12AX7 tubes...matched and balanced triodes...any value here?

Discussion in 'Amps/Cabs Tech Corner: Amplifier, Cab & Speakers' started by clay49, Sep 13, 2012.

  1. clay49

    clay49 Silver Supporting Member

    Messages:
    3,355
    Joined:
    Dec 28, 2006
    Location:
    Birmingham, AL
    So what's the scoop on paying extra for "matched triodes" or "balanced triodes" or both in a 12AX7 tube? And while we're on the subject, what about "gold pins"?

    Does it REALLY matter? Can anyone, other than a dog or bat, really hear the sonic difference???
     
  2. Timbre Wolf

    Timbre Wolf GoldMember Supporter Gold Supporting Member

    Messages:
    9,097
    Joined:
    Oct 14, 2005
    Location:
    Berkeley, CA
    Gold Pins: no difference, except maybe risking unevenly applied plating. Some people can't resist the allure of yellow metal, though, so here it is.

    Matched Triodes: could make a difference in a HiFi system where one triode feeds channel A, and the other channel B of a stereo preamp. Other than that, save your money.

    - Thom
     
    teleman1, Yamariv, noctilux1 and 2 others like this.
  3. clay49

    clay49 Silver Supporting Member

    Messages:
    3,355
    Joined:
    Dec 28, 2006
    Location:
    Birmingham, AL
    Thanks, Thom...was hoping you would weigh in on this!
     
  4. Onioner

    Onioner Member

    Messages:
    2,863
    Joined:
    Feb 2, 2011
    Location:
    SF bay area
    What he said, w/ the minor change that they do make a difference in some positions, but not necessarily a positive one. My conclusion so far is that it's either an irrelevant difference (it has nothing to do with how well the tube functions overall), or a detrimental quality, at least in the PI. Still haven't totally made up my mind, as I haven't tried that many balanced tubes. I'm pretty convinced that it does a make a difference in the PI, just, so far, not a positive difference.

    But, yes, save your money, unless you absolutely have to hear it for yourself. When the time comes, talk to me, and I'll probably have a few around I'm looking to dump...
     
  5. clay49

    clay49 Silver Supporting Member

    Messages:
    3,355
    Joined:
    Dec 28, 2006
    Location:
    Birmingham, AL
    Thanks for these responses. I've seen a few posts here and there from guys who say that they actually prefer preamp tubes to not be matched or balanced, just wanted to get some expert opinion!
     
  6. CyanideJunkie

    CyanideJunkie Member

    Messages:
    273
    Joined:
    Sep 6, 2012
    Location:
    Singapore
    I think that the use of balanced and matched tubes helps ensure a predictable level of performance in an amp, which I think is very important if you intend to gig with said amp.
     
    Raidermofojeff likes this.
  7. Onioner

    Onioner Member

    Messages:
    2,863
    Joined:
    Feb 2, 2011
    Location:
    SF bay area
    I have heard from some knowledgeable folks who i respect that a matched pi is a good idea. That's not what I've experienced, but those guys have a lot more experience than i do, so ymmv and all...
     
    Steppin' Wolfe likes this.
  8. Raidermofojeff

    Raidermofojeff Member

    Messages:
    156
    Joined:
    Feb 28, 2009
    Outside of PI's consider a matched triode 12AX7 in a 6G15 reverb unit

    At Doug's tubes
    Matched= both sides of triode matched for gain
    balanced=same current draw....
    You're given the choice......

    I just ordered a few to have in my stock pile, I opted for matched gain.....in a 12ax7.......
    My mindset was to match the output of my wet signal to my dry signal in my Fender 6G15 build.......
    This makes a difference! I know first hand from having a weak wet signal due to an unmatched 12ax7!
    I had put in something random out of my anos collection, it took me a long time to figure out why my unit wouldn't do
    a reasonable Dick Dale sound......It sounded good, just not super wet! A matched dual triode 12ax7 was the ticket....

    I like the extra grading done too, just in general.......and as quiet as can be.......worth the extra $......Just to have good tubes!
    That said, I've never bothered to worry about this for PI tubes...
    Only a couple times have I ever had PI tubes go out, and never heard any audible issue due to balance/match.....Only when PI tube was bad!
    I'm rolling with the RUBY 12AX7AC5 HG+ with the latest order..
    Currently in the 6G15 is
    RUBY 12AX7AC7 HG+ .......I do think the former is likely more robust....but reverb unit sounds great with this in it..

    Opinions and additional info invited......
     
    Last edited: Apr 7, 2017
  9. 71strat

    71strat Member

    Messages:
    7,968
    Joined:
    Oct 7, 2013
    I always buy them matched when I can. Brent Jessie only charges $5 for matching internals.
     
    guitar007 likes this.
  10. HotBluePlates

    HotBluePlates Member

    Messages:
    6,034
    Joined:
    Mar 27, 2017
    IMO, matched sections is mostly a plus if you use the tube in the phase inverter slot for a long-tail inverter (or a diff-amp stage after a split-load inverter), if you seek the most clean output possible from the amp. In that case, tightly-matched output tubes is also a plus.

