triode noise

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

abbey road d enfer

Well-known member
Staff member
GDIY Supporter
Moderator
Joined
Jan 22, 2008
Messages
17,034
Location
Marcelland
For those who have access to the AES journal, I should mention an excellent article about triode noise calculation AND practical measurements, done by a Neve/AMS designer, one Merlin Blencowe, in the 2013 November issue (Vol 61 Nr 11).
 
Typical. I was a member of the AES  for years but got fed up with articles on surround sound and game sound so last year I did not renew. Then they publish something right up my street.

Cheers

Ian
 
Have not read the entire article but found a trouble point. He includes tubes made from the 1950's to 1990's "worldwide production". The 1990's tubes, if from Russia & China, pollute the results fro tubes made by Western countries from the 1950's to about 1970.
Too bad, it was otherwise a rigorous approach.
 
bockaudio said:
Have not read the entire article but found a trouble point. He includes tubes made from the 1950's to 1990's "worldwide production". The 1990's tubes, if from Russia & China, pollute the results fro tubes made by Western countries from the 1950's to about 1970.
Too bad, it was otherwise a rigorous approach.

Actually I think that was a good thing. First it is clear from the results that from the noise point of view there is very little difference  between tubes from widely differing sources and times. Secondly it means you don't need to pay an arm and a leg to get a branded European tube in order to get decent noise figures.

Cheers

Ian
 
bockaudio said:
Have not read the entire article but found a trouble point. He includes tubes made from the 1950's to 1990's "worldwide production". The 1990's tubes, if from Russia & China, pollute the results fro tubes made by Western countries from the 1950's to about 1970.
Too bad, it was otherwise a rigorous approach.
That's the problem with statistics, selecting the categories. As soon as you make categories, you influence the results.
In Marcelland, car insurers have found that drivers aged between 16 and 99 are the most accident-prone, so they have decided to increase their rates and conversely give a large discount to the others. :eek:
More seriously, I've receently read a couple of reviews of China, Russia, Czech vs. NOS German, British, US. In most cases, the average performance was similar, the difference being that some tubes of Eastern provenance were outrageously defective, meaning that, in terms of industrial production of tube-based equipment, selecting the tubes would almost completely solve any issue.
I understand that for condenser mic applications, the issue is probably different since grid leakage due to improper vacuum is not a specification.
 
..since grid leakage due to improper vacuum is not a specification...

Do we know that main component in grid leakage is improper vacuum? I can see a potential business in re-vacuuming stock tubes...

Back on topic, my main objection against new-production tubes is that they (in general) seem to vary enormously in average life time, and has huge shifts in operating parameters over time. Which has more or less forced me to stick with NOS tubes for the commercial Gyraf range - because I'm running minimum-feedback, and am thus dependent on somewhat stable parameters.

Jakob E.
 
gyraf said:
Do we know that main component in grid leakage is improper vacuum?
If the leakage is serious, then yes, it is probably due to poor vaccum. In a 'normal' tube it is dominated by direct leakage between anode and grid, and there isn't a lot you can do about that (except keep the valve socket clean).

Back on topic, my main objection against new-production tubes is that they (in general) seem to vary enormously in average life time, and has huge shifts in operating parameters over time.

Rapid decline in gm is normal for the first 100 hours or so, and this used to be taken care of at the factory, who would burn the devices in as part of the normal quality control procedure. I speculate that these days they perhaps don't spend as long burning in, so when the consumer receives his valves, he still has to suffer several hours of declinging gm before things settle down?
Incidentally, you can greatly stabilise gm over time by operating at a slightly reduce heater voltage.
 
gyraf said:
..since grid leakage due to improper vacuum is not a specification...

Do we know that main component in grid leakage is improper vacuum?
Yes. Not enough vacuum either/or pollution. That can be due to leaks, so would be detectable after many years. There may be other reasons that I don't know of. BTW other parameters may vary over time, in particular s parameter decreases with time due to cathode losing its emissive properties.
I can see a potential business in re-vacuuming stock tubes...
Interesting idea! Does anybody currently do this?
Back on topic, my main objection against new-production tubes is that they (in general) seem to vary enormously in average life time, and has huge shifts in operating parameters over time. Which has more or less forced me to stick with NOS tubes for the commercial Gyraf range - because I'm running minimum-feedback, and am thus dependent on somewhat stable parameters.

