Tube swaps and frequency response

[Chuck H]

I'd like to clarify what I mean to say. I expect tubes operated ideally to be similarly linear, rendering all tubes as simple and predictable components and only one part of a more complex circuit. But tubes, especially as used in guitar amps, are often used in a way that fails their proper operation as linear amplifiers. So as far as I can tell the real question here is: Do all tubes of different constructions operated outside their intended parameters as amplifiers fail this function in a same and linear way??? As far as I know there are no good studies about this. Only mojo and guru arguments against techs and engineers that want to stick everything into the same old boxes that have been used since "distortion" was something to be designed out of an amplifier, not into it. And as far as I know, all the tube data and information we have to work with is expressly limited to their ideal operation as amplifiers to limit distortions.

[jazbo8]

*sigh* So much data, so little use.

Agree there is a lot fluff in his writing, but Myles was the person responsible for testing, qualifying and grading the tubes from the factories for Groove Tubes, the methodology and equipment that they used, while not fully disclosed (for trade secret, marketing purpose, etc.), are sprinkled here and there in his writings and on GT's website. So if you must know how the tests were done, you can probably get a good idea. But the key takeaway for me from the article, was simply, that the average user really has no way to tell just how good or bad the tubes that he is buying would sound in the amp, it's pretty much a hit or miss situation. So when people give specific recommendations on such and such tube from this or that brand as many are prone to do, those are basically useless information, where as if they were to recommned certain brand/tube because of better construction, or less microphonic, then those claims probably have a bit more creditability.

Besides, we all know that the OPT and the speaker make even bigger differences in how the amp would sound anyway.

[potatofarmer]

If you don't know how an amp works, or even own a screwdriver, you can experiment with changing preamp tubes. If you can bias (or can't), you can experiment with power tubes.

I think this is a big part of it; tubes are basically magic to most guitarists, and if each little glass bottle holds a fairy, who's to say that different fairies don't have different personalities? I don't really see the harm in tube rolling, but I've never heard a difference from one tube to the next that wouldn't be swamped by changing one resistor.

I don't remember if these were ever brought up here but here are some pretty plots:
https://www.tubesandmore.com/tech_co...ent_made_tubes
https://www.tubesandmore.com/tech_co...ent-made-tubes

And you could look at those frequency plots and say "wow, what a difference from one brand to the next!" ...or you could say, "huh, you take ten tubes and get +/- 1dB regardless of who made 'em. Probably less if you correct for gain." They could have done a whole lot better - larger number of tubes, more linear amp - but it would still come down to a Rorschach.

Herding cats is pretty easy. Just open up a wet food at the anode and they all come running.

[teemuk]

...Do all tubes of different constructions operated outside their intended parameters as amplifiers fail this function in a same and linear way??? As far as I know there are no good studies about this.

I would love to see one too. I think though that arranging a proper test setup is far out of reach for most DIY'ers: There's not much point to test just a single tube of each "specimen", for a valid sample you need at least a handful and then you "average" their performance, if possible. Understandably proper amount of tube samples would be a quite a gigantic pile in the end. To cover even a tiny bit of today's tube supply in such test you would have to invest plenty of money into it.

We are probably never going to see it happen.

[g1]

I wish people who mention spec sheets would just drop it already .
For modern production tubes they are a rough guideline at best. The tolerances are so wide that you end up with greatly different "grades" or even new tube numbers/names for tubes that fall outside of the already huge tolerance range.
Or tubes marketed as something they are not (Russian production of Russian tube numbers called some other American tube number).
Would anyone here try to argue that a 7591 sounds the same as a 6BQ5? Would you call me an idiot for saying they sound different? Well, that's probably how far apart some examples of tubes with the same designation can be.

Put aside the premise that tube spec sheets are strictly adhered to and this whole argument becomes very different.

And Stan, thanks for your clarification of your stance, it makes sense to me that attaching particular tonal characteristics to specific tube brands is misleading marketing hype.

[J M Fahey]

FWIW and to make it clear what I say:

1) when I write "gain" is the watered down version trying to convey the main idea.

What actually varies from tube to tube, I mean same model, even same brand , simply because tubes are handmade and even a slight variation in grid wire turns spacing , part to part spacing, etc. changes it is transconductance.