    But perfect balance in the push-pull output section guarantees maximum cancellation of even harmonic distortion in the output tubes. For me that's not usually a goal, nor do I often seek all the clean output power possible (I like output tube/OT/speaker distortion before I like preamp distortion).

    Some people play older tweed amps or the earlier tweed Tremolux a dig on its grind that the later tweed non-Trem Deluxe doesn't do the same way. The earlier amps (and definitely the Tremolux) used a paraphase inverter which was not likely to ever have perfectly matched outputs driving the output tubes. Some folks really like the resulting "slightly lo-fi" sound.
    _____________________________

    That said, I don't like any preamp tube to have one side mostly-dead. After all, what's the point of that?
     
  11. Timbre Wolf

    Timbre Wolf GoldMember Supporter Gold Supporting Member

    Messages:
    9,097
    Joined:
    Oct 14, 2005
    Location:
    Berkeley, CA
    Only one triode is used in a paraphase phase inverter. Balancing triodes is utterly pointless.

    - T
     
    pdf64 likes this.
  12. VaughnC

    VaughnC Supporting Member

    Messages:
    16,745
    Joined:
    Jan 4, 2002
    Location:
    Pennsylvania
    While any tube may subjectively sound different, in a guitar amp there's no technical benefit to using matched triode 12AX7's. And even though some phase inverter's use both halves of a 12AX7, the circuit itself is in seldom in balance.
     
  13. pdf64

    pdf64 Member

    Messages:
    5,790
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    It's great that your reverb is working to your satisfaction now.
    But the reason is unlikely to have been dependant on the 12AX7 (7025) sections being well matched.
    I suspect that pretty much any of the higher gain brands of current manufacture 12AX7 would have worked just as well, given a reasonable degree of QA dealer pre-screening for suitability in MI gear.

    Please see the signal paths within the equipment http://schems.com/schematicheaven.net/manu/fender/reverb_6g15_schem.pdf

    The wet signal is amplified by a common cathode (CC) 12AT7 gain stage, then a gain pot (dwell) and further amplified by a second CC 12AT7 gain stage driving the 6K6 single ended CC power amp.
    The OT secondary of this then drives the reverb tank, the output of which feeds a 7025 CC reverb recovery gain stage, which feeds the tone, and wet end of the mix, pots.
    The dry signal is buffered by a unity gain cathode follower (which uses the other 7025 section), which feeds the dry end of the mix pot.

    Due to the overwhelming differences in the 2 signal paths, and in the application of the 2 7025 sections, I hope you can see that there is absolutely no conceivable benefit from the 12AX7 / 7025 sections being matched?

    My guess is that you may have lower than intended gain in the wet signal path of your 6G15, perhaps due to drifted resistors / bad ecaps. Or, as good gain is required by the reverb recovery stage (hence a 7025, which is a 12AX7 tweaked / selected for high gain and low noise/microphonics, being specified), perhaps your ANOS (?) 12AX7 are a bit low on gain.
    But pretty much any functional triode will provide suitable performance for the dry signal CF buffer.
    How can it be optional?
    For the system to be balanced, to my thinking it seems pointless to specify that one stage have one stage has active elements with closely similar gains whilst it's only a 'nice to have' on following stages.
    Also, bearing in mind the significant role of power tube screen grid sag in a tube guitar amp, how many tube vendors check for similar g2 characteristics in their matched power tube sets?
    If the screen grid draw differing currents under high signal conditions, then (in an actual amp, rather than test kit) their plate transfer curves get pulled down to differing points.
    And I wonder what specs are used for matching the number of turns of each leg of the PT primary, both back then, and now?

    Even assuming that an LTP is used, it seems a rather blinkered approach to focus on the phase splitter tube and ignore the above.

    And, as is the case with many amps (lottsa Princetons, 5E3s etc out there), if an LTP is not used as the phase splitter, then the whole concept of a 'balanced PI tube' is nonsensical.

    But, if it gives tube buyers a warm fuzzy glow to spend a little more, and gives test vendors gainful employment, then I suppose it has some benefit.
     
    Last edited: Apr 8, 2017
    Advisable Owl likes this.
  14. HotBluePlates

    HotBluePlates Member

    Messages:
    6,034
    Joined:
    Mar 27, 2017
    The paraphase inverter I'm talking about applies preamp signal to one triode, whose output then goes to one output tube grid (labeled "inverted" in the image below). But that triode output is also applied to a voltage divider (loss approximately equal to the triode's gain), which then feeds a 2nd triode (labeled "non-inverted" in the image). That 2nd triode is mainly there to create an opposite-polarity output.

