GENERAL CONSIDERATION ON TESTING.
- 1: Objective Standards:
The first and most important point: A tube under test can never meet a “real” standard to data book specification. That is the reason why a well functioning tube is said to be “on bogey”. Bogey is a phantasm- a ghost as in Casper the Ghost! You can come close in a high quality tube or in a low quality tube tested to very exact specifications. Take the idea of a positive benchmark or reference point out of your cognitive infrastructure. If you were to take a metal ruler from your desk, and leave it in the elements in Alaska in January it would be objectively shorter. If you would then fly that self same ruler to the Mojave Desert in August and leave it on a rock for an hour, it would be objectively longer. The current definition of a metre (the correct spelling incidentally) has since 1983 been “the length of the path travelled by light in a vacuum during a time interval of 1/299,792,458th of a second”. But we know that light affects magnetic objects, and the reverse, that magnetic objects deflect light. And what exactly is a second? The point here is that objective measures themselves are as problematic as the subjects it seeks to measure. For our purposes we only need to understand that the largest variable spoiling our metreology is heat. Just like with our metal ruler, every measurement is heat related.
The first use of the term “metre” came in 1675 from Italian scientist Tito Livio Burratini, in his treatise “Universal Measure” when he referred to “metro cattolico” directly translated as “catholic metre” or conceptually translated as “universal measure”. Thus the idea of a objective standard, endowed with truth, came into the vocabulary. You know where I am going with this. As long as you accept the notion that there are no fixed universal measurements endowed with truth, you will do okay. Truth is a metaphysical concept, it does not belong in our investigations. If you want truth read whatever sacred text floats your spiritual boat. I do, and so should you.
- 2: Positive or a negative?
We need to get this out of the way as quickly as possible.
Alternating Current does not have polarity, it moves back and forth.
Electricity, as conceived by Volta, Galvani, Romagnosi, Zantedeschi, Pacinotti, Faraday, Ohm, Ampere, Joule, Marconi all the way to Fermi assume cathodic potential and anodic potential. How that is expressed is irrelevant. Electricity has a direction of propulsive drive regardless of whether it’s a function of a chemical process or of a magnetic field. That direction is from the cathode to the anode and back unless directed to ground. As long as it was referred to as Galvanic Current there was no confusion.
The confusion arose because in a common lead acid electrochemical cell, electric currents are composed of positive hydrogen ions (protons) flowing in one direction, and negative sulfate ions flowing in the other. But, when a metal wire is connected across the two terminals of said cell, the source places an electric potential across the conductor. The moment contact is made the free electrons available within the conductor move towards the positive terminal under the influence of this field. The free electrons are therefore the charge carrier in a solid conductor. To summarize, in metallic solids, electric charge flows by means of electrons, from lower to higher potential. In gasses, ion gasses move in both directions, but absent a means to excite them they stay put, they have neither propulsion nor potential nor directionality unless connected to an external power source (unless you can split them...).
Electric current produces a magnetic field, and a magnetic field can be made to produce electricity through induction. When a loop of wire rotates in a magnetic field, the potential induced in it reverses with each half turn, generating an alternating current. By inserting a commutator, essentially a rotating switch, the connection at the windings is reversed and a direct current is produced.
A Direct Current can only move in one direction, and that is forward. Short of engaging in a semantic argument, positive is a synonym for forward, and negative for backward. So whoever coined the term Positive and Negative polarities of Direct Current was confused and is confusing us. The cathodic “moves forward” and the anodic receives and sends back. At this point somebody will point out that to charge a battery and reverse the chemical process, you need to feed it anodic current… That might be so, but its intended purpose is not to use electrical potential but rather to reverse a chemical process within the cell (battery). So unless you are using the potential to reverse a chemical process (to re-charge the battery) so as to produce heat, current only moves in one direction, from cathode to anode.
So we are stuck with two very confusing terms: Positive + and Negative -.
When discussing thermionic valves, current moves only in one direction: from the cathode to the anode, that is from Negative to Positive either completing the circuit back to origin, or to ground.
It would have been easier in Direct Current terminology to have indicated the forward potential as K and the return as A, unfortunately we are stuck with Negative and Positive. And if I had known that years ago I would not have placed that diode in the wrong branch of the circuit when connecting the solar panel to the battery….
In Alternating Current you have Live, Neutral and Earth Ground. And while the Live might appear the more vivacious it would appear that the “potential” is between the Neutral and the ground potential. But that discussion does not interest us.
2.3: Testing Considerations.
Below you will find some of the concepts that you need to employ in order to test thermionic tubes. No concept stands in isolation. All the concepts are inter related, some causing failure by and of themselves while others only become significant in relation to other issues. Again we remind the reader that there are no objective thresholds or triggers but rather a confluence of factors shift the output of vacuum tubes into dangerous territory, between diode and conductor.
2.4: Vacuum Tube Segregation.
Tube Segregation: All electronic devices are born from the same mother. It does not matter if its solid state or filaments in a vacuum. Take for example a computer processor (CPU)- they all come off the same fabrication line. Then they are tested as silicon die so as to meet minimum parameters. If they pass, the Known Good Die then get sent to packaging where the silicon is bonded to an alluminum wire frame with gold wires. After which its encapsulated in a plastic/resin case with tin leads protruding. Its only at that point that the “chip” gets parametrically tested to exacting standards. Speed grading follows: assuming a Mean Time Between Failures (MTBF) set at X hours times Y speed, chips then get segregated by their lifetime expectation. If MTBF is set for 50,000 hours the determination is made as to what is the maximum speed it should be allowed to operate at. There was a time when CPUs could be overclocked- they could be made to run faster but only for more limited time periods, obviously a practice that manufacturers offering consumer warranties has to discontinue.
2.5: Vacuum Tube Thermal Shock.
Another issue affecting any electronic device was and is “thermal shock”. A silicon die attached via gold wires to an aluminum lead frame attached to tin leads inserted into a copper socket is one of thermal mismatch. Metals expand and contract with differing temperatures. The greater the number of repeated cold-hot cycles (thermal cycling in life time expectation tests) the greater the risk of connections coming loose. “Burn in” really only tries to prove the quality of those connections. If a short or break in the integrated circuit will occur you want it to occur in the first couple of hours, and indeed most electronic failures occur when the equipment is new.
2.6: Vacuum Tube Galvanic Corrosion.
All metals have dissimilar electrode potentials. This goes back to Volta’s Electrochemical Series. Once two dissimilar metals come into contact one metal acts as anode and the other as cathode. The electrical potential between the dissimilar elements is what causes the nobler metal (K) to steal from the less noble (A) metal, in the so called galvanic couple. Obviously in a vacuum there is no direct electrical path for this to occur. But in a physical tube there are connections between the pins and the tube elements within the enclosure. I will willfully ignore the issue of what resistance those joints offer. Suffice it to say that galvanic corrosion is rarely a contributing factor to failure, other issues pre-empting it, nonetheless in a hundred year old tube it might be an issue. Corrosion on pins is determined by galvanic corrosion. Tubes that were left for years if not decades, inserted in equipment will obviously show signs of pitting, this being a consequence of the copper contacts on the socket robbing the tube pins. Older tubes stored in their original boxes are instead usually oxidized by the acidity of the cardboard. This is sometime the reason why pins on miniature tubes break. Yes, oxidized pins do offer some resistance. If you are obsessed with the issue, buy tubes with gold plated pins.
