Tubes

[JC@]

I know that as they age, tubes can get noisy, gassy, weak, shorted...
But why really do tubes change their sound quality over the time?
I mean, why do they become dull sounding, for example?
This is not the transistor vs tube dilemma or something like that, as everybody knows how "crisp" they sound when new.

[Enzo]

The cathode inside the tube has a coating, and that coating is designed to offer up electrons when heated. The heater then raises the temperature of the cathode to - I forget what exactly, 1200 degrees? - the point the coating emits these electrons. The positive charge on the plate then attracts them.

The coating eventually erodes away, and the electron emission is compromised. Like a scratch and sniff, after a while you can scratch but there is no sniff left.

Not only that, but the vacuum within the glass tube can get weak. More air molecules get inside and interfere with electron flow.

Those are a couple main factors.

[JC@]

So there's no magic at all, just numbers

[Robert M. Martinelli]

So there's no magic at all, just numbers

No magic? Just Numbers? Galileo once said "Math is the alphabet God used when he wrote the universe" - And math really IS everywhere....each time we strap on our guitars we tend to forget that notes, scales, chords, are all tied together by a mathematical relationship, and so is each component that can be found in any amp ( including tubes, with their ongoing electron dances ). Our being human adds the necessary unpredictability needed to make things moving, exciting, boring, better, worse....ooops, I'm turning into a cheap philosopher...

Cheers

Bob

[Steve Conner]

I don't believe that tubes become dull sounding as they age. An amp with worn-out tubes may well sound duller than an amp with fresh tubes, but that's not the same thing.

To clarify: The tubes can't make as much power when they wear out. As Enzo said, you can scratch all you want, but the sniff is gone. Because guitar amps are treble boosted, you probably notice the lack of power in the treble first.

While we're on the subject of cheap philosophy: Someone here recently claimed that the human mind is the most complex thing in the known universe. I'd go one step further, and say that the known universe is the most complex thing in the human mind.

In other words, the universe is not "made out of math". The known universe is, but that's only because it's a figment of our collective imaginations, just as math is. The real universe, however, is something much more mysterious. Claiming that it's the same as the known universe is claiming that you know the mind of God, which used to be called the sin of hubris.

[ThermionicScott]

I'll throw out a theory, and if I'm wrong, would somebody smack me?

As a tube ages, the transconductance goes down and the plate resistance goes up. This has the effect of pushing the plate curves downward on the chart. If you were to design an amp around these old tubes, you would need a higher load for full power and bandwidth. Since you haven't changed the load in your amp, this is like hooking up a lower-than-optimal load to a healthy amp. Running a lower-than-optimal load generally results in a slightly darker/smoother tone. So perhaps that's why old tubes are reputed to sound "duller."

- Scott

[Robert M. Martinelli]

Scott,
Don't want to "smack" you, but, AFAIK, the amplification factor u and the transconductance Gm remain pretty constant throughout the whole life of a valve. Performance usually starts to decay when the cathode wears out and it becomes more and more difficult for the heater to "dislodge" the same amount of electrons away from it, so the emission gets weaker.

As Steve said, most amps put some emphasis on highs when amplifying the signal, and this could be the reason why amps with worn out tubes sometimes tend to sound duller.

JM2CW

Cheers

Bob

[ThermionicScott]

Transconductance for tubes is usually quoted for a couple of operating points (and always for class A conditions), so it might appear to be a constant.

Transconductance and plate current affect each other quite a bit. Increasing the plate current has the result of increasing the transconductance to some degree -- you can see that for tubes that are listed with several entries for "typical operation", and visually by noticing how the plate curves are spread further apart at the top of the chart than at the bottom.

So as the cathode wears out, and the plate is no longer able to attract as many electrons, the plate curves sink to the floor, taking the transconductance for a given operating point with them.

That said, the impedance mismatch might not be the culprit for the "dull" frequency response, but I thought I'd throw it out there.