Which varies gain, but also internal impedance and probably a couple other factors too, such as exact bias point.

They are made that way even today.



See that the operator has those naked EL34 structures carelessly staked at random, and she isn't even wearing a mask, so the instant she sneezes, 20 EL34 of that batch will have instantly changed parameters ... no kidding.

2) agree that modern tube datasheets are probably just a scan or redrawing of the old ones, actual measured ones are probably not flattering to put it mildly.

But from that absence of fresh evidence to deduce that old tubes were flat but modern ones have audible bumps or dips at different points within the Audio range (i.e. 5 or 6 dB at least to be easily heard by a lot of people) is an unjustified assumption.

I mean: for something to be describes as "full" we need such a bump somewhere around 200Hz; to be described as "glassy" or "biting" we need a similar one around, say, 3500 to 4500 Hz, to be described as "dark" we need to lose at least said amount above , say, 1500 Hz and so on.

I'm open to see measured curves showing that and consistently matching brands .

The only proof I'd accept that a blanket statement such as "JJ are warm, Sino are bright, Mullards are midrangey" has some merit.

Anecdotal evidence (meaning samples of 1 , a single tester/listener, random amp/guitar/speaker, absolute lack of even the crudest measurements, etc.) does not rank high in my book.

Meaning that it is valid ... to that one person, that one tube, that one setup .... but meaning it will be the Revealed Truth to somebody else who's asking for advice in a Forum is stretching it a little bit too much.

EDIT: FWIW Stan actually has tubes custom made for him in China, which he distributes under his own brand , and he has all the necessary equipment (and then some) and has traced them for his own use (mainly quality/consistency checking) , so he's not trusting old datasheets and hoping they apply, but checking that the OEM supplier sends him what he asked and paid for and not the factory floor sweeps and rejects.

So he's the only person in this Forum who can speak with authority about what modern tubes actually do.

[Tooboob]

You can herd cats with food, but some come from nearby, and some take longer to get there. The one that ran away two days ago might not show up at all. Another might be afraid of some of the other cats and wait until they leave.

The mechanics of the tube, design and precision, must affect the abruptness of cutoff and saturation. It still doesn't explain frequency response differences, which, we see from your linked plots, are small.

From the discussion, it looks like Miller capacitance and plate resistance are the best explanations for any real difference in frequency response that the listener hears, and the effect of both are very dependant on the external impedances feeding the grid and loading the plate.

I'm still not sure what's going on with 6L6 vs. EL34s (or why EL84s sound like the cats are angry at the bees?). I keep reading the chuggers saying that 6L6 types give more bass. I guess (not much more than a guess) that 6L6s herd cats better, witnessed by the higher clean headroom and more abrupt saturation, and the fuzz on the EL34's low notes is throwing people off, or Rp is frequency dependant for some reason I can't explain.

Thanks to all of you for voicing you opinions, experience, and even some cold, hard facts.

[km6xz]

Being involved with GT relating to technical aspect causes anyone with a lick of sense to view his writing with suspicion. I have never run across a brand in electronics where the hype far ran the results so spectacularly. GT tubes are least consistent I have seen of any of the rebranders, proven to be less reliable than other labels buying from the same factories. I always assumed it was because testing was not a high priority for the company, the budget went to promotion. Comparing the consistency and reliability of, say Ruby which I knew exactly how their tubes were tested...worked on their test rigs whenever they needed calibration or repair....which made money as a shop owner by not having units return in a couple months for warrant repair and a free set of tubes out of my pocket. I could not, and refused to, afford GT in a service environment where our rep depended on things staying fixed. Even New Sensor has a much better track record for testing and has invested in custom build testing and burn-in equipment.

