JAN TESTING of Electron tubes
JAN = Joint Army Navy
Military tested tubes are strange items in the world of HiFi. As if they do not belong there, because whatever the army is doing, has nothing to do with listening to music. In this small article I will try to describe the advantage of JAN TESTED tubes. Perhaps I will convince the reader, but this will not convince the "crowd" because once something is believed, it is supposed to be that way and no other way. So use this knowledge in your advantage when you check those uninteresting products, that not everybody is shouting about. It is just this what JAN TESTED tubes are. These provide higher reliability, as this was the intention of JAN TESTING.
6SL7 Sylvania JAN TESTED / Order Number: Order Nr:114-087-09
I read this talk on a forum: (Link here)
'Also, I am unaware of any additional testing (torture test, expected life measurements, whatever) that went into the military version. As far as I know, they are the exact same tube, just labelled JG-6072/JG-6072A/JAN 6072A. GE produced tubes for a number of different clients and the military was no different; they just put whatever brand needed to be printed on the tube. Again, if someone has documented information and not just internet hearsay that counters this, please post and correct me'.Â
So this talker informs here, the military buys whatever they can get, and brand it 6072 because that's what they do. Now that is really silly. Even more confusing, people even write "Thanks... Good information!.... Wow... thank you!". Isn't that silly?! Read it better....the author writes there,"as far as he is aware", and "as far as he knows". So in fact, he knows nothing.
Here is more like this: JAN Specifications force 500 hours abusive testing on a certain number of samples from the same lot. Then, it requires a 90% survival rate of the tested samples, and after that comes a documented destruction of the abused tubes (date, quantity, operator), and the manufacturer must allow unannounced audits. These 90% untested tubes from this lot, have a certain mathematically proven failure rate, which can be calculated from the 10% that were abused (and destructed). In reverse, with that failure rate under excessive conditions, the probability on NOT having a failure (=reliability), during a specific number of hours, and normal use, can exactly be calculated with a special formula and data tables. To my personal opinion no other program can replace JAN testing, because it gives the exactly that probability in percent. From this may result such requirements, to replace all tubes of vital electronics, after a given number of hours, even when the tubes are all still testing strong. JAN Testing was introduced by the USA Army and the Navy, in order to deal with tube defects in a predictable way. This method is still used today, with semi conductors.
Let's start from the beginning: What is JAN Testing
When electron tubes found their way into army systems, they needed a system to define reliability in a defined, controlable way. The method which they designed was aiming at electron tubes, but later on this system was working so well, it was also used for semi conductors.
What is reliability in the first place?
This is a misused word. Do not confuse it with quality.
Quality means something has much better properties than average. Specific about Quality, is the missing metrics. It exists of course, and it's very real, but what means quality means to you, is not what it means to others. A person can have all qualities a partner is looking for, but how about his or her reliability? And how can you find that out? Not by looking at the qualities.
Also with products, Reliability is not quality. A low quality product can have high reliability. Or vice versa. A hammer can be made as a cheap, low quality product, or high quality product. Each of them you can use for 50 years, and have no defect with it. So reliability of each is the same. Even so, the quality hammer may loose it's quality properties (for instance said to be none-rusting, but it did rust).
Quality is sometimes described like: Say what you do, and do what you say. However the military doesn't take that for an answer, and wants numerical reliability.
Uptesting: Increasing the reliability of a product, AFTER production. This is done by screening out the unwanted products. Like you can increase the precision of a test instrument, just by screen out the ones with lower precision. So you change nothing to the production process, and yet improve your products. This is called uptesting. A word to remember.
+ + + +
At Hewlett Packard. I worked there for 17 years. I went to JAN Training courses, and they gave me this example: Suppose you need to purchase 100.000 drinking glasses each year, for use in airplanes. You want only such that do not break if the passenger makes it drop on the floor. Take standard glasses, and some will break, some will not. How will you verify that? Will you buy 100.000 glasses, drop them all one by one, and only use those that survive? Probably not! You want to do sample testing on glasses of different vendors. The brand that performs best, is your favorite. But hold on.... Do we want to test this ourself? Do we have a certified glass testing lab? No we don't! So we let the vendor do this, the way WE tell him, and he must document it. We have the option to show up at his production site by surprise, and they must show how they are documenting things that very moment. This gets close to JAN Testing already.
