0.68 ohm resister. It does not take much to drop 0.3 volts

Yes - I was going to mention this, but was going to discuss one item at a time.

The current draw for a 6dj8 is listed at 385 mA.
When I was tinkering with a 6dj8, using a voltage regulated supply for the heater,
When first switched on, the current jumped up past 700mA for the first second - yikes !
People have discussed current regulation for the heater elements,
but converting existing equipment is not practical.

However, for voltage regulated heaters, I'm wondering if a thermistor should be connected in series with the heater elements.
Maybe something like a CL-30 thermistor ?
.
 
Higher heater current at switch-on is to be expected. The valve is designed to cope with this. A series thermistor is only needed for long series heater chain (e.g. in AC/DC radios and TVs), so that the smaller valves don't overheat while the large valves are still warming up. To reduce the problem, some valves were made with heaters which warmed up at a known rate - sometimes indicated by an A suffix on the valve name. This problem does not arise with parallel voltage fed heaters.
 
...
However, for voltage regulated heaters, I'm wondering if a thermistor should be connected in series with the heater elements.
Maybe something like a CL-30 thermistor ?
.

Don't worry about it. All heaters have low resistance when cold draw more power then use less current when hot. The resistance depends on the temperature of the heater element. Light bulbs are the same way.

Why is it safe? Because it is self limiting, the heater is never over heated by the high current because just before it starts to over-heat the current goes down. It is a kind of feedback temperature regulation. The hotter the tube heater gets the less current it draws. So you get a current regulated system with just a transformer and nothing more.

What I typically do is place a 0.1 to 0.68 ohm resister in series with the transformer. Many transforms have to high of a heater voltage when used on 120VAC mains. I like to drop it to 6.0 volts with 120V input You can calculate using ohms law but that never works partly because resisters only come in a few standard values and the effect is not linear (tube heaters do not follow ohms law) I just try a few resisters, then wait 10 minutes and measure voltage

The series resisters do hold back the turn-on serge a little. But as I said this hardly matters

Your thermister idea could only work if you found a NTG thermister with the exact required specs, I doubt you'd find one. It would need very fast time constant and the EXACT resistance when hot (a 0.1 ohm error matters) And as I said above, this is not a problem you need to fix.

Back to tube testing: The only reason to have a tube tester is i you are in the business of selling used tubes and you want to assure your customers the tube is not dead. For testing your own tubes, a simple test circuit is easy to rig, you just place it in the amplifier. Back in the "day" in the 1950's the real use of tube testers was to sell tubes. You would show the customer that the tube tests "low" and then sell her (Almost always female) a new tube. Not that the low tube was really all that bad. The repair techs would simply swap in a known good tube and not bother with the meter.
 
Last edited:
I would like to have some method of measuring the state of a vintage twin triode.

Referring back to the Class A resistance coupled amp from the data sheet,
this is a common cathode amp with the cathode resistor bypassed.

I think the equation for Av is correct. Certainly, from an AC point of view, Rp and Rs are in parallel.

However, the problem from measuring gm from this circuit, is the wide range in a tube's plate resistance.

For example, my RCA data book says for a 6sn7, rp min and max is 6.7K to 7.7K ohms.

Unfortunately, rp dominates the other resistor values.

How about using a new Sovtek tube in this circuit and then comparing it to the vintage tube in question ?
.
 

Attachments

  • img_0876_2.jpg
    img_0876_2.jpg
    154.4 KB · Views: 267
Last edited:
Uunderhill said:
However, the problem from measuring gm from this circuit, is the wide range in a tube's plate resistance.
gm and plate resistance are related: Rp = mu / gm

Measure any two of these and you can calculate the third. You could measure Rp by varying Rs in your circuit and seeing how gain varies. Then, knowing Rp, you can calculate gm from your results.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.