Whats a good tube tester thats affordable... and why?

[Frankenamp]

I've got a few tubes that have appeared from my scrounging about and I want to get a decent tester to see whats good and whats not. I lost out on a Lafayette tester that looked really nice (but was just a emission tester) so no great loss there... Looking at a few dozen on fleabay and the Tone Lizard's article has me still a bit confused, and he seems like he's trying to find something good about all his chilluns' unless they are really bad boys. Not gonna get a Hickock because the 'all buss-up ones' are still wanting over a hundred!  and the nearly as badly beat ones are outrageous! I'm hoping to find a 'sleeper' of sorts (and a pot of gold at the end of the rainbow- like thats ever gonna happen) and maybe through cussin' and dis-cussin' we all might learn something of value.

I'm lookin at B%K's, Sencore's, Heathkits, Precisions, Mercury's and US Navy TV-7's basically mutual conductance testers...



(Yeah, I was the kid in class asking all the dumb questions- first because I didn't know, second because I knew nobody else knew either... but didn't want to look stupid asking  : )

[FYL]

What kind of tubes, what kind of testing?

[Frankenamp]

Well, the tubes are what ever I can get my mitts on of course! I'm assuming that they'll mostly be audio types: various permutations of 6v6, 6l6, 6a/bq5's, 12ax/t/u7's, 6bm8's, 6gw8's, 6ca7's, 5Y3's, 6X4's, I have a few tv tubes, but dog knows what I'll do with 'em. As far as testing, I was hoping to do both emission and mutual conductivity, & maybe gas. 

[FYL]

Try to find a B&K 550 or 650 and modify it:
- Add voltage and current measuring jacks
- Wire unused sockets for tubes not directly supported

[DummyLoad]

stay away from emissions testers - junk.

get dynamic mutual conductance tester. stay away from the B&K-707 - half the panel is an emissions tester.

hickok, triplett, weston, heathkit tt-1a, some B&K (747 is my favorite), and some of the military testers (TV7-U) will work well.  link below is to a heathkit tt-1a - if i had the cash, i'd buy it - as long as the meter were guaranteed good.

http://cgi.ebay.com/Heathkit-TT-1A-TT-1-Vintage-Tube-Tester-Good-Shape-/140404215582?cmd=ViewItem&pt=Vintage_Electronics_R2&hash=item20b0be4f1e

[PRR]

Build a Champ. Wire extra sockets and jumpers for oddball bases and pinouts. Leave space to change cathode and plate resistors.

You don't need to verify ALL original specs. You wanna know "do it amplify audio OK?"

> both emission and mutual conductivity, & maybe gas

Testing emission on oxide cathode is like testing a lawnmower engine by jamming the governor wide-open no-load 5,700RPM. It's bad for the machine, and doesn't actually say how it works at 3,500RPM.

Gm tests require being able to set specific (and reasonable!) voltage and current. They don't actually mean much. I don't think I have ever cared. The self-bias cathode voltage with "reasonable" Rc and Vp will say if Gm is near-right or way-off.

Gas... in a Champ, if grid voltage is not zero, the tube is gassy. Very-hi-Z work like electrometers and condenser mikes may need a finer test; audio work doesn't.

[Frankenamp]

I like the Champ idea, on one hand, but on the other, I like to have some numbers as a basis of comparason. ie "matching" tubes for PP output and checking the sides of dual triodes... (some of which sounds reasonable, and some like voodoo) The Champ sounds like a great functional tester though. What is your take on the tone lizard's article? Anything else anyone has found that is also informative? I'll probably go and do likewise as PRR suggests, but the compulsive perfectionist wants numbers...

[Rich]

http://www.tone-lizard.com/Tube_Testers.html

[Frankenamp]

YEP!  That's the article!

[HotBluePlates]

I like the Champ idea, on one hand, but on the other, I like to have some numbers as a basis of comparason. ie "matching" tubes for PP output and checking the sides of dual triodes... (some of which sounds reasonable, and some like voodoo).

I'm like you. I wanted numbers.

It is informative that the guy that wrote the book on tube testers (Alan Douglas; no kidding, he wrote a book on tube testers and classic test equipment) generally agrees with the guys on a test equipment forum that tube testers (even gm testers) do not generally give absolute numbers. Meaning if you had 3 of the same model tester, you could easily wind up with 3 different measurements for a given tube, even with good calibration of each tester. That's because of how the circuit works in the tester (I'm mainly thinking of the Hickok type testers, others may be different).