2.7: Voltage and Heat in a Vacuum Tube.
A thermionic valve is no different from the cpu example given previously. But obviously they present different problems which we must address. First and foremost substitute speed with voltage. Higher voltage results in higher resistance which in turn generates heat. Higher electron speed is heat. The higher the heat the lower the expected MTBF. Running a high cathode voltage counter balanced by a higher grid voltage might produce the same exact emission as running a low cathodic voltage counter balanced by a lower grid voltage, but the tube will run way hotter, and consequently last less. It might give you a faster response, true, but not for very long.
What gross liberties we take in the name of commerce. Okay, if we wanted to do this subject some justice we would have to fill volumes. Let me just say that a direct current circuit, voltage and current are independent of time. If a capacitor or inductor is added to a DC circuit its technically not a DC circuit. What we are achieving in a DC circuit is a steady state. By using a constant current source connected to a capacitor and a constant voltage source connected to an inductor we are using a non DC solution to achieve our goals, namely to power the circuit by a DC Source by non DC means. Regardless, a tube operates in a time domain by virtue of the circuit its in, even though we are not concerning our self directly with that issue, other than to point out the relationship between intensity and mean time between failure. But in so far as a vacuum tube has resistance and capacitance, it has a time component.
As a practical matter, if you increase the voltage on the DC B Supply from 1 to 100V, you will get a nearly instantaneous result. If you decrease the voltage from 200V to 100V you will have to wait for the excess capacitance to dissipate before you get the correct current (I) output. That of course is dissipated either to ground or via thermal transfer.
Similarly, when talking about the filament it can be said that the higher the voltage across the heater element, the hotter it will be, the higher the emission. Similarly in a tube with a separate filament and cathode, the hotter the filament gets the better the ability of the cathode to emit. But one must be careful with this because outside of the bogey values, at a higher voltage, a filament will emit separately from the cathode. What it emits of course interferes with the job of the cathode, grids and anode. Grossly it can be referred to as EMI or electromagnetic interference. That results in noise or distortion. We will get back to this when we discuss limiting values.
So its absolutely critical to set the voltage across the filament very precisely. Moreover, precise in our case means “consistent and repeatable”.
An AC circuit is of course in a time domain. Which is why we say 120V AC 60 Hertz (60 cycles per second). If you buy a tube tester built in the 50’s it will really be set for 115V operation with the expectation of real life mains current at 112-114V. But to use an example, the “voltage at the meter” in our building is 244.8V AC two phase. Split into separate phases you get your standard 120V AC mains current. Mains voltage drops as a function of distance from the transformer. But in our case we are six feet from the mains transformer so our “120V AC” line delivers a strong and steady 122.4V AC. Some simple math- 122.4 – 114 = 8.4 / 115 = 0.073 or simply stated the voltage is 7% off design spec, should we chose to use an AC Mains tube tester from the 50’s... The consequence in a perfectly linear world (which tubes are not) is that in my 1950’s tube tester the heater filament is 7% hotter and the emission more than 7% higher than “spec”. If the secondaries on the B supply are 7% higher so will the emission. Its not that the test is invalid. Its simply that whatever measurement is given will be different on similar setups on differing mains current. Repeatability between two identical pieces of equipment in two separate locations is the hardest issue to resolve. Whatever bogey value is given will never be realized on mains dependent equipment. Similarly a German tester designed for 220V will be off 4.5% if plugged into an English 230V outlet. Which is why we are trying to go the DC route.
2.8: Mechanical Distances Between Differing Elements within a Vacuum Tube.
Of equal importance is the distance between the heater filament and the emitting cathode. The closer the heater the greater the heat transfer, the higher the emission. Too far and it will not emit much and too close and it will contact the cathode and “short” the circuit its in, mostly with catastrophic results short of the fuse blowing. As a gross generalization the “stronger” the tube the higher the probability of a manufacturing defect. I am not usually worried about a “weak” tube- strong tubes are the problem and they must be immediately be disposed of. Similarly the distance between the cathode and the anode is of primary importance. The further away the lower the emission. The closer they are together the higher the probability that either the cathode or the anode will contact with the controlling filament, namely the grid(s). If a heater to cathode “short” is bad so is either the cathode or the anode contacting the grid. Incidentally, if you place a tube under test on its side there is a good probability that the heater filament will sag and contact the cathode.
As we have previously mentioned, the relationships are not linear. How they get assessed is not as important as understanding that those emission curves are actually exponential. I will spare you the mathematics. What is important to understand is that a tube when presented with too high a voltage will eventually go into what is called “runaway emission”- the tube will put out increasingly higher emission until the elements saturate and the tube shuts down. It might “come back alive” when it cools down, but in my experience a tube that survives that process will soon be proven defective. Of course the grid can control higher voltages, but even there eventually the amount of grid voltage necessary will be so high that it will itself over heat the tube and cause much damage. I recently, and to my horror, read a whole chapter on testing tubes with a “home made rig” that suggested 400V at the cathode with the Grid at -50V “only for a few seconds so as not to destroy the tube”.
The only thing that concerns us here is that a mechanically defective tube is one where the elements are too far apart which means lower emission, or too close which results in too high an emission in relation to the controlling elements (say grid to cathode resulting in run away emission), and in the case of the heater, a temperature coefficient by virtue of proximity exceeding the operating temperature of the cathode directly and other elements indirectly. Soviet tubes are notorious for presenting these mechanical deviations. Arcing is the usual result, with catastrophic result on the circuit it resides in. “Popping” in audio equipment usually results in transformer failure.
2.9: Runaway Emission in Vacuum Tubes.
We cannot postpone this any further. Lets quickly define runaway emission. We are using the potential of the power supply, however defined. In the case of a transformer, that transformer can only supply X amount of potential. A thermionic valve will control how much of that potential you are using in a time variable. You can use that potential to move a magnet in a speaker, or to charge a capacitor or to heat a filament. If you allow “unlimited” potential to be realized you will throw that transformer itself into runaway potential, the upper limit being what the mains current can supply. Either it will “saturate” and the magnetic field collapse, or you will bring its temperature beyond its mechanical physical limits. Most power values are instant on instant off. They are not continuous use values. So there is the issue of exceeding the limit and the issue of duration. A tube in a runaway condition is the equivalent of shorting a battery. Take a standard household cell and short it with a wire and the wire will burn red hot, drop a wrench between the leads on a car battery and you will get a small explosion. Runaway emission is thus the process of transforming a diode into a short circuit. Expressed differently it’s the inability of the grid to control the cathode emission resulting in an uncontrolled power discharge.
2.10: Consistency of Emission in Vacuum Tubes.