[g1]

As far as I'm concerned, GT pioneered the whole idea of marketing tubes that were so far out of spec they should have been tossed in the trash bin. So from this we get the whole "grade" of tubes issue. How much tolerance difference would there have been between a GT grade 1 and a grade 10?
Maybe some of the current manufacturers are adhering closer to spec, but I believe it took Stan quite some time to find a manufacturer that did. High consistency is certainly not the norm with modern production tubes.
And back to the different tubes with same designation issue, the early Sovtek wafer base "5881" was actually a tube with a russian designator. It was somewhat similar to a 6L6 but I wouldn't expect it to sound much like a real NOS 5881, and it could withstand a lot more abuse.
So now that New Sensor have most guitar players used to the idea that a "5881" is the same as a 6L6 (many of the bottles have both numbers), they have got hold of the tung-sol name brand and are building a tube based on the original 5881 specs (lower wattage and plate voltage). Sure death in many higher voltage 6L6 amps.
Sometimes I wonder if it would have been better if tubes went extinct like they thought would happen back in the 90's.
I guess the bottom line of my thinking on the whole modern tube business is that we have to be very careful to compare apples to apples, and often when we think we are, we aren't.

[Pedro Vecino]

As far as I'm concerned, GT pioneered the whole idea of marketing tubes that were so far out of spec they should have been tossed in the trash bin. So from this we get the whole "grade" of tubes issue. How much tolerance difference would there have been between a GT grade 1 and a grade 10?

GT has distributed in the worst times real Tesla EL34, 6550 GE, 6L6GC Sylvania, Tungsram ECC83, ECC83 Ei, 5751 GE, 6072 GE and some others good tubes than do not remember. He also distributed the first Chinese EL34 (ravioli getter type) in matched pairs and quartets offering security when buying them in bulk it was like playing with a gun. He has also made mistakes as offering the 6Pi3C as a 6V6 (900 mA heaters consumption). But overall it was the company that drove the future trade of tubes in the late 80s and that has a lot of merit.
At present in GT, the division between degrees is a bit random when TAD and Ruby makes in a systematic way. With EL34, for example can be calculated 40mA approx. of maximum difference between the lowest and highest grade. It should be theoretically 4mA/step. But I have seen many times how the measures on different levels overlap. And the same happens in Mesa.

[Merlinb]

Being involved with GT relating to technical aspect causes anyone with a lick of sense to view his writing with suspicion. I have never run across a brand in electronics where the hype far ran the results so spectacularly.

+1.
There's nothing new to be known about tube parameters. If Myles refuses to divulge his test methods it is not because there's something clever about them, but because there's something fishy about them. Whenever anyone says "I can't tell, it's a secret" you know they're bull$h1tting. That's a truth universally accepted, by the way

[jazbo8]

+1.
There's nothing new to be known about tube parameters. If Myles refuses to divulge his test methods it is not because there's something clever about them, but because there's something fishy about them. Whenever anyone says "I can't tell, it's a secret" you know they're bull$h1tting. That's a truth universally accepted, by the way

Aren't you guys being a bit cynical? Just because he worked at GT, you simply discount everything he wrote? He worked for GT and may very well be bound by certain confidentiality/employment agreement to prevent him from fully disclose what was done. I only posted the link to show that there seemed to be a large variability among the brands/types of preamp tubes, albeit from an industry insider, does that mean it's all BS with not a grain of truth? If you guys are saying from experience that there is very little variability in the preamp tubes or the QC coming from the factories are actually quite acceptable, then by all means please share the data, so we can all learn something.

[Merlinb]

If you guys are saying from experience that there is very little variability in the preamp tubes or the QC coming from the factories are actually quite acceptable,

We're not sayings that. There is, and always has been, very large variation in tubes from different factories and even from the same batch of one factory. That's hardly a revelation. Even Mullard specified the tolerance to be ±20% for ra and gm, and ±10% for μ.
We're just pointing out that the exact numbers presented by Myles don't really mean anything because he doesn't say what they're supposed to mean. When he claims tube X has a 'tolerance' of 35%, what does that mean? Who knows. All we can guess is that he means tube X ain't as good as one with a tolerance of 1%, presumably. But whether he's saying tube X is 'barely usuable' or 'pretty darn good, just not perfect' is a mystery. So his data don't tell us anything we didn't already know: tubes vary.

[jazbo8]

he doesn't say what they're supposed to mean. When he claims tube X has a 'tolerance' of 35%, what does that mean?

I see now (says the blind man)... He mentioned the "1.2mA/1600 transconductance industry standard specification" (for ther 12AX7) in the article, could that be the "X" that the tests were referenced to?

[Enzo]

We can all rationalize things it COULD mean. But we shouldn't have to fill in assumptions for someone else's pitch.

"I see," said the blind man, so he picked up his hammer and saw.