Jakob E.
Agreed. It seems the current pumping systems are optimized for speed of production. A big contrast to the high-rel tubes of the 50's and 60's, which took as much as 100 hours of pumping in order to evacuate the sublimates.
 
abbey road d enfer said:
I can see a potential business in re-vacuuming stock tubes...
Interesting idea! Does anybody currently do this?

Not that I've ever heard of - but someone should take it up.

On a similar path, me and Tim C. did quite a few experiments with glass-envelope modifications to calm down microphonics in EF41/42 type tubes (these sound beautiful as frontend for microphones when underheated, but are sadly horribly microphonic) - we heated small spots on the glass envelope with a torch, allowing vacuum to suck in a spike/dent of molten glass that would touch and thereby support internal structure. Actually worked in some cases - but the tube ends up looking like something out of a nightmare  ;D

Jakob E.
 
abbey road d enfer said:
Do we know that main component in grid leakage is improper vacuum?
Yes. Not enough vacuum either/or pollution. That can be due to leaks, so would be detectable after many years. There may be other reasons that I don't know of. BTW other parameters may vary over time, in particular s parameter decreases with time due to cathode losing its emissive properties.
I can see a potential business in re-vacuuming stock tubes...
Interesting idea! Does anybody currently do this?

[/quote]

A brief internet search revealed a service for X-ray tubes.
 
Serious discussion of vacuum tubes would not include anything made after 1980. We certainly don't use any in our mics, and that's where we make our living.
But, I was glad to see the article anyway, as he did investigate into the contrary realm (currents greater than 1mA) where I had not.
 
bockaudio said:
Serious discussion of vacuum tubes would not include anything made after 1980. We certainly don't use any in our mics, and that's where we make our living.
David, did you have a bad experience with post 1980 tubes?

Can you provide details?
 
Years ago when I first built the U99 at Soundelux we used the Russian EF86. Trays and trays of tubes just to get a couple of mics out. Failures in the field. Then developed pre-check/cycle burn in/final check testing. Still failures. Was told by another in the field I should sped "a second or two listening to the tube or perhaps comparing it to an NOS", did, and then changed the entire production (to NOS) as the new Russian tube was entirely unsuited to both our sound and reliability (hint: the cathode materiAL CHEMISTRY IS DIFFERENT THUS THE z OF THE TUBE IS NOT THE SAME). Fast forward  two or three years and as the first manufacturer to "reissue" the ela M251, used (don't ask me why!) a Russian 6072. Nice & quiet (they copied the RCA not the GE!), some even lasted ten years. But again (not as extreme as the ef86) reliability, and no comparison in sound. So I found other tubes (NOS) to do the job. Today the entire company is based on NOS. We do repairs at the shop as well, and in every case, our customers come back because we don't bullshit them with new (non NOS) tubes. This includes gear built by people I respect greatly (rare).
A simple test: try an NOS vs new China or Russian 12zxy in the first stage in your McIntosh Mc225. A simple amp, not crazy hi-Z like a mic. If you're ok with that, move on and ignore everything I've said. (FWIW I have also done this experiment with power tubes in different apps and my position became More Galvanized).
This is why I say, if it's not NOS, it's not serious.
 
My firsthand experience is not as extensive as David's, but has been exactly and without exception the same.

In my own gear, if funds are extremely tight, I'd MUCH sooner use a quality used, tested old-stock tube as opposed to compromising with new production.

From people intimately acquainted with such matters, there are entire raw materials supply chains that no longer exist to support manufacture of current tubes that would compare to the quality of the better NOS.
 
Incidentally, you can greatly stabilise gm over time by operating at a slightly reduce heater voltage.
This is very interesting and useful.  When you say slight reduction what exactly does this mean? 
I remember reading somewhere a discussion of reducing 6,3 filament voltages to around 5 Volts for longer life
but I thought this merely decreased gm not also thereby stabilising it. 