Now look at just one such a glass from an approved lot. There is high belief, if you drop it from 50cm, it will not break. Though this is not guaranteed, it's just the chance on breakage that is lower. From the test results you had before, this chance can even be seen, as you know for instance it was 1:15 for the best vendor, and 1:4 for the worst vendor. But you want more. You want better than 1:25.
Now that becomes difficult, because you can not write this required result as a "must". Then you get stuck in sample testing again. And most of all, any conflict would end stupid, because the vendor questions what you are doing, and you question what the vendor is doing. So now comes JAN TESTING. You describe all vendors how to do the testing, and they get simple and clear rules. Each time he must makes a lot of 1000 glasses, he must take 100 random pieces out. These 100 he must drop from 1 meter, not 50 cm as in real life. This is definitely too high, but some will still survive, and it's just THIS that you want to know. And no, you don't want to have those that have survived, as these may cracks now. So the lot which was used for testing, the factory can sell it as commercial products. The remaining lot (of 90%) these are the good ones, and it's so weird do say this, because they are the untested. Yet this works, and it works good.
+ + + +
Reliability means the PROBABILITY a product keeps working during a SPECIFIED period of time. Now isn't that what we all expect, and this a simple engineering parameter, called reliability. When it's puts into a number, it's expressed as MTBF (Mean Time Between Failures) So same as weight is specified in kilogram, reliability is expressed in MTBF.
With that number, engineers can calculate the chance on a product error during a specified number of use hours. With tubes, reliability is well recognized, as used tubes no matter how fine they test, have much less value as unused tubes. Prices are are also related to a specific brand, type, construction or date code. That is, because some people seem to know how a lot about this, and others no so much at all. Some believe Gurus, and some others only what they can verify. It is just here where JAN TESTING is a great a help.
With tubes, JAN TESTING exposes the tubes to an overload situation, above the datasheet limits. It is obvious this creates some defective percentage within some time. From a lot of 10.000 tubes 100 are taken, and these are overheated slightly, and run for 1000 hours that way. All you want now, is a pre defined survival percentage for this lot. It is obvious, you want only those 9900 tubes that are not tested by overloading them, but if the survival rate of those 100 tested tubes was good, you have better belief in this lot.
In a practical situation, this survival rate is defined, and also the overload situation is defined and the number of hours. The result of this, after some difficult math, is a parameter, called MTBF. (Mean Time Between Failures). Once you have the MTBF, you can use that number to do any calculation. Like what is the defect rate in percent, after 9000 hours, at 70% of maximum dissipation. There are formulas and tables for this in the military handbook.
Can you repeat those calculations? Unfortunately not. With JAN TESTED tubes the MTFB is known to the buyer, but it is not communicated externally. All we have is the code numbers on the box. Sure is however, these have a higher reliability than commercial tubes, and lots that did not pass, were sold normally as commercial tubes.
JAN means: Joint-Army-Navy. The army does not do the testing themself. Tests are done by tube manufacturers in house, and must be documented. The lots must be numbered, lot numbers on the boxes, and army inspectors will visit the production sites for inspections. So you always will see a lot number, the manufacturer code, the date code, and the actual test specification code. This leads a government inspector always to the original factory documentation for any failed tube.
So JAN doesn't say anything about initial values like transconductance or anything like that. For that there is a datasheet. JAN means the tube has a higher change to KEEP the datasheet values during it's specified lifetime. It is recognized this chance is never 100%, and the number MTBF can be used with formulas and tables, to make a forecast about this chance. I would say for buyers of NOS tubes it is just a good thought, knowing this chance is HIGHER with JAN tubes as with none JAN tubes.
Done so ever since, and JAN testing is still used for semiconductors as well.
You see, the good part is, the crowd doesn't know and doesn't understand this. So take your advantage :)