Dug myself a deep hole this time...

Discussion in 'Tube Audio' started by trainbuftony, Dec 22, 2017.

  1. trainbuftony

    trainbuftony Electron Herder Subscriber

    So I pulled one of my amps off the shelf to play with it for a day. The amp is a early LK72 Scott. I thought I had it all working but when I pulled it down I had a problem which resulted in a hilarious troubleshooting experience.

    On the phono setting, the gain was wrong between channels. One side was like 6 to 10 times the amplitude of the other and I spent nearly all day trying to figure out why. I learned something (or realized something I already knew, and I'm going to learn a lot more before I'm finished with this. So I spent all day trying to figure out why the channel with the lower gain wasn't performing like the channel which sounded great! I replaced nearly every resistor in the phonostage with the best matching resistors (channel to channel) which I had in my stock. None of them measured substantially off and I had my doubts about replacing them. I replaced each capacitor 2 or 3 times. I took the pins for the suspect triode out of the socket and cleaned them individually then put them back in the socket and still had low gain on that channel. I swapped tubes and tested them 2 or 3 times. Tubes tested good each time, swapping them produced the same result, the same triode section producing way too little gain. At this point I've been focusing on this 1/2 of a 12AX7 for about 6 to 8 hours and the frustration is mounting. I'm ready to rip the socket out of the chassis and replace it, with little confidence in that course of action.

    Deep breath. Chill out Tony, this is supposed to be fun...

    Late that evening I found my culprit, My thinking was off, and I learned a lot. You already may know where my thinking was off. I had been focused on the triode with the lower gain. As it turns out that triode was working as designed. (or close enough to it) The fault in this case was found by turning up the amplitude on my signal generator. I noticed that the "good channel" developed several distortions in the wave form, while the Lower gain