Bill
 
> entire raw materials supply chains that no longer exist

If you just buy commercially pure Nickel (for cathode sleeves), you get different (and sometimes useless) results with every batch. The oxide coating apparently reacts with "minor" impurities in ways that can be bad OR good.

INCO got tired of tube makers whining for special batches. Back in the 1950s (1940s war production?), once a year they would pour several batches which appeared to suit most tube makers' processes. Roll some sheet and send samples around for testing. The batch that most tube makers approved would be set aside for rolling cathode stock, until it ran low and new samples sent out.

A particular problem is "interface", which acts-like a layer of insulating scuzz between the Nickel and the oxide, one cause (of many) for Merlin's "decline in gm".

Modern metallurgy "should" give us a spec for cathode Nickel. But I suspect that even if current tube makers were sophisticated enough (and had the cash) to do the research, the total demand is too small (less than a ton a year??) to justify a special batch at the Nickel smelters.
__________________________

I'm uncomfortable with the idea that noise is grid leakage and grid leakage is vacuum leakage.

There's several causes of grid current leakage. All will happen in varying amounts. These may be sorted by direction and how they vary with Vp and Vg; I think RDH4 has a summary. In the 1950s papers by both RCA and Bell Labs, stray cathode-stuff on the grid (cathode activation is a violent process) is often blamed.

Also grid current noise ought to go to zero for a shorted grid. The main market for tubes today is guitar amps which may have significant grid impedances, so the makers must control how many too-hissy tubes they make. Mike transformer impedances are lower (so is tolerable hiss level). Of course condenser mike heads are super high impedance and grid current hiss may dominate.

Other papers explore crappy oxide as a cause for voltage noise. This is not unlike early transistor fabrication where "surface states" (dirt) on the Ge/Si die caused 5dB-40dB excess hiss. This was cured with passivated planar process where surface is sealed and nearly inactive. But we can't do that to a cathode.

Another problem is that mica leaks. And I suspect that much of the best mica was worked-out in mid 20th century (much as we killed the gutta-percha sources). Even selected best mica has to be treated with sealers.

Tubes "don't leak". Or rather, if a tube really leaks it will become useless very soon (yeah, probably in the customer's hands soon after sale, argh). There are issues with absorbed gas ("pollution") in the mica and metal. Cooking while sucking gets most of this out, but there is an exponential tradeoff between what comes out and how much it costs. Even an all-week pump-down won't remove all the absorbed gas. Also a LARGE amount of crap comes out in cathode activation. We would not have tubes as good as most are if we only relied on the pump. Both all-week and half-hour cycles RELY on the getter to clean-up after the stem is sealed.

For when it REALLY REALLY matters.... here's some of what AT&T and BPO did to make tubes they threw in the sea:
http://www3.alcatel-lucent.com/bstj/vol36-1957/articles/bstj36-1-163.pdf (15MB PDF file)
175HQ is an AT&T low-Gm tube, pages 4-16. 6P12 is a BPO tube pages 17-25. The BPO tubes were used in shallower water where recovery and repair were reasonably easy, the tubes are very good and used redundantly. The AT&T tubes were used in deep water where redundancy and recovery were extremely costly, and were built for "no" failures (across 306 tubes) in 20 years.

Fig 6 page 11 shows a 120,000 hour (14 years steady) run, Gm holding to ~~80% of initial.

Of interest is the not-uncommon case of slow emission from trapped gas pockets. Page 20 shows a test where GM does fall dramatically for the first 5,000 hours but then recovers at a rate which suggests return to initial Gm at 50,000 hours (6 years steady). Page 24 shows well-processed tubes holding 95% of initial Gm at 26,000 hours (3 years steady).
 
a lot of the chemicals used in the production of NOS tubes are no longer available or illegal to use, so exact reproductions of old tubes is something that would be hard to achieve,

you could probably do it with a massive R&D budget like they had in the old days, but the market no longer supports such investments,  :-\

RCA and GE were probably as big as General Motors and Ford back in the day,
 
Back
Top