A further testing parameter which is related to run away emission is the consistency of the result. I have placed tubes in old testers only to see the initial result low, followed by a slow increase in the measured parameter all the way to an above bogey value. Those are the most dangerous tubes. Whatever emission there is, is variable. A high quality tube is one that given the proper temperature (2-3minutes with heaters live -be patient- and then apply voltage) goes immediately to a set value and stays there. At 45V all tubes are “stable”. At 300V or whatever the maximum value given is, the tube is very unstable. Exponential amounts of emission require exponential amounts of grid voltage to control. The problem here is that if you push the grids very hard they will not only over heat but they will indeed start emitting electrons themselves. 2V DC at the Grid is not really going to cause any measurable emission. 10V DC will cause emission. If you see any current (Ig on C Supply) on the grid supply mA meter you have exceeded the tube’s operating parameters. Try taking a 6.3V heater supply in an output tube to 12.6V and you can clearly measure 20-30mA worth of emission on the B Supply mA meter. As a matter of course once you have kept the tube there for 2 minutes, if it survives, you can then wait for it to cool down and then proceed to throw the tube in the garbage. Short version: trying to work out how much emission a heater can output is a destructive test. Running too high a voltage on any element is a destructive test.
2.11: Non Destructive VS Destructive Testing in a Vacuum Tube.
I bring it up because you must differentiate between survivable tests and destructive tests. You should be able to test your tube without damaging it. You have no reason to test “maximum values”. They yield no useful information. At maximum cathode voltage, regardless of how much grid voltage you apply the tube is unstable, prone to runaway emission and most likely to over heat and damage the connections between the different materials. Clearly if the distance between the cathode and the anode is of vital importance, you could remove the glass enclosure and measure the distance with your trusted micrometer. But what good would that do you? The tube needs to survive the test in order for the test to be valid. Otherwise its not testing, its failure analysis or reverse engineering- and that is a separate profession.
To test an unknown tube you picked up at a flee market in your high end audio rig is also an invitation to disaster. With a good tester even a clean short could potentially do damage. In our case we used a Constant Voltage to Constant Current design. When a short is found the current draw will reach its maximum after which the voltage will drop until current draw drops. Its not fool proof and the meter will not like it one little bit. But if caught in time no harm done. The point here is that testing a known defective tube is also a bad idea. You don’t need to know how bad a tube is. If at 60V at the cathode the emission is unstable, or too high or too low, it will not get any better at 95V. If a heater is supposed to draw 1.5A and after 15 seconds its still at 2.6A, you are a time component away from a full blow short circuit. If you are bringing a tube up to temperature and have zero volts at the cathode and the tube starts emitting (mA meter on B supply is registering current), you have a short. If your grid supply is showing current on the meter, its shorting. Again, a short is any value outside of the element parameters.
2.12: Tube to Tube Comparisons.
Another consideration at test is the issue of tube to tube comparisons. As we mentioned previously all tubes come out the same oven. If you test a statistically relevant number of tubes you should be able to categorize those tubes. Again we are faced with a bogey if you assume a parametric data book specification. But if you test 1000 tubes you should be able to get some number of tubes that match. Unfortunately considering the nature of the manufacturing technology that is a pretty random event. You have six independent variables in this equation, the heater, the cathode, three grids, the anode and all the internal connections. Some manufacturers are better than others at process control. Some manufacturers test better than others. Fortunately even in the bad old days manufacturers segregated out the tubes into different categories and marked them accordingly. If we need to make a statement on this regard, I would have to say that the only thing that matters is a consistent current output at given non stressing voltage inputs. If you know what voltage your equipment is supplying the cathode and grid, test it at that value. You could say, the equipment requires 70mA at the anode and work backwards from there, but that is rarely the case. In some circuits the cathode voltage is a given and the grid is the variable (bias adjust). A “matched pair” is thus an oxymoron unless you know the parameters, and know if those parameters are relevant to your application. Knowing what the tube will do at twice its optimal voltage at the cathode is also not relevant. Nor is it relevant to know what a high power output tube does at 45V at the cathode.
- 13: Life Expectancy of a Vacuum Tube
Life expectancy is another tricky subject. The military was a large user of tubes back in the fifties. The air force for example erroneously swapped out all the tubes (say in a bomber) every 500 hours. With disastrous consequences. The highest incidence of failure is always in the first 50 hours, after which the tube stabilizes and its pretty much consistent up to its upper life limits. In the case of “consumer” tubes that could be 2-4000 hours in the case of higher quality tubes 10,000 hours. Replacing tubes every 500 hours meant that most of those tubes were actually perfectly operational with a good fifteen hundred hours left on them, while those used to replace them had a very high probability of failing in the first fifty hours. Like the proverbial dog chasing its tail, they could never “get ahead of the curve”. That is a failure of statistical analysis. Fortunately for us it means a huge quantity of very high quality tubes with “500 hours” on them. Not quite New Old Stock but darn close. And in any event most of those tubes ended up in the packaging the new tubes arrived in guaranteeing physical integrity (and a prosperous grey market). Personally those Mil-Spec JAN (Joint Army Navy) are my favorites- they were usually called into service in high frequency communication applications, thus were sharp and fast and consistently on bogey. Test a lot of 100 tubes and 60 would be identical. Test 100 consumer tubes and the results are all over the place. Of course “black plate RCA” from the fifties and sixties test badly, as in all over the place- but sound really really good in audio applications. Ultimately, the two aforementioned examples were designed for specific applications and the empirical results are consistent with the intended purpose.
The issue of Mil Spec tubes is brought up for a simple reason other than covering the issue of when failures will occur. No tester can be said accurate or predictable. A parametric databook specification requires precise parameters. Nobody can know in what application the tube will end up. Which leads us to the difference between parametric testing and “functional testing”. Ultimately functional testing is the best even though most electrical engineers will cringe at the very idea. Simply put, place the tube in your application and apply a signal and see what comes out the other end. If setting the bias on an electric guitar amplifier, feeding a signal and looking at the sine wave in an oscilloscope is still the best method. But here comes he who shall not be mentioned. Now place four other tubes in that circuit. Now place four small signal tubes in the circuit before the four output tubes. What is what? No way to tell.
Let me illustrate a further concept with a hypothetical circuit. The design has 48 resistors. Your resistors have a 2% failure rate or a 20% probability of being off spec- AKA 20% tolerance parts- which actually are two separate concepts if you test loosely, but just because the resistor resists does not mean its on spec, and off spec might as well be considered a failure in its particular application (like a piece of medical equipment in an emergency room). You build 1000 boards. That equals 48,000 resistors. Assuming an average distribution that means 48k X 0.02 = 960 boards that parametrically at least are defective. 980/1000 means that for all intensive purposes all the boards under test are defective. Now take your board with 48 resistors and add 8 capacitors with also a 2% defect rate. And then add four diode bridges giving you another 16 components for a grand total of 48+8+16= 72 components. 72,000 X .02 = 1440 out of 1000 boards will be “defective”. Now throw in 8 small signal and 8 output tubes with three grids each and see what the math looks like. So the notion of a “bad tube” is a notion loaded with prejudice. Having said that a high quality piece of audio gear will probably use 1% tolerance parts, and a failure rate of 1/1000 on components is achievable (even 1/5000 with a relative cost of 10X). But it does not take away from the mathematics of statistical control: the more components in a circuit, the higher the probability of circuit failure, whether random, consistent or occasional. Ultimately when testing components the issue of specification tolerance must be brought up. The more components in a circuit, the lower the tolerance must be, regardless of application. A two tube audio circuit with one capacitor and a couple of resistors will work, even very well, regardless of the tolerance of the parts. A circuit with 500 components (say a high end pre amplifier) will not. And lastly there is the matter of expectation of quality, a high subjective issue.