What they will all give you is a good indication of how strong/weak each tube is compared to a batch that you test on the same tester. So you wind up learning that you particular tester reads high or low compared to the number given on the roll chart and/or which tube types tend to read high or low.

A lot of this has to do with fine details of how each tester's circuit works, and what limitations there are for the test method.

Without getting into that too much (and yes, I have a Hickok tube tester and have plugged close to all my tubes into it at some point), I will say that what I'm interested in doing soon is much more akin to PRR's suggestion, but not necessarily a complete amp (although that gives me a good excuse to build a Champ). I was on Pete Millett's site and downloaded Basic Audio, Volume 2 by Norman Crowhurst. The first section, Electronic Amplification, gives a good explanation of vacuum tube parameters and how to measure them with a meter and a couple of voltage supplies.

Yes, it is slower than having a tube tester. Yes, you have to do some math instead of having a direct-reading meter. But there are practical problems with owning a tube tester; if the wrong part burns out (and there are a couple) then the only way to get a replacement is to steal it off of another tube tester.

But if you have power supplies and a signal source of some kind (fortunately, I have both adjustable power supplies and way too many audio signal generators), you can measure gain, transconductance and plate resistance on your own. You can also measure amplification factor, but that is usually less interesting with triodes.

Matching output tubes is another matter. Even with a Gm tester, you only have a measurement at one operating point, and it's probably not too realistic. You'd still need to verify idle-current matching, which has often been close for me when there is a gm match, but could vary. And you don't know for sure that gm matches well over the range of operation.

I haven't cared enough about matched output tubes to tell you a good way to verify output tube matching, but I suspect the best would involve actually operating the tubes in a push-pull output stage.

[RicharD]

>I'm like you. I wanted numbers.

What is transconductance?
(loosely) the amplification factor of a tube divided by its anode resistance.
OK, so feed it with a sine wave and measure Vo/Vi.  Better yet, measure the cathode current and cross it to a curve chart.  I can't tell you how many times I've had a 12AX7 test as new on my now deceased Hickok only to plug it into my audio test circuit and have it be microphonic.  My Hickok before it's death had been demoted to simply makiing sure a tube was at least alive enough to plug into a real circuit.

[sluckey]

My Hickok before it's death had been demoted to simply makiing sure a tube was at least alive enough to plug into a real circuit.

Ain't it the truth! Even back in the late '60s when I got into this stuff, substitution had become the 'real' tube checker. It still is today in my line of work.

[PRR]

> What is transconductance? (loosely)

Your definition is correct. But you must measure TWO numbers (Mu and Rp). But Rp is hard to measure while feeding the DC unless you have a really great choke or CSS. And a Pentode's Rp is too high to measure conveniently.

Set plate-cathode voltage plus a hair more. Insert -small- resistor in plate lead to read current. Set grid-cathode voltage, check expected DC current. Insert large grid resistor, check grid leakage. Insert -small- AC voltage to grid. Measure AC plate current. (I have not read Crowhurst in years, but he probably gives all the details for this exact process.)

But on a triode, knowing Gm is not enough (on a pentode it is close-enough). The Rp has a major effect in most useful circuits. So you gotta either measure Rp or Mu. And a good reading on these is actually harder than getting Gm.

That allows you to check a tube against the one or two spec-points given in the EIA registration data (essentially the published data-sheet). But we rarely run a tube at the datasheet values (12AX7 a notable exception-- Fender put a lot of AX7s very close to 250V 1.6mA). What about when we find a 200V B+ and 470K plate resistor? It's probably run close to 100V and 0.2mA. We have a 100V spec but that's at 0.5mA.

Even with specialized test gear, finding "all" parameters for all voltages and current (even "all likely") is a job for well-salaried white-shirted junior engineers working for a company which will sell millions of tubes. And when you have "all" the data, what do you do with it?

I'm building a stud-wall. Do I test every stud for tensile strength, compressive yield, and modulus of elasticity? In all three axis? True, the sawmill already did some of this, culling the very-good sticks for structural grades and the sub-stud sticks for utility grade. But still within one market-grade wood can vary 3:1 in strength (mod elast can be closely predicted by dry weight and defects). Do I test and sort my sticks 2,900psi 3,600psi 4,100psi and use the better sticks around the door? Or maybe between doors, since the door is double-studded? No, generally I stick a stick and unless it jiggles or splits while nailing, I'm happy.