    [​IMG]

    The flaw in the regular paraphase is the assumption that the fixed resistive voltage divider will have exactly the same loss as the amplification provided by the 2nd triode. That's not necessarily true from tube-to-tube, or even as a single tube ages.

    There are self-balancing paraphase inverters that use a feedback mechanism to jigger divider loss to be equal to 2nd paraphase stage gain, but those seem to be less common in the classic amps. I (perhaps wrongly) guess by that time the amp manufacturers quickly moved on to long-tail inverters.

    I agree it would be pointless to spec matched sections of a phase inverter, then not care about matching in the output stage. But there are a number of conditions to meet to call output tubes truly "matched".

    My personal approach is to look for "reasonably close" idle current of the output tubes (maybe up to 5-7mA difference in idle current), because wildly mismatched idle currents erode some of the power through-put capability of the OT. However, I don't need "max clean output power".

    What's more, the Radiotron Designer's Handbook 4th Edition (RDH4) included a section on the effects of tube matching. They tested the effects of mismatching by usuing a push-pull pair made of two different types of output tubes, one having half the Gm of the other tube. This was done to simulate the worst-case of one new tube and one "dead" tube. The result? Power output was down a little, and the output section had 5% harmonic distortion. RDH4's verdict was that mismatching is not a critical problem, but that clean out is reduced and distortion is increased.

    How many guitarists completely shun distortion? (5% THD was considered at the time of RDH4's publication to be the standard for hi-fi-reproduction)

    Power tube screen grid sag is mostly a non-issue in guitar amps, except where builders/tinkerers purposely install oversized screen resistors to induce sag (I can prove this by analyzing max power conditions of a 6L6 output stage if you wish).

    This is a misapplication of something published in Guitar Amplifier Overdrive, where the relevant section is discussing preamp pentodes. Throughout the book, they're slamming the snot out of the gain stage they're discussing, with grid input signal as much as 20 times the bias voltage of the tube. If you do that with an output stage, you're so far into control grid current and grid blocking distortion that any influence due to the screen grid is far overshadowed.

    What is relevant with respect to screen grid current and resulting screen voltage drop is that the potential peak plate current for the tube can be reduced/constricted.

    Some aspects of that book were challenged on another forum, but the person referencing the book hasn't shown back up in the past couple months, so no telling if we'll reach a shared understanding. I hadn't read the book until this precise claim of output tube gain/distortion changing with a screen voltage shift was put forward. So I and another very knowledgable member bought & read the book. Sadly, unless you know as much on the subject as the authors, it's hard to catch how the information presented only applies under specific conditions that are not guaranteed to apply to every amp or every player's situation.

    Number of turns of each half-primary. That's it, and that's all there ever was. When you come to really understand how transformers work, you'll know why that's all that really mattered anyway.
     
    guitar007 likes this.
  15. Malcolm Irving

    Malcolm Irving Member

    Messages:
    827
    Joined:
    Aug 27, 2014
    Location:
    Derbyshire, UK
    It would be good to see the analysis, as I think most people believe that sag in screen grid voltage at full output is significant.

    EDIT: Maybe in a new thread - we seem to have diverged from the original topic (as usual). :)
     
    HotBluePlates likes this.
  16. pdf64

    pdf64 Member

    Messages:
    5,790
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    It seems valid to me to view the 1st CC stage in the image, rather than being part of the paraphase, to be the last gain stage of the preamp? As it could equally serve a SE power tube as well as a paraphase, cathodyne or LTP p-p power amp.
    Whatever, the point that specifying that the 2 sections are matched being pointless still seems to stand.

    I think that analysis was for triodes in class A; I suspect that the impact would be somewhat greater in the case of pentodes/beam tetrodes in AB, as is typical in our application.

    Hence the call for matched sections in a tube intended for a guitar amp phase splitter seems misapplied, which I think was the thrust of the opinions expressed in posts #2 and #12.

    Thanks, I'd be grateful for your take on that.
    It would be interesting to compare it to a typical EL34 p-p amp, with a ~400ohm dcR choke in the HT g2 supply node and 1k current limiting resistors; I suspect with such amps, g2 sag is significant, and perhaps a key contributor to their characteristic response.

    Yes, hence for power tubes to be truly matched, it may be beneficial for their g2 characteristics to be included in the comparison.

    Yes, but it may be noted that may not equate that to the resistances being equal, as that would not necessarily occur unless a bifilar winding technique was employed.
     
    Advisable Owl likes this.
  17. HotBluePlates

    HotBluePlates Member

    Messages:
    6,034
    Joined:
    Mar 27, 2017
    I fully agree "sag" can be related to dropped screen voltage. I disagree with the original assertion that G2 voltage changes in the output tubes significantly alters the operation of a typical guitar amp.