Fortunately for us most tubes are generally acceptable in our application. But will a Mil Spec JAN tube designed for a radar receiver work in our amplifier? Allot of tubes the military used were “sharp cut off pentodes”. What that means is that given a certain grid voltage emission would cut off immediately instead of following your standard descending parabola. An audio amplifier would optimally present the following scenario: an attack curve of a certain duration, a dwell period of a certain duration, and a decay of a certain duration. A good amp closely follows what real notes on an instrument sound like. Now take a perfectly good sharp cut off tube and place it in your audio application. What happened to the decay? If it’s a very analytical amp the notes might be truncated. If it’s a fat and sloppy and very warm sounding amp, a sharp cut off might just give you the right parabola. Now lets take the example of an output tube that instead of requiring the bogey value of -13.9V on Grid 1 to cut emission instead requires 30Volts. And now lets say that your C supply only puts out 24 Volts. Do you have a bad tube? Or a good tube in the wrong application? How about we state that the amp is used to drive an electric bass guitar? The best sound coming out of said beast is when the tube goes into runaway emission, distortion is at its inconceivable highest, and the amp’s speaker is wildly trashing about. In such an application an “on bogey” tube would require you feed the cathode a HUGE amount of voltage in order to achieve your stated goal of runaway emission. Practically, that would mean raising the volume to ear splitting levels. On the other hand a tube with an off bogey grid would only require modest K-A voltage and still be able to go into “controlled runaway emission”. If you are buying used tubes, or “Nused” JAN tubes or tubes claimed to be NOS you still have to consider in what circuit they resided in, however briefly. The question in regard to longevity would have to be “would you buy low hours EL34s from somebody who plays in a rock band?”. The answer is no.
That brings up the last point. If you do not know in what circuit the tube is going in, you cannot make a statistical prediction as to how long that tube will stay on bogey. Some manufacturers of tube amplifiers crank up the voltage at the cathode and at the grid to compensate for an under powered design. You know a-priori that the amplifier will burn through its output tube real fast. Other amplifiers instead run all their voltages low- but an exponential increase in equipment and tube longevity. Circuits with 1% tolerance parts will be kindler to their tubes, assuming they were properly designed, by virtue of the fact that the tubes will never go into runaway emission because of incorrect values at the heater, at the cathode and at the grid.
Longevity is ultimately an in application variable.
- 14: Tubes that are Fit for Use
The above paragraphs of course raise the issue of tube selection in a specific application. Good tube resellers will simply ask you “in what application?”. And then answer “For that amp I have gotten good results with this combination of tubes…”. Or ask “at what voltage do you like to run the cathode/grid?”. But then again, equipment manufacturers never give you that information and instead private label their own tubes and sell them at a premium. Which of course brings us to some further considerations on this last point.
We could bring us some further issues, namely of the materials used in the construction of the tube elements and of the metallurgy and the chemistry. The problem here is that we cannot in any manner test for the above. There is no practical way to do so without breaking the tube apart. You can do failure analysis but that is destructive analysis, and as we have demonstrated above, the why should not concern us too much.
There are many things that cannot be solved at test. This is one of them. The answer is not one of test but rather one of manufacturing process control. You can eliminate these problems by careful manufacturing, but once the good is manufactured, testing for them becomes counter logical. You can test your materials before you manufacture but not afterwards short of destroying the good produced. Vacuum tubes and most electronic components fall within this category. Material selection has of course vastly improved in the last hundred years, but the technology employed for manufacturing tubes has not greatly been changed. Its still the same tooling used forty years ago in many cases. And the same manufacturing process.
Back to statistical control. If it costs you 10$ to manufacture a tube with a manufacturing yield of 80% or 5$ to manufacture the same tube with a yield of 60% which one would you build? Lets do some math on a 1000 unit run:
Manufacturing Company X: 10 X 1000 = 10,000$; 1000 X 0.8 = 800 Units; 10,000/800 = 12.50$ per unit.
Manufacturing Company Y: 5 X 1000 = 5,000$; 1000 x 0.6 = 600 units; 5000 /600 = 8.33$ per unit.
What is important for our purposes is that they burn in and test every tube produced. Problem is what does the factory do with the “borderline” tubes? What if they re-sell them as “off spec” “B Grade” parts with no markings? What if somebody else buys them and counterfeits a known brand trade mark on them? They look identical. A Trade Mark is not a manufacturing guarantee strictly. A trade mark owner might specify materials and test, consequently what you are buying is the brand equity invested over time in the trade mark. If counterfeits are available it’s because manufacturers release those components. Its not that those tubes are “bad”- its simply that they are tested to different parameters or for different applications. One needs to understand that two physically identical tubes from the same factory could be two completely different tubes. Sometimes the tubes are physically identical having been built on the same tooling, but manufactured with completely different materials. There is no way to know that. What is important to understand here is that a “product” is only as good as its brand name. A trademark is a quality mark for a testing methodology, a factory is just a place where physical objects are assembled. When manufacturing was done in house, one had total control over every variable. Nowadays all “brands” have is the ability to specify what they want in a purchase order. What they get could be something entirely different. Quality considerations are therefore batch specific. The date code (expressed by week and year as in 5212) is sometimes related but there is no objective way to know what the factory did with the balance of the off bogey tubes.
Nonetheless it would be fair to say that a tube built with higher quality materials, with a higher quality process will ultimately be a better tube in so far as it has greater longevity. It might even sound better, empirically. Where the line is though is murky, and worst still, the information upon which to make that decision, is usually unavailable. Thus people buy brands.
2.15: Gas Tests and Gassy Tubes.
If you place two metallic parts in close proximity to each other and connect each one to one of the two battery terminals you will get current that will jump the gap. That is what the spark plug in your car engine does. The only way to mitigate this problem is to expel all the “air” and whatever else is present in the environment as a result of the manufacturing process. Achieving a perfect vacuum is hard. Achieving a perfect seal is hard. How well they burn off the remaining impurities in the getters is hard. Any impurity will eventually result in a spark. An electrical arc between two elements in a tube is a short. And a short can do massive damage in the equipment its placed in.
Lets look at the “gassy tube” issue by taking a small detour into particle physics. Once an applied electric field reaches the breakdown value, freed electrons reach sufficient speed to create additional free electrons that in turn collide and ionize neutral atoms resulting in a process called avalanche breakdown. The breakdown process forms a plasma that contains enough mobile electrons and positive ions to make it a conductor. That results is a short- physically demonstrated by the creation of an arc, spark or lightening event. Field electron emission occurs when the electric field at the surface of the metal is high enough to cause tunneling, which results in the ejection of free electrons from the metal into the vacuum. Small electron emitting regions can form rapidly on the cathode and anode (both metal surfaces) when they are subjected to a high electrical field. If an incandescent state is reached on either the cathode or anode an avalanche breakdown occurs. We can assume that is a terminal condition in a vacuum tube. In short: high voltages will destroy a tube
All old tube testers have a “Gas Test”. All a gas test does is reduce the voltage at the heater by whatever amount you decide (by selecting the proper resistor) and seeing by how much emission drops all other variables being fixed. A better tube will maintain a consistent emission profile. A tube with excessive hours on it will produce less emission. It’s a valid test that has nothing to do with “Gas” since you cannot test the composition of the gas or the amount. But if gas is present in the tube, the circuit will short either with a bang or with a series of bangs. If you get an electrical corona around filaments that you can actually see, the tube has lost its vacuum. Its about to short regardless of how well it tests. As a matter of fact if you see a corona in the tube do not even insert it in the tester- throw it out, you don’t need to know what parameter is failing or has failed. Its useless knowledge. Its no longer a diode, it’s a short. Someone will raise the fact that the military used gas filled tubes. And at the first opportunity they gave Fairchild a gazillion dollars to deliver them semiconductors. Tubes that are purposefully designed to go bang don’t last too long. They were designed as triggers; once the appropriate temperature was reached, and a specific voltage was achieved within the corona, the tube would short and release instantly the full potential of the supply. They were essentially remotely triggered “instant on” devices.
The confusing bit is that there actually is no correlation between “gas present” and “diminished emission in a spent thermionic valve” other than for the coincidental fact that old tubes happen to be gassy. But the tube can be old and spent and not be gassy. So even if there is a correlation there is no cause and effect. Whatever emitting materials were used might have lost physical integrity, or spacing within the emitting design, or the filament might have dimmed down- there are a thousand reasons why older tubes cannot sustain emission at a given cathode voltage.
What was once assumed was that as the elements within a tube age with use, any remaining gasses contained in the element coatings (not burnt off or absorbed by the getters) would slowly be released. In fact the coating might not contain any gasses. The coatings could just as well just turn to ash and fall to the bottom of the tube, and the lost emitting coating being responsible for the lost emission. The only studies I have seen are fifty years old. The statistical methodologies could have been entirely wrong.
In any event, whether gassy or not, a spent tube can be correlated with the following characteristic. Lower the heater voltage by 5%, and if the emission drops precipitously, the tube is spent (whether old or gassy irrespective). Remember that emission is heat dependent, which means for the test to be valid you must allow the tube to cool down, after all the filament is not there to illuminate but rather to heat. Military specification MIL-STD-1311C calls for a 10% reduction in heater voltage. That might be excessive. In any event many old military tube testers usually have multiple “gas tests” which means they preferred different values depending on the tube.
Yes, there is a way to test for gas composition by correlating the current curve to the cathode temperature and voltage, but that is not what the tube was designed to do.
- 16: Tube Heater Variables
Lumen output: A heater filament is no different from the filament inside your average lightbulb. If you look up average lumen output you will get two figures. The first is the amount of light it puts out when new and the second is the amount it will put out in the last 40% of its lifespan. This is most evident with neon light bulbs which towards the end of their life puts out substantially less light. The thing to understand is that the instrument will tell you how many watts it consumes expressed as V DC and Amps. But just like with a light bulb that is labeled 75W, those 75 Watts will not tell you how much light or heat to expect. In our application though the principals are not dissimilar. Brand new the heater filament in a tube will put out 110%, within a couple of hundred hours it will reach its intended parametric target and will eventually settle on 95% efficiency- some more some less. When tubes were used in battery applications current draw was an ongoing concern. The solution to lowering the consumption was to place the heater filament as close as practical to the cathode. Any problem with the filament or cathode would consequently result in a heater to cathode short. Just some observations. Big high draw amplifier tubes like a 6550 were not designed for battery operation. Small pentodes were. Take greater care when testing smaller tubes- more things can go wrong. A 6550 will over the course of 2-3 minutes go into runaway emission and hopefully blow the fuse as the transformer is forced to draw more and more power from the mains. A small triode like a 12AX7 will instead instantly “pop” when turned on- and the results are often more catastrophic to the circuit its in.
If you have an amp with 8 output tubes and one is proven defective, there is no reason to replace all of them. Even though the newest tube has higher output at the anode, within a couple hundred hours the heater will settle somewhere in the 95% heat output range and stay there for 2-3000 hours. So when mixing old and new tubes discount the emission in the new ones by 10%. What matters is its performance over 3000 hours not its performance when new.
Another problem is the fact that in most amps the heater connections are serial, and regardless of what some electrical engineers claim the first tube will draw more current than the last one. Again that brings us back to our so called “gas test”. If you reduce the voltage at the heater by half a volt, how much does the output decrease? Does the left channel on your amp put out less volume? It could be a simple matter of not enough current at the heater. Likewise a brand new tube with heaters tied to the AC mains might just tolerate a 118V line but what if you are a half mile from the transformer and that main gets browner and browner? What if the tube has 4000 hours on it?
Heater filaments are very sensitive to thermal and voltage shock, smaller tubes much more so. The older they get, the more prone they are to fail when turned on. That is the same reason why light bulbs are more like to fail when you hit the switch rather than when on. That is why some fancy gear upon being turned on goes into standby mode and slowly brings thing up to temperature before allowing the user to pass a signal.
The military usually specified beefed up heater filaments. Essentially like buying a 7000 hour light bulb instead of a 1200 hour one. The reason they did this was because of vibration. That is application dependent, and in most cases we could ignore it. You could test for it by tapping the tube with a pencil. That is actually not a bad idea if the tube will sit within a guitar amplifier that also incorporates within the enclosure a speaker. That is the worse case situation for a tube. Premature failure is a given, which is why manufacturers give such short warranties on replacement tubes. If you ask your self why do manufacturers not offer application specific tubes like “guitar amp tubes” its because they are still using the original tooling from the fifties and sixties. But eventually that too will wear out and will have to be replaced.
In most cases the filament was designed to outlive the tube’s usefulness.
Smaller tubes are more prone to catastrophic failure (detached filaments) than larger tubes. The good news there is that once they do, they don’t come on and thus cannot be resold “Our mutual conductance recently calibrated Woolworth’s tester shows them in Green Good Tube section”.
2.17: Limiting Values on Tubes.
Limiting Value: If you look carefully at any specification sheet, you should seek to find what the limiting values are. They are sometimes expressed as Maximum Ratings. The problem here is that if you look at the maximum cathode value it might say Vk = 300. That does not mean that you should run it at that level and “play around”. You will fry the tube under test. The more important values are current at the anode (Ia) and voltage at the grid (Vg). What you need to make sure of is that, regardless of maximum voltage on the cathode, the current maximum at the anode is not exceeded, regardless of what voltage you throw at the cathode counter balanced by the grid.
If you are for example testing a 12AU7 with a maximum emission of 20mA, you should never exceed that value. If you are testing with 0 V at the grid (Vg = 0) that might in practice limit you to 150V at the cathode (Vk = 150). If on the other hand your fixed grid value, or your maximum grid value is indicated as 8.7V (Vg = 8.7) you might be able to bring the tube up to 300V without exceeding your maximum emission value (Ia = mA).
Regardless of what voltages you select never exceed the current value at the anode, and never exceed the voltage maximum value at the grid.
In manufacturing thermionic valves this constant is expressed in terms of the current per square centimeter of cathode area. Unless you change the metallurgy or the chemistry, you cannot exceed that limit in that area. If you do exceed that limit physical damage to the cathode surface will result. Likewise there is a limit to what any of the grids can accomplish. If you exceed those limits, you will start to burn out the filaments.
A grid, instead of slowing down or limiting emission by counteracting the cathode, can also be run positively (anodically) and be made to attract emission. The problem here is that a maximum rating for a cathodic current is substantially higher than for an anodic current. Some of the better specification sheets might even spell it out that you cannot run the grids anodically by indicating Max Vg = -20V, Min Vg = 0, or in the case of some tubes other tubes Max Vg2 = -300 Min Vg2 = 100.
The last issue is a practical one, if you want to test at what point the grid or filament starts emitting enough current to be measured at the cathode, you will reach the point where you have damaged grid and filament.
A word of advice: NEVER use maximum values- you will fry the tube. Maximum values are instant on instant off parameters. Use the lower end of the range.
- 18: AC vs DC Tube Tests:
The big problem. Do you test a tube with alternating current or with direct current? Again, no clean answer. You could take a Variac and run a precise “mains current” to your test rig and get proper repeatable results. The problem is this: all AC based testers are differential testers. They take a measurement then a second measurement and compare the two. They are essentially in circuit designs. If a circuit design is in play, its no longer a parametric test, it’s a functional in circuit test, the test result only as good as the circuit design itself. You are making a value judgment as to the circuit itself. In a DC tester you are applying precise amount of variables to the tube under test. Other than the resistance of the wires involved over very short distances there is nothing to qualify the result. Don’t get me wrong, the mA meter is a very complex circuit, but with minimal impendence and resistance, 100mA in is 100mA on the LED display. Similarly the voltage output of the three supplies are all complex circuits. But then again, if the meter says 12.6V how you arrive at that value is irrelevant provided it can output that value precisely, more to the point consistently. To stretch this point to exasperation we even provided a separate power supply for the main meters, not even the 100mA necessary to light the 200mA meter is connected to the B Supply. Less interference, less draw and a separate path to ground ensures an accurate measurement. Again, once the meter reads 12.6V DC how you arrived at outputting that value is no longer an in circuit discussion. Parametric testing is thus a static DC measurement.
2.19: Rotary Switches in Tube testers.
Why no rotary switches? It would have been allot easier to just place six rotary switches from the output of the three primary supplies. Then the user could have just selected what pin to send that voltage to. All vintage tube testers have such switches. The problem here is that works only up to 60V DC, and indeed all commercially available tube testers run at a very low voltage (usually under 45V). After that threshold rotary switches simply stop holding current. Granted, there are Mil-Spec switches that are truly isolated, but they are prohibitively expensive and hard to consistently source. We thought it preferable to put the money into the power supplies. What happens above 60V DC is that the current jumps to adjacent pins. At 300V DC one can observe with a multimeter 60V DC across all nine pins. Try it your self, run 300V to the input pin on a rotary switch, set the switch to output to pin 1, and then test all the remaining pins. Then repeat with increasingly cheaper switches… you will marvel at how well fiber glass transmits electricity. Plastic is a little better and porcelain better yet. But as with a spark plug, a 0.1” (2.54mm) gap between two leads is a walk in the park for a 300V 500mA power supply. However good the switch is it has resistance. Quite allot of it actually. A final observation: its impossible to throw a tube into runaway emission at 45V. What would appear as a “good tube” on your vintage tester will look precariously dangerous at 200V. In fact most “audio” tubes are unstable at that voltage. And if they can be made to output greater than 200mA at the anode its only for very very short periods of time. Not enough time to take a visual measurement. Older tube testers with rotary switches are 45V testers with a differential value taken via a resistor.
2.20: Vacuum Tube Shorts Tests.
In a thermionic valve, absent physical damage, there is no such thing as a short. Its not a contact on, contact off proposition but rather a question of exceeding an emission value. The term as generally understood in the context of tubes has often given the appearance of indicating an unwanted electrical connection between two parts of a tube. It has been conceptually used as if that was a physical connection as in a light switch completing a circuit, which it is not. The important thing to understand is that two elements don’t need to touch in order for them to “short the circuit”. But shorting the circuit does not mean you have a short within the tube. You need to keep in mind that all elements within a tube can emit emission given the right temperature and voltage potential. Consequently a short in a tube does not exist. What does exist is thermionic emission that is outside of the working parameters. Thus runaway emission is a short proper in both the tube or the circuit its in.
This is of course the reason why some tubes do not short when cold, but do so when they reach or exceed their operating temperature. Emission is temperature dependent. Runaway emission is temperature dependent.
Of course if any one element within a tube physically breaks, it might just contact another. Heaters are 10,000 hour filaments, and usually outlive the tube’s usefulness. We already covered this in the heater section. The military often violated the physical integrity of tubes which is why on MIL SPEC JAN tubes you would see a “G” gravity designation not to be exceeded. Which is of course why they ditched the damn things as soon as semiconductors came around.
In the interest of being thorough though, on old testers the “Short” button would do the following: it would run the full mains current AC between the two selected elements and complete the circuit with a neon bulb (with no filament in line but with gas to make the emission connection). If a stated value was exceeded (as defined by the minimum firing voltage of neon) the bulb would glow. The problem with that is that the grid filament was not designed for 100+ volts. Some double triodes have a limiting value at the grid of only 2V- so why stress them at 50X their expected voltage carrying specification? The only proper way to test this “short” business is to actually calculate the resistance between two adjoining elements at a given voltage with the use of an ohm meter. That is how the military do it (see abbreviated reading list) and that is how it should be done. If you are designing circuits its critical, otherwise its not, assume the tube is within the parameters unless anything else is amiss.
2.21: In Circuit Shorts.
Shorts are not a bad thing in examining a tube within a circuit. As we mentioned above a short is only a bad thing if the thermionic profile of a tube goes “out of bounds”.
A “short” is a very useful “in circuit” term also. If we say “short the cathode to the B Supply positive terminal” we are assuming an in circuit understanding of the term. Generally the aforementioned would not require resistance between the two branches of the circuit so unless we specifically indicate a resistance value a simple connection will do. Similarly if we say “short the anode to grid 2” that is functional to the circuit, not the tube and could require some resistance in line, and if so, the resistance value would be indicated. But generally “short” simply refers to “make a connection” between two points in a circuit. If its qualified, its via a resistor of a given value.