Take care,
- Scott

[Robert M. Martinelli]

Hi Scott,
The point I wanted to highlight is that Gm and u remain "constant" (they vary slightly at different operating conditions as you rightfully pointed out) throughout the valve's life. When the valve is at the end of its useful life, the cathode's emission (number of charges dislodged in a time unit) diminishes, and that's the main factor. In a worn out valve, Gm moves down mainly as a consequence of the cathode's diminished emission (and partly because of the Ra slightly increasing as you correctly stated). If the cathode was able to go on emitting the same number of charges vs. time, Gm would still remain more or less the same.

There also are other factors, as Enzo pointed out, but cathode wear is by far the most significant IMHO.

Cheers

Bob

[Steve Conner]

Another possibility is that the "clipping" due to running out of cathode emission is soft. Whereas, if you've scoped the speaker output from an overdriven power stage, you can see that the clipping is normally pretty hard, caused by the onset of grid current.

So as power tubes age, the clipping of the power stage could get softer overall, which would mean less harmonic content and a duller sound.

I'm prepared to believe that the measured gm doesn't change, provided it's always measured at a current that the worn cathode can still emit without saturation. Above this current, though, the effective gm would fall. And lower gm means less gain inside the power amp's NFB loop, which means the presence control will boost less, another possible reason for the dulling.

[dd2]

Testing Tube Life

Given mu & Gm do not change much over operating life, any suggestions on how to develop a test (or series of tests) that could roughly determine whether a tube is nearing end of life? (Assume a tester with variable load current & variable supplies.) Can this be done without knowing the specific early life characteristics of any given tube? Slope of load line perhaps?

[Steve Conner]

Best test I know of is to check whether your amp still makes its rated power without clipping.

When you install fresh tubes, measure the max undistorted power output and note it down somewhere. Then recheck it every so often, and if it's dropped significantly, time for new tubes.

This amounts to the same thing as an emission test of about 500mA cathode current, in the case of EL34s and 6L6s.

It also assumes the amp has a clean channel with a gain structure that can actually drive the power tubes fully.

And last but not least, it assumes constant line voltage, because power output varies strongly with line voltage.

[dd2]

Thanks Steve, I'm looking for a more general way to test for near EOL via parameter measurements; a way to sort through boxes of pre-amp & power tubes. I've built several testers so I can vary input parameters & output current load, however based on previous discussions in this thread mu & Gm don't change much as a function of wear-out.

[Steve Conner]

Well then, build an emission tester like I suggested. Emission is just another name for the maximum current the tube can pass. It's a non-linear thing, so mu and gm can't quantify it.

You could maybe do it like this: Connect G1 and cathode to ground. Connect the screen to some supply voltage (250v? 300? 350?) via a resistor, and connect the plate to the same supply via another resistor, or even a couple of light bulbs in series.

If the tube can pass 500mA, then it can light two 120V, 60 watt bulbs in series. If it can't, the bulbs will only light dimly or not at all, and the plate will go red.

Note that you need to test in excess of the absolute maximum cathode current on the tube datasheet. This is implicitly the value that you'd measure in a working amp under sine wave test, with an old analog VOM, so it's an average, and the tubes need to make the peak current, which is about 1.5 times the average.

Used power tubes are just as often "bad" because of excessive grid leakage (caused by gas or grid contamination) or partly melted screen grids, so you need to test those too. I've come across examples of all of these, worst of all are a beautiful pair of brown base GEC KT66s, except one of them has a hole burnt in the screen by the previous owner. :-(

[Old Tele man]

...a tubes' transconductance (gm) value is a cubic-root funtion of its' plate current, hence anything that diminishes usable plate current ends up reducing the tubes' usable gm:

gm = (3/2)*(Ip * G^2)^(1/3)

where:
gm = tube transconductance, amps-per-volt
Ip = plate current, amps
G = tube perveance, amps-per-volt^(3/2)