If it amplifies audio, nearly the right amount, the tube is nearly OK.

> as new on my ...Hickok only to ...have it be microphonic.

Right, there's no real rest for that except the REAL test.

There are bulk-tests in mil-spec. You put a tube in an amplifier circuit and mount that on a shaker table, run the output to a recording voltmeter, and set a limit on the shake-induced voltage you will accept. That's pretty good for knowing when the mica-punch is worn or mis-matched to the grid-rods: tubes will test OK on bench but go nuts on the shaker. That's maybe even useful weeding for tube to go in Jeeps or missles where shake is quasi-random. In audio we can accept a lot of sub-sonic wobble, and depending on cabinet we may take a lot of >1KHz ring. And even some low-mid boinging, but not one that aligns with a speaker or cabinet peak. Which pretty much means: put it in the actual application, use it.

> substitution ...the 'real' tube checker. ...my line of work.

Some of the jugs you work on would crush the Hickok.

[Frankenamp]

Well, as Alice was wont to say: :curiouser and curiouser..."
It's a lot more complex than sticking a pipette tip in a jig and determining if it's straight enough, the walls are thick enough and it not a short shot, or has bubbles in it. Oh My. I'll have to revisit Pete Millet's site (sounds like a good read). The way PRR describes the process- it seems the electronic equivalent of juggling  bowling pins on a highwire while riding a unicycle and reciting Shakespeare simultaneously. I'll have to study summore and see how far I can wrap my pea-brain around the concepts. Thanks fer the direction, every time I think I have a handle on something thermonic, I find that there's more to learn (or unlearn)! 

[simonallaway]

What is "shake-induced voltage"? Is that anything to do with the parts inside the tube literally moving and inducing current?

[HotBluePlates]

What is transconductance?
(loosely) the amplification factor of a tube divided by its anode resistance.
OK, so feed it with a sine wave and measure Vo/Vi.

The first statement is on-target. The second is a "yeah, but..." And PRR mentioned why: you can't get an output voltage without a plate load resistor, and the internal plate resistance interacts with that to give a gain less than the amplification factor. Mu is what you'd get with Vo/Vi if internal plate resistance didn't enter the picture.

Strict terms, transconductance is the transfer of conductance; conductance is how easy current flows, and is opposite of resistance which is opposition to current. So if Resistance = Volts/Current, Conductance = Current / Voltage. It conductance is high, there is a big change of current with a given change of voltage. So, transconductance (of grid 1, commonly Gm) is the change of plate current with a given change of grid 1 voltage.

All the american data sheets commonly use micromhos; the British system is more intuitive, and would call a Gm of 4000 micromhos as 4mA/V. You change g1 voltage by 1v and get a 4mA change on plate current.

Crowhurst's description is probably pretty accurate for what tube manufacturers might do. Attach a voltage supply to the plate with just a series ammeter (or milliammeter), and apply a variable dc voltage to the grid, observing the change in plate current while maintaining a constant plate voltage. That doesn't really lend itslef to automatic testing, as there are a number of variables to adjust/control at the same time. And the change of grid voltage is what is important for transconductance. You could do 2 discrete steps of grid voltage and calculate based on the resulting 2 values of plate current. Or just keep grid voltage constant while changing the plate voltage and observe/plot the change of plate current to arrive at our familiar plate curves.

Hickok did it a different way, and had an ac milliammeter in the plate circuit while applying a 60Hz signal to the grid of about 5v for most of their models. That's fine for power tubes, but large for high gain tubes like a 12AX7. So certain Hickok testers will be inherently better at testing large tubes or lower mu tubes than high mu tubes.

Which brings us around to testing the tube in the actual circuit as a preferred method. If it's an audio circuit with a speaker, you can also listen for noise and/or microphonics. Slower than the automatic testers, perhaps, but you definitely know where the tube stands.

For me, the primary case where I would consider the slower and more eleaborate methods Crowhurst shows to test Gm, gain and rp is when I want to characterize a group of tubes from different manufacturers or with differing construction styles, in an attempt to find out why some sound different than others. And fortunately, I have a good enough selection of old-production tubes that I know perform audibly different that I could follow through with this type of test. The goal would be to find out if there is an objective defining difference between the tubes which leads to certain subjective results, which I could then look for in new tubes/cheaper tubes rather than hunt for expensive older brands. The ultimate goal would be to not do a complete series of tests for every tube, but find the most representative test(s) for what I'm trying to achieve.