The above is again part of the confusion surrounding tube testing and electronics generally. Just like with a “Gas Test” a “Short Test” is not quite what it would indicate colloquially. The term is context sensitive and can have some contradictory meanings. Its not the only term though that has an apparent contradictory meaning. Take the word “shunt”. A shunt is properly a “low resistance path” between two points in a circuit. So it would seem appropriate to use “short” for when you use a resistor to connect two branches of a circuit and “shunt” when you make a direct connection via copper wire. But the fanciest device in our tester is the mA meter. And in meter design to “shunt” actually means to run resistance: An ammeter shunt allows current values exceeding an ammeters specifications to be brought down to a useful range. In this example a manganin resistor of a known value is placed in series with the load so that all the current can be measured. The voltage drop across the shunt is proportional to the current flowing through it and since its resistance is known a meter connected across the shunt can be scaled to display the current value. Without going any further into the mathematics of this, the “low resistance path” refers to the fact that the resistance of the shunt is very small in order not to affect the value we are trying to measure. Again its context sensitive. Even if our ammeter is capable of being shunted, we decided not to. We simply chose a wide range between 0.01mA and 200mA. We could have used a meter with a narrower range, say in 50mA increments and used shunts to work the full range 0-500mA. But even though it’s a “low resistance” process its resistance nonetheless and that would have affected the validity of the measurement. It would also require a switch, introducing more resistance. A 1% tolerance resistor would mean a test result off by 1% either way, consequently our result would at the very least be theoretically 2% off. So instead it was chosen to give the user the ability to bypass the meter and use another instrument.
Short, of course is also an indicator of distance. And the greater the length of a wire, the greater the resistance. Copper has less resistance than aluminum. But it has resistance. So ultimately we have come full circle: the shorter the connection, the less resistance you have. To short thus really means connect with the least resistance possible, and if qualified, with a given value. This comes from the fact that in circuit design every connection and branch posses resistance and creates emission, so even though you are only calculating the values of the components you are using (resistors capacitors, diodes et c.) the wires/connections within the circuit itself affects the final output, so ultimately you need to keep them “as short” as possible, and if high potentials are involved, as far from each other . Its not much of a stretch to describe this in terms “ghost values” chasing “bogey values”.
The only thing we are trying to convey in this section is that a short is not an off switch misapplied. Everything in a circuit is shorting. Its all a matter of degree. What most people consider a short is an unwanted connection between two potentials. Unwanted in the case of thermionic valves is a condition where run away emission occurs.
2.22: Vacuum Tube Emission Drift.
Emission drift is the observable change in the emission profile over time. I say profile because we have defined emission (Ia) as a combination of Vk – Vg at a given Temperature. If the temperature changes with heater filament ageing, then so will the emission. The issue though is that the heater will stabilize after a couple hundred hours, and remain consistent over the course of a couple thousand hours. Similarly if you were to take your measurement at Vk= 45, Vg = 1, and run the tube that conservatively in between, there would be very little change in emission profile. If you were to run the tube at twice that voltage or three times that voltage you would see far greater loss of emission over time. Emission like longevity is temperature (T) determined.
This brings up two related issues:
Tubes must in practice be compared to other similar tubes within the same circuit. Gain is dependent upon the change in anode current, which takes place across a particular load. If you change the load the results change, the expected drift changes, life expectancy changes. Similarly a transconductance test absent knowledge of the load is meaningless. What we are getting to here is that if you insert multiple pairs of output tubes within the same circuit, their drift should be similar. If you have eight 6550’s in the same circuit and when new test within the same bogey value, and you use them for 1000 hours (time enough for the filaments to stabilize) you should have similar bogey values at that time constant. If one of those eight instead drifts way off, that particular tube then functionally becomes “off spec” compared to the other seven. Thus a perfectly good tube, could be off spec compared to the other tubes within the same circuit.
This brings us to some observations:
You need a New Tube measurement.
You need a Burnt In (first 50-100 hours) measurement.
You need a 1000 hour measurement.
AND you need to compare “one of many” tubes in the same circuit to each other.
Matched tubes are only realistically possible if you have burnt in the tubes. In our experience 20-40 hours is sufficient to see correlations between the 20 hour value and the 200 hour value. The reason why nobody does this is not because of the cost but because of the fact that you would no longer be selling new tubes. And in any event the load you place on them is different circuit to circuit. The last reason nobody brings this up is because no equipment manufacturer want to be held responsible for the tube value (Ia) in its equipment. After all they have no idea how that circuit will be employed (and in audio circuits at what volume playing what music).
The only practical conclusion is that tube testing has a strong tube to tube in circuit component to it. If you buy new equipment you should mark the tubes, keep the observable values for reference in a log, and then compare those original findings to how the tube performs after a determined period of time, compared to its peers. What matters is that those tubes all age in comparatively the same way.
Similarly, a new tube might survive a high voltage stress test with stable emission when new, but not after some use. Which is why some audio equipment tubes prematurely die at louder volumes when older. Less powerful amplifiers will be driven allot harder than more powerful ones.
- 23: Tube Sweet Spot
In audio equipment, a time period is necessary for the equipment to “burn in”. That is of course an empirical observation based on subjective criteria. But, it is also an objectively observable event. Resistor values drift, capacitance drifts, every component’s value drifts however you measure it. If you measure the circuit its observable objective values drift. Better components drift less, but not if excessively stressed by the circuit.
Nonetheless, anybody that has owned a vacuum tube amplifier knows, that the amp will “open up” in time and sound better after 500 hours than when new. There comes a time though when the tubes will start lacking in some empirical objective measure. I am sure that the tube emission bogey will by that point have drifted. But it will also have drifted in comparison to the other tubes within the circuit. Which means you have two separate events, same tube drift, and tube to tube drift. Either way there always comes a time when tubes must be replaced. More often than not that is because older tubes are more prone to going into runaway emission when pushed to higher volumes. If the distortion becomes more noticeable when the volume is raised, it might be time to replace those tubes regardless of how well they test at a standard test cathode voltage. Finally you must also remember that all the other components in the circuit will have drifted, which means that even identical tubes to the originals will sound different by virtue of everything else in the circuit.
2.24: Some Final Observations.
A hurricane blowing at 132 miles per hour is no different from a hurricane blowing at 128 miles per hour. Empirically you can say this though: trees will loose their leaves at 70mph, will loose their branches at 90mph, and will fall at around 120mph. Similarly the building code demands that the roof withstand a 130mph wind. But when presented with a 128mph wind my roof might just blow off. 130mph is not a demonstrable scientific mathematical trigger value. Its an empirical indication of what range of wind speed will do what. Similarly we have a problem of focus. In an attempt to eliminate all variables and give a single value answer, we fail to see that hurricanes are not just wind events, they are rain events, they are flooding events and ultimately storm surge events at the shore. A single measure of anything is empirically not usefull.