    A person dropped that nugget on a different forum, citing the book I noted. I (and another forum member) withheld comment until we actually bought & read the book. The authors were talking about preamp pentodes, which are fundamentally different in application & construction/characteristics. But the problem is people read modern books and don't grok the practical realities of designing with tubes, then over-inflate the importance of what are really minor issues.

    I do agree this is best addressed in its own thread.
     
    Last edited: Apr 9, 2017
  18. HotBluePlates

    HotBluePlates Member

    Messages:
    6,034
    Joined:
    Mar 27, 2017
    We both agree on this. On another forum, we developed a way to intentionally unbalance a long-tail pair to reduce output power before distortion. I personally don't worry about matched sections outside of hi-fi & recording equipment.

    I can see how it can be useful to think of the divider & 2nd half of the paraphase could be seen as a "functional block" apart from the 1st triode in a paraphase. Especially if you're starting at the output tube grid and planning the gain requirements of the preamp to drive that output tube.

    I often look at any element included in the negative feedback loop from the speaker as part of the output section. That would put the 1st triode of the paraphase as part of the output section in the 5B6 Bassman.

    Separately, analyzing Leo Fender's first big patent is more intuitive if you think of both triodes as being included in the paraphase. The tweed Tremolux sends anti-phase outputs from the plates of the paraphase triodes to drive the output tubes (like every inverter). But Leo's invention was to apply a tremolo signal to the shared cathode resistor of the paraphase section, to avoid pumping heard when applying trem at a preamp tube cathode. The trem modulates both outputs but is a common-mode signal, and the output tubes only respond to differential signals. So trem is applied, and pumping minimized because the trem signal is effectively discarded at the OT.

    You are right that RDH4 demonstrated with Class A triodes. But class of operation is immaterial, because the issue is whether the composite characteristics formed are straight & parallel. Below is an image of this for 2 identical tubes (dashed lines are original plate curves, solid lines are the composite characteristics of the push-pull stage).



    This was repeated in RDH4 for two different tubes, which produced the composite characteristics below (which are surprisingly only a little bent):



    RDH4 doesn't explain the warning that "effects of mismatching with pentodes are more serious than with triodes" or that "Class AB is more sensitive to mismatching than Class A". Without going into a more-extensive derailment, it really boils down to how much distortion is no longer being cancelled by balanced push-pull operation, and how that erodes some of the advantages of pentodes over triodes.

    So I'm just saying if you don't need "Clean as possible, as Loud as possible" then precise matching & balancing is irrelevant.

    No problem. I'll begin another thread if it hasn't happened already.

    I've never seen a single vendor of tubes that does this. Maybe I've overlooked someone...

    Correct that DCR for each half-widning is unlikely to be equal. But fundamental properties of a transformer are based on numbers of turns and voltage or current, for inducing a secondary voltage or current. DCR is mostly about copper loss heating the windings, though DCR differences between halves of a push-pull primary are made a small % of total primary impedance.

    I just measured the Heyboer OT in my 5E3 Deluxe clone. There is 208Ω from CT to one end of the primary, and 207.5Ω on the other half-primary. The 0.5Ω difference is nearly-nothing, and I'm gonna assume it's not bifilar. But even if there were 100Ω difference between the half-primaries, that would be very little in the face of the 4kΩ reflected impedance for a half-primary on this 8kΩ OT. 100Ω/4kΩ = 2.5% imbalance, and you can't count on anything else in the circuit being better than 10-20% tolerance, without taking special steps.

    Further, when Frank McIntosh invented the bifilar transformer he sought a way to solve the contradictory requirements of high incremental inductance (lots of turns; needed for good bass) while reducing distributed capacitance in the transformer (more turns = more interwinding capacitance; kills treble response). "Balance" and "DCR" weren't key drivers for the invention.
     
  19. 71strat

    71strat Member

    Messages:
    7,968
    Joined:
    Oct 7, 2013
    The only reason I ever have mine matched is, I figure theyre going to deteriorate at different rates, and wont be balanced for long anyway. So I usually get them as evenly matched, and testing as strong as I can find.

    Also if were talking 5% distortion with mismatch, which is normal for amps ect, but whats matched going to be 4%? 20% cleaner? Maybe? Maybe Less?

    What is Factory Spec for Triode Matching ??? How close/far %%% wise, is Tolerable to be at full factory spec, but not match? I know many tubes were screened from the factory. I'm fine if they test really strong, x NOS, but don't match.

    Heres a good little look at Brent Jessie testing some tubes.

     
  20. HotBluePlates

    HotBluePlates Member

    Messages:
    6,034
    Joined:
    Mar 27, 2017
    Screen voltage sag thread posted here. The thread kicks off with an in-depth look at a 6L6 output stage and 470Ω screen resistors; I'll need to repeat for EL34 (which draws greater screen current) and 1kΩ screen resistors.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice