Tube swaps and frequency response

[Tooboob]

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.

We all read the results of such experiments, as the experimenter searches for words to describe the differences, as if he's tasting fine wine, and numerous responses respect the findings.

Now, even in the 12AX7/ECC83, there are great variations in construction. We can attribute changes in the abruptness of saturation or cutoff to these differences. Getting all the electrons to behave the same way is like herding cats. The spec 1V bias is a very rough number, so we can expect different tubes to bias up significantly differently (especially if you wander off to 12AY7, or other semicompatible higher-current parts). This would greatly affect the character of the distortion.

The one that bothers me is the claim of altered frequency response. Plate resistance is very significant in these circuits, and it could change frequency response. Are there other factors? Miller capacitance?

[pdf64]

My thinking is that the standard tube characteristics are only specified for linear operation, whereas the sharp attack envelope of a guitar signal may cause even perceived clean tones to include a degree of overdrive.
In such conditions, differing construction materials / processes may result in 12AX7 from different manufacturers to perform differently.

[Chuck H]

My thinking is that the standard tube characteristics are only specified for linear operation, whereas the sharp attack envelope of a guitar signal may cause even perceived clean tones to include a degree of overdrive.
In such conditions, differing construction materials / processes may result in 12AX7 from different manufacturers to perform differently.

Not to mention differences in interelectrode capacitance. Multiplied some tens of thousands of times from the early stages in a cascade preamp this can make a very audible difference.

Frankly I'm tired of the "tubes sound the same" argument. It only qualifies under operating parameters that most guitar amps don't adhere to, making the observation utterly moot. As noted, it's precisely when a tube is no longer being operated under ideal conditions in a low distortion circuit that the differences become audible. With different cutoff and capacitive characteristics, bias points and minor Mu differences in overdriven circuits it should be obvious to the pocket protector ilk that WRT guitar amps the observation of different brand/construction preamp tubes of the same designation sounding different is entirely valid and even quantifiable (albeit in less than universal technical terms most often). My point is that even saying a certain tube "sounds" a certain way has at least some validity. Unfortunately the people that know there is a difference and are doing tube swaps don't usually have the bench gear results or electronic chops to convince the naysayers, who only recognize things THEY know how to quantify. Indeed, most tubes amplifying within ideal operating parameters for their design will be similarly linear. But that sort of blanket thinking can't be applied to a four or five stage cascade preamp.

Going to put on my flame retardant suit now

[teemuk]

They are known to have a lot of variance... And especially in guitar amps they get fitted to circuits that don't even try to compensate this variance. The usual mu for 12AX7 is supposed to be "higher than 100", so device gain simply must fall within certain tolerance. However, in simple circuits commonly used in guitar amps mu can greatly affect the amplitude range of filters implemented to gain stages. ...This effect plus several aforementioned, like tolerance shifts in interelectrode capacitances, slightly different responses to "overdriving", different plate impedances causing different tubes to operate at different "loadlines", etc.

[Pedro Vecino]

My appreciation is that 12AX7s from different manufacturers yet with identical transconductances sound different. Or behave differently, as one prefers.
Shuguang, JJ, Sovtek WA, WB, LPS, EH, Tungsol (Siemens, Telefunken, Phillips Holland, RFT, Ei, Tungsram...) they all sound different. Are said to be flat but I fear that the complexity of what is "transported" through them in the form of electrical signal is not enough valued. The sound of an instrument is an incredibly complex skein in continous movement. It is usually represented on a basic form by a curve with decibels up and down but that's just a eggshell. The real "activity" associated to the timbre (the magic word) is below. All this is what really is transported/transformed through a simple triode and its annexes components.
Different frequency response width, different linearity, different capacity of handle complex harmonic content and their interactions, different compression levels, focused capriciously... are some of the differentiators between different 12Ax7s models. The final effect depends on the circuit/design but these differentiators are a starting point to predict their behavior and find a relation allowing establish a suitability for a particular function. The relationship with the mechanical/electrical construction of the tube .., no idea, but these variables in compression it´s possible that they are associated with the intensity of the vacuum (it is a theory) interacting with other parameters.

How to recognize the effect of different tubes on an amplifier is plugging the guitar and playing. With good technique and many musical resources is simple in most cases. Without it, very complicated or even impossible.
Many of its effects are interrelated with own overdrive textures and that is something that the guitarist feels through his hands-brain connection modulating the expression differently and orienting his playing towards new forms. It is not just a matter of decibels above or below but something much more complex.

[Tooboob]

I agree with everyone. I have a bunch of preamp tubes, and when I get an amp running well, I swap them in. I find the results unpredictable, and the differences are very noticeable. All I'm asking is what, besides Rp, can account for varying frequency response? I don't think the tubes are bandwidth challenged.

[pdf64]

'All I'm asking is what, besides Rp, can account for varying frequency response?'
I question whether the actual frequency response of the various makes of (functional) 12AX7 differs to more than a trivial degree.
Run a frequency sweep and and any seem give a closely similar response.

Carrying on with my previous post, maybe someone with a distortion analyser could run some tests a various signal levels (including various degrees of overdrive) on a few different types?

[Leo_Gnardo]

Wasn't it only a couple weeks ago Stan wrote that all 12AX7's have a measured frequency response ruler flat from zero to what 20, 40, 60 megahertz? "Anybody claiming a difference is either repeating misinformation they read on interweb sites, or saying these things to try and fool readers." Please pardon if the quote isn't 100% exact. So folks, we're doing that, or we're having a hallucination. Put head-spinning eye-rolling smiley thing here.

FWIW I sure as hell can hear a difference between shrieky-bright jangly sounding Shuguang 12AX7's and the well-balanced response of SELECTED (not all!) JJ 12AX7's plus the JJ's nice feature of "fade to black" low hiss. But remember folks, I'm just hallucinating a mile a minute. Seeing as how I don't frequent sites other than MEF and TAG, I don't see how I'm repeating claptrap I read elsewhere including dealers' ads or fanboy sites. I hear a difference and so do my customers. We must be ka ka ka-ray-zeeeeeee.

While some guitarists extol the bell-clear hi frequency response of Telefunkens, I hear a warmth in the low mids like a plummy FM radio announcer's voice thru an EV RE-20 mic, pleasant indeed. Even with the same tube, we hear different things. Curious, innit? Nutty musicians . . . . . .

IMHO it has a lot to do with resonances, especially in combo amps. This effect will not be measured in a curve tracer. I find the "warm" low end of =C= authentic Svetlana 6L6GC's eventually gives way to a low frequency microphony "wuuuuuuuuuuuh" as those tubes' interior parts loosen up with heating & cooling cycles in use.

Not to knock Stan, he's a heck of a knowledgable and talented guy, and I'm thankful he posts on MEF. We agree on about 99.5% of everything. On a couple of minor matters, not so much. Oh well.

[loudthud]

In theory, when you measure frequency response, you are not looking at distortion, you are probably jaming a 50 Ohm generator right into the grid with no stopper.

In theory, a passive network of resistors and capacitors can not create distortion so sine wave in, sine wave out. Only the amplitude will change.

In reality, feed a sine wave into a Fender Blackface preamp and the output of the tone stack won't be a sine wave. Different tubes will have different gain and distortion characteristics and will give different waveforms.

[J M Fahey]

Wasn't it only a couple weeks ago Stan wrote that all 12AX7's have a measured frequency response ruler flat from zero to what 20, 40, 60 megahertz? "Anybody claiming a difference is either repeating misinformation they read on interweb sites, or saying these things to try and fool readers." Please pardon if the quote isn't 100% exact. So folks, we're doing that, or we're having a hallucination. Put head-spinning eye-rolling smiley thing here.

Well, let me bring sad news to you:
TUBE response is flat from DC to several MHz.

Not only Stan says so, but more important Sylvania/RCA/Philips/Telefunken/GE/etc. say so.

Just read the datasheets[tm]

"But I hear a difference" ........
Well, in some cases you may, in others it's placebo/expectations/whatever but in the few cases there is a real, measurable difference, it's from external resistors or capacitors interacting.
TUBE response is flat from DC to several MHz.

One tubeless example :

- "are resistors flat? (within audible range, forget self inductance at 40 MHz)

- "yes"

- "but ... but ... I tested a resistor in a circuit , the 10K one was bright, the 1M one was unbearably dull ..... it also sounds weaker ... all my friends, my wife and a neighbour also noticed it.
Also there's 54 threads in DIY Audio and 43 in TGP saying the same ... not forgetting 14 YT Videos, the Lord Valve page, also Howard Dumble, Diaz and G. Weber ... funny that Rodd Elliott says it isn't possible and there must be an error somewhere ...."

- "Please describe your circuit"

- "Oh, that's easy, just a tube 15W Hi Fi amp ... 1 M input impedance with 470pF in parallel for RF suppression ... but all I changed was the resistor in series with the input, so any audible change must be attributed to it.

Everything else was the same.
Scientific Method they call it "

[Chuck H]

Juan!!! Whyyyyyyy!!!!

C'mon. Really?

Piss and grape soda have near identical moisture content. I bet they taste different

Are you telling us that the only difference between tubes of like designation is gain? and therefor any tonal differences are a direct result of how the other components in the circuit are affected by more or less gain? Phooey!

Do sharp cutoff tubes clip the same as less sharp, remote, semi remote, other cutoff tubes? And isn't it primarily the physical construction that affects this? And aren't most modern preamp tubes notably different in construction?