Take tube measurements with the same empirical approach.
Some of you probably wondered why I was so mean to the late unlamented Georg Simon Ohm in the previous chapter. The reason is simple. He was wrong in his conclusions. His logical and systematic use of new mathematical tools outlived him though. I bring this up because you should look at this mathematically, and record measurements mathematically. But there is a difference between reason and verifiable reason. There is also the fact that method is important on its own merits. But the conclusion from rigorous method not always correlates with observable empirical results. Science is great when you can isolate a single component. When you have an object and a subject. But it is very hard to come to any certainty when you have multiple independent variables. You can seek to isolate them and treat them separately, but you always reach a point where too many variables invalidate the empirical observable results. So use caution when expressing judgments, especially scientific ones.
Here is the mathematical formula for Ohm’s Law in its present day iteration:
The ohm is defined as a resistance between two points of a conductor when a constant potential difference of one volt, applied to these points, produces in the conductor a current of one ampere, the conductor not being the seat of any electromotive force, where V= volt, A = ampere, m = metre, kg = kilogram, s= second, C = coulomb, J = joule, S = Siemens, F = farad.
The siemens (S) is the derived unit of electric conductance and admittance, and is also known as the mho (ohm spelled backwards) and it’s the reciprocal of resistance in ohms.
The problem with that is that cathodes and anodes don’t conduct per se, thus they cannot be the reciprocal of anything. What they do do, is offer qualified resistance to the path of electrons jumping a physical gap, depending on distance, temperature and potential. Thus a parametric direct current thermionic tube tester does not calculate electrical conductance, or resistance, but rather the force necessary for electricity to jump a defined gap, or more specifically to allow a measured amount of electricity to jump a gap, in a circuit, at a defined speed. As mentioned in the historical section Mr Ohm was 100% wrong, the science of electricity is not really concerned with contiguous contact or effort but rather with keeping it within bounds, or generating it out of bounds. If you insert gas in a tube enclosure you are allowing it to bridge by virtue of the fact that you are providing it “something” to step on (creating contiguous action one molecule exciting the next). Arcing or sparking is simply a way of visualizing the current path through known chemicals. Its simply bringing it up to a different temperature, vaporizing it or burning it, creating yet different compounds. Absent any gas you are jumping, you are generating a electrical field at a distance.
Electricity must be viewed in terms of its potentials, in terms of its magnetic field, and in terms ultimately of its temperature. Our eyes are only temperature detectors.
In a tube centric world, of course tube elements and internal wiring have resistance. But even a super conductor has resistance- that is a given. What is important for our purposes is that the reader understand that one can always jump that gap between the cathode and the anode. All that is required are three variables: heat, voltage at the cathode and voltage at the grid. Two more grids complicate the matter but not theoretically, they just add more variables. Yes, some cathode coatings are better than others at conducting or allowing jumping the gap, but that still has to be viewed in term of resistance not in terms of a reverse ohm Siemens value of “conductivity”. Everything poses resistance. And what we are ultimately talking about is the difference in potential between anode and cathode, which like two poles of a magnet are complementary. One pushes, one pulls. The grids attract or repel.
Similarly the grid wire must be viewed in terms of how well it controls the cathode to anode emission. Yes, the wire has resistance regardless of how well it conducts electricity, for everything has resistance. But regardless of whether its emitting or getting over powered by the cathode to anode emission (the cathodic voltage being pushed to be more anodic than the other elements) its principal function is not to emit or receive but to control the cathode to anode emission at a given voltage potential, and provided that all over variables are within certain verifiable parameters it will do so. If it does not, something is amiss. Similarly the heater filament is designed and constructed so that it produces a precise amount of heat. It obviously puts out some visible light but as a light bulb it is inefficient (more heat less light). The heater filament is also by design, not very good at emitting. If brought up to temperature with cathode and grids at zero voltage, it should not emit anything noticeable, or better, not emit in any manner that cannot be overcome by the cathode emission. I have accidentally run a 6.3V heater at 12.6V and the only reason I knew is because I saw 30mA at the anode meter and 10mA at the grid meter- where the value should be zero because no voltage was applied to either the cathode or the grid. But attempting to calculate how well a heater or grid “conducts” is a destructive test by virtue of the fact that it was not designed to so. Running conductivity comparisons between cathode and any of the grids is similarly a destructive test. Clearly in a heater to cathode comparison the cathode has a higher Siemens value. But the purpose of the heater is to maximize resistance and create heat. It would be oxymoronic to ascribe it a Siemens value by comparison to the cathode.
I think we have gone as far as I can go without touching on circuit design and without using mathematics. Nonetheless a quick note. Whether the “bias” is a fixed grid value, or a operator adjustable value, or a self adjusting variable, what should matter to the user is that the tubes he inserts are largely within the industry parameters for a given family of tubes. If your circuit is operating on a fixed bias, then it could be critical that all tubes have the same emission when presented with the same voltage at the cathode. In a design with user adjustable bias, it is less critical because you have control over the variable. In a self biasing design as long as they are within 10% of each other, the circuit can compensate. The point that I am making here is that 1mA differences in a power output tube is irrelevant. In a 12AX7 it is relevant. So what matters is that the tube in the application remains within the envelope of acceptable behavior and does not go into run away emission.
Similarly if one manufacturer has a less conductive cathode material, it does not matter. What matters is that the design engineer figures out what the square footage of his cathode must be so that the final emission at a given industry accepted cathode voltage and grid voltage is consistent with other thermionic valves in that family of tubes. Similarly the grid could offer more or less resistance depending on material selection. What matters is again that at a given voltage it does its job. An EL34 with a shut down grid value of 13.2V is just as good as an EL34B with a shut down value of 14V. In a self bias design the amplifier circuit does not care. There are speed differences, but those are in circuit considerations.
2.25: Safety First.
BEFORE WE GO ANY FURTHER WE MUST DO THE “DON’T TRY THIS AT HOME” SECTION.
I mention the rotary switch test with some concern. 300V DC is allot of power even if there is a limit to that power. The B supply on our tester will not go much beyond 500mA. But 300 V DC at 500mA is allot of power. You need to take great care not to electrocute your self. You need to take that seriously. If you buy a DC Regulated Power Supply you should know what are you are doing. Like wise with our A,B,C supplies, you need to take great care. With the test socket inserted in the test fixture and all hold down screws properly secured you will never be in contact with any electricity. But if you disable the safety features you will get shocked. A tube tester is a powerful device. The joke on powerful tube amplifiers is that if you don’t put the speaker connections cover on, your cat might accidentally shock himself. I don’t like cats so it does not matter to me much but what does matter is that you don’t let children play with any equipment that can put out the kind of voltages we are talking about. If you still fail to heed our warning we suggest you go to your nearest store and buy a 9V battery (like the ones used in your smoke detectors) and take the live side with the pos/neg leads and touch your tongue. It will sting. Now multiply it by 33.333 and visualize the result on a 300V DC supply. Even Galvani’s frogs would have been upset!