How Do Tube Rectifiers Fail?

[rp]

Like curve A or like curve B?

And what is the difference btwn one that tests excellent and one that tests good? Is it only the life left or do you experience a gradual decline in output in voltage or capacity in current, or higher noise as it declines? What are the audible indications it's going to fail if any? I don't really care about visual indications (sparks) as I'm usually not looking at the back of the amp as I play.

BTW got some 6087/5Y3s off eBay and the guy shipped them in a padded Jiffy envelope with an extra turn of small bubble wrap over the 3 tubes together (they survived). Note to self: don't buy Genelec KT66s from the guy.

[tubeswell]

They can fail like Curve B, esp if they develop a plate to cathode short. The only thing that'll really protect you from that is SS 'protection' diodes in series with the tube's anodes and the PT's High Tension winding (banded ends pointing towards the tube anodes).

On the other hand if they simply go 'open' (without shorting) that shouldn't be a problem for your PT, (but the tube will be haddit). Usually a 'B' mode of failure, but could be prompted by a gradual decline in performance over time.

If they lose their vacuum, they've haddit, and that's more likely to be 'B' mode of failure (if its a sudden loss of vacuum).

[tictac]

The experienced tube guys I know say that in guitar amps tubes usually fail mechanically (due to vibration) more than anything else.

Which is why I only build heads with the glass envelope facing up ala Marshall...

Tubes run cooler and they are away from the sound field of the speakers...

TT

[gary sanders]

All I know is I got a used 5Y3 and hit the switch and it looked like this inside.Kind of looked cool but Im sure not so cool for the amp.

[Prairie Dawg]

I don't think you can extrapolate a general rule of failure characteristics without some inside knowledge. I think you would have an initial high rate, then a settling in, and then near the end of the design life a trending upwards-remarkably like turbine blade stretch characteristics.

Furthermore you might have a family of curves based on operating conditions. The rougher the service the shorter the life span.

Of course to establish that curve you'd have to premise it on similar baseline operating conditions, voltages, etc and we know that's not the case in guitar amps.

[Zippy]

I don't think you can extrapolate a general rule of failure characteristics without some inside knowledge. I think you would have an initial high rate, then a settling in, and then near the end of the design life a trending upwards-remarkably like turbine blade stretch characteristics.

Why would you posit that? Have you any of the "inside knowledge" that you profess is required?

"Initial high rate" of what?

This makes little to no sense as you state it.

[VacuumVoodoo]

rp: What do the axis represent in your two graphs?

[Zippy]

I'm guessing voltage vs lifetime - and that's why I don't understand Prairie Dawg's idea that voltage will increase in early life.

Failure mode is not going to be a very great fraction of the life span.

[rp]

rp: What do the axis represent in your two graphs?

'Goodness' and time. What I really want to know is what happens to 'goodness' as time goes by? Goodness being the sum total of everything a rectifier does but I'm not sure what that sum is - unlike pre and power tubes which I can hear the sonics of and I can hear them degrade over time (at least in hifi amps) rectifiers are pretty two dimensional just voltage and current, they either work or they don't, and besides sag I don't think they have a sonic effect, though I've never a/b'd rectifiers.

I'm not really sure what the difference is btwn a rectifier rated weak and one rated excellent? Do they put out less voltage as they age or become inconsistent? Is there any sonic disadvantage to buying a weaker tube or are you only buying less life-time and increased risk of a catastrophic failure? I wouldn't buy a cheaper power tube tested at say '70' with a minimum of '60' since I think it would sound bad, but I might buy a cheaper rectifier rated the same and just use it till it fails, hoping the fuses save the PT.

In short is the only difference btwn an new rectifier and an old one just the # if hours left in it?

[Prairie Dawg]

I'm guessing voltage vs lifetime - and that's why I don't understand Prairie Dawg's idea that voltage will increase in early life.

Failure mode is not going to be a very great fraction of the life span.

Where did I say voltage would increase in early life? If I did I misspoke. Happens some times. Zippy, the inside knowledge I was referring to was that which might be possessed by someone who works in a tube factory doing quality testing....not me.

Let me clarify what I was getting at. If a rectifier tube is like most other devices, looking at a population of them there would be an initial high rate of failure, then service life, and when the devices near the end of their design life the rate of failure will trend upwards.

That would be fairly predictable under a standardized set of test conditions and I suspect that is how MTBF figures are generated. What we know is, all guitar amps are not equal, and the operating conditions therein vary. That, of course will change the failure rate and increase it as the operating limits are approached-which is why I opined that what you might have in those cases is a set of characteristic failure curves where the operating conditions are known.

On the other hand if we're talking about an individual tube it would be much harder to characterize the failure as one of old age or because of an inherent defect, or because of exceeding the operating limitations.

the chart is what I think a population would look like.

[Zippy]

Where did I say voltage would increase in early life? If I did I misspoke.

This is the quote I cited:

I think you would have an initial high rate, then a settling in, and then near the end of the design life a trending upwards-remarkably like turbine blade stretch characteristics.

Since neither you nor rp stipulated nor labeled the axes of your implied plots, I had to guess at what you meant. "High rate" usually means a change with respect to time. If you were talking about failure rate, then "settling in" makes even less sense.

rp now cites the axes as "goodness" v "time" - so we are back to talking about the output characteristics of a single tube during its life span (not the time-to-failure for a population of tubes).

[Prairie Dawg]

It's all good. It makes for an interesting discussion on a rainy Wednesday morning. I'm not an engineer but my old man was and I worked in a technical profession (aircraft engines) so maybe I got a little bit by osmosis.

It seems as if we have two discussions going on-a general one about failure rates in a population (which I may have started)-and the o/p's question about what makes an individual rectifier good/not good?

I guess the general answer for one tube would be "when it stops working as it should". Whether that's a straight line decline or a curve is another story that would require a lot more specific information, and the people who might have had that information are probably long since gone, along with the RCA Harrison plant.

All Langford Smith says about the subject is that a new tube starts out with a certain degree of emission that increases over time and then starts its inevitable decline, but the user does not suffer any detriment until 'the emission is insufficient to provide peak currents without distortion', whatever that means.