[HotBluePlates]

OK, so feed it with a sine wave and measure Vo/Vi.

And more definitions...

*If* the tube had zero internal resistance, Vo/Vi would be equal to mu, as mu is defined as the change of plate voltage divided by the change of grid 1 voltage, with plate current being held constant.

But real triodes have plate resistance, and the plate current does not stay constant when you change the grid voltage. If you remember that the equation for gain of a triode is Output voltage = Input voltage * mu * [Rl/(rp+Rl)], then you'll see that the output voltage would be the input voltage times mu, if it weren't for the fact that the output voltage is made smaller by the voltage divider formed from the plate load resistance and the tube's internal resistance. Ya have to have a plate load resistor for there to be any voltage output anyway, and you could indirectly determine rp by observing the resulting output voltage with an input signal and noting the output signal size, input signal size and plate load resistance. But we still have the variable "mu" hanging out there, and in measuring pure tube characteristics, the "useless circuits" without a plate load sometimes come in handy.
 
Or you turn things on their head and forget trying to measure mu directly, and measure rp instead. Take a tube and apply a fixed grid 1 voltage. Apply a plate voltage, noting plate current. Keep g1 voltage constant, and change the plate voltage, while also noting the change in plate current. That gives you a by-the-definition way of measuring rp; rp = change in plate voltage / change in plate current, while all other conditions remain constant.

Another big problem, and why "automatic tube testers" are "automatic". Tube characteristic curves aren't straight lines... they're "curves". Which means the most accurate answer is arrived at with a change in any place where the definition says "a change of..." that is infinitely small. Infinitely small change is nice with calculus and graphical measuring of printed curves, but pretty hard to realize with physical test gear. The bigger that any of the "changes" are made, the less-accurate your results are.

Which is why the 5v grid signal for some Hickoks work great with some tube types to give more-absolute numbers, while on other types they are simply representative numbers. I think I have some test gear that will do well at measuring very small voltage and current changes (uV and pA), but I have some serious doubts as to whether my voltage supplies and signal sources are up to the task; I've got some ideas about workarounds for that, though...

But this is a lot of talk about theorizing and lab-type results, and tubes are not high-precision devices, at least in the range of tolerance that has to be allowed for a manufacturable product.

[HotBluePlates]

What is "shake-induced voltage"? Is that anything to do with the parts inside the tube literally moving and inducing current?

What makes a 12AX7 different than a 12AU7? Why does one amplify more than the other?

Let's ignore the different current and power ratings for the tubes. The 12AU7 has a mu of about 17, the 12AX7 about 100. Mu, or amplification factor is not constant, but let's pretend it is since it is more constant than other tube parameters.

The control grid is the element that determines how much plate current flows in a triode if plate to cathode voltage is held constant. The 12AX7 amplifies more than the 12AU7, so maybe there is something different about their grids. In fact, there is.

The control grid is a spiral of wire wound around 2 support rods, and physically between the cathode and plate. If you apply a voltage to the grid that is negative compared to the cathode, it tends to electrostatically repel the electron flow back towards the cathode. In other words, more negative g1 voltage lowers plate current. Indirectly, if the grid becomes more effective at doing this, meaning a bigger current change is caused with the same change of g1 voltage, then you can wind up with more amplification after inserting a plate load resistor.

How do you make the grid more effective at controlling plate current? How about winding that spiral tighter and tighter, so that the wires are closer together? Now the electrostatic force acts over a smaller physical space, because the wires are closer together, so a smaller voltage will achieve the same effective control of plate current, or it is more effective with the same g1 voltage applied.

This is great if those wires are perfectly immobile, but what happens if there is slop and the wires can move? Then an outside force shaking the tube can cause more and less amplification to occur. Viewed from outside the tube, you can't really see that the internal structure is moving, you just know that when you use physical force to shake the tube, or when sound waves hit it and shake it, there is a change in output of the tube that is directly related to the force applied. From your frame of reference outside the tube, this all looks like a change you cause by shaking, or a "shake induced voltage".

Tube geometry dictates many of the tube's parameters; distance from the grid to the cathode, distance between grid wires (winding pitch), distance from cathode to plate, distance from grid to plate, size and weight of the plate material, and many more. Those mica spacers at the top and bottom of the tube structure do a number of things, but the first and foremost is to attempt to hold the entire tube system rigidly in place and prevent the tube elements from shifting and changing the tube's performance.

But maybe the holes cut in the micas are just a tiny, tiny bit too big. Now the tube structure has some slop, and any force that might act to move the innards will cause a momentary change in characteristics.

But spacer holes could be too big, the spacer could be cut just a hair too small to hold against the inside of the tube envelope, the outer envelope could be just a bit too large, grid wires could be wound just a little too loose, support rods could be just a bit too small, plate metal or the cathode could be just a little too flexible and move in mid-span (changing the distance between elements), and on and on.

There might be no such thing as a microphonic-free tube; but there are tubes that have so many design and construction features/considerations that the level of microphonics is too low to affect the circuit they will be used in.

There is an audiophile tube vendor that has high prices, but may very well be on the up-and-up. He sells preamp tubes in at least 4 different grades, based on measured noise and microphonics. The poorest performers are graded for line-stage use, because while they do perform well, there is some noise or microphonic behavior, but the tube is intended to be used in a circuit where the signal level will be so much higher than the offending behavior as to mask it. Used in accordance with his grading system, there will be no perceived defect. His highest grade of tube is for use in extremely low-level phono stages, where even the slightest hint of noise or microphonic behavior will be very large in comparison with the intended signal. Tubes meeting this criteria are very hard to come by, and his prices reflect that.

[simonallaway]

Best...explanation...ever!

Thank you!

[Fresh_Start]

Thinking out loud... OK, so I build a Champ/AX84 P1 with two 9-pin sockets and two 8-pin sockets.  Make the first stage switchable between the triodes of V1, the second stage switchable between the triodes of V2, and the power amp switchable between two power tubes.  Need adjustable bias for the power tubes, both to balance output and test different tube types.  Use on-off-on switches so any tube/triode swap puts the amp in standby.  Probably need VVR for the power amp section to allow for different tubes and different applications.  Run "known good" tubes in each spot at different signal levels and record results.  Viola!  A "tube tester" which lets you find matched triodes (if you care), match output tubes under real world conditions (if you care), test for microphonics (combo cab of course), and find out how the tubes sound.

Heck, it would be even better if you could make the four preamp triodes interchangeable so that you could do double blind tests between different preamp tubes to your heart's delight (and FYL's too ).

Is that the basic idea?  I like it but it'll have to be project #143...

Chip

[mrr3000gt]

I have a Sencore Mightmite and it looks OLD.

What is frustrating about the 50s and 60s era tube testers is that if a tube tests good, IT PROBABLY IS GOOD. If it tests bad, its bad. And anything inbetween is fair game: your 'good' tube might sound like total poop.

The only thing I really use it for is to identify a totally burned tube to confirm a diagnosis of some cheapskate playing a set of power tubes to the very end only to burn them up in their amp.

For me, the tube tester is a tool in making diagnosis and have very little value by itself - kind of like a lie detector.

But as we know, many tubes are good liars and easily pass the tubetester test!
 

[Frankenamp]

OK, I found one that is Mutual Conductance tester and also isn't in too bad (I hope) condition, for what I hope is a reasonable price.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=300446097491

It's an old B&K model 500. My plan is to see if it fires up and works. One of my wife's Avon friends has a husband who is a retired Electronics Engineer who has decided he likes building tube radios & transmitters. He also has a Hickock 534 (transition model) that still works. I ran a few tubes through it and it seems to be functioning well... I can probably test a few tubes in each tester and see how closely they correlate... but at this point it's still cart before the horse time untill it's here and functioning (quick prayer to the postal demigods that the package is treated better than some I have seen).

Am I headed in the right direction?

[mrr3000gt]

Ancient thread reviver here :-)


I snagged a B&K 550 is nice and usable shape.


EDIT: I have verified it checks all the tubes I wanna check. It also uses the Hickok circuit for DM/GM from what I can determine. Its my first DM/GM tester and I am pretty amp'd about having one (sorry I could not resist).


It is a subset of the B&K 707- does anyone out there in Hoffmanland have a B&K 707 or a 550? Looking for 1st-hand user experiences with it.


Thx!

Try to find a B&K 550 or 650 and modify it:
- Add voltage and current measuring jacks
- Wire unused sockets for tubes not directly supported

[HotBluePlates]

Ancient thread reviver here :-)

I snagged a B&K 550 is nice and usable shape.

Check your other thread.