My argument above stipulated that when tubes are operated within their design parameters for proper performance they should all be relatively linear. It's only when you get away from operational ideals that things become notably different. Different in what way? I dunno. I've never tested the phenomenon for evaluation. I wouldn't have all the proper bench tools for something like that anyway.

As to frequency and linearity, the Sovtek 12ax7wa and wb tubes are VERY dull sounding. Why? I dunno. And I'm probably not going to run any tests for this either because it's so incredibly obvious to anyone doing a comparison that there is no mistaking the effect for more or less gain changing the behavior of the circuit. And I don't care what any data sheets say because I've seen way too many rusky tube data sheets that are a verbatim data copy of old specs when the tubes are very obviously performing otherwise. Just take the challenge for me. Get a Shuguang 12ax7B and a Sovtek 12ax7wa. You may test them both for gain if you like. Even adjust circuit values as you see fit to adjust for bias and gain. Then A/B them for a listen. Then scratch your head for awhile as you come to terms with what you're hearing.

[Merlinb]

Are you telling us that the only difference between tubes of like designation is gain?

Yes, because it is the gain that gives rise to:

1) Distortion. Distortion is caused by the gain not being constant throughout the audio cycle. More distortion, and more high-harmonic distortion, is likely to be perceived as 'brightness'. Grid current limiting is a form of 'gain limiting', so grid current also influences gain.

2) Miller capacitance. Obvious this can affect the top end of the frequency response to a greater or less extent, depending on other circuit details. But the geometry of the tube does not change with frequency, hence there can be no difference in frequency response between different tubes, at least up to the megahertz range where transit time becomes important. It is the variation in Miller capacitance that can causes differences in frequency response of the circuit as a whole. There are small difference in interelectrode capacitance of course, but it is the gain of the tube that 'magnifies' these capacitances so their differences show up. Ultimately, gain is to blame for almost everything...

...but then there's microphonics. Different parts of the valve resonante at different frequencies, and it could be argued that this provides a plausible mechanism for changes in frequency response. Such changes would only appear when sound is actually being played through a nearby speaker, so it wouldn't show up during normal bench tests...

[jazbo8]

Take a look at some tube test data here (near the bottom of the page), and it is pretty clear that there is a huge variation between the tubes, not just based on tube type (even within the same family) but also from different factories and production batches, so unless you personally test each and every single tube that you put into the amp and know what their actual characterisitcs are, no generalization could really be made...

Theory is good to have - it's the gain (or the transconductance & plate resistance which make up the gain) - but you cannot buy ideal tubes off-the-shelf, so like the author said - it's really a crap shoot unless you spend some time and money to make sure the tubes are doing what you think they are doing, unlike the passive components with +-20% tolerances, the preamp tubes could have as high as +-50% variation vs. the datasheets, yes, they would still "work" just fine, but that does not mean the amps would sound the same, if fact, they shouldn't, which is what we observe in real life.

[km6xz]

It is funny how everyone needs to create a strawman to argue against, by ignoring the prior posts, and offering a biased interpretation of what was meant in the posts you guys want to criticize.

My statements that all tubes "sound the same" is not what I wrote. Tubes do not have sounds, they are one part of a complex interaction between many factors that combine to create a transfer function that describes the operation of the whole circuit. My comments were about the bizarre notion that "XYZ has weak highs" or "ABC tubes are warm sounding" being told to credulous beginners is flat out wrong and either intentional lying or intentionally ignorant. A tube + circuit + operating conditions will sound different than another combination of those factors obviously but the hoo-doo snake oil claim that tube sounds are transferable to other people's amps is wrong on every front. A better, less deceptive comment would go something like this "in my amp, the way it is, and with my playing and instrument, I like the brightness of the tubes + amplifier+speaker+playing+acoustic space but your milage will vary". Besides, tonal characteristics are also individual and come from both cultural environment and physical traits of the ear and brain. Telling others which tubes to get because of some attribute you assign to the tubes incorrectly is not being helpful in the least. There are 100 other factors in your amp and playing that are determining the sonic traits. And yes, tubes have incredibly flat response over very wide bandwidth and if you are hearing more highs there are a lot of reasons that would be expected if all the parameters of your system are known.
In engineering we try to isolate factors so results are predictable so to hear the difference in two amps that can be attributed to the tubes requires normalizing everything except the tubes and in those double blind tests, golden ears consistently fail to hear difference with any accuracy.

[Merlinb]

Take a look at some tube test data here

*sigh* So much data, so little use. Unless I'm missing something, he doesn't say how he measured it or even what "tolerance" and "Ave output" are supposed to mean. Just a short list of some bench equipment randomly dumped half way down the page. Things like this would be so useful if people would take time to explain their method.
"Test equipment used:
RCA WT-100A
Tektronix 570 vacuum tube curve tracer
Vacuum tube valley small tube characterizer for quick balance checks prior the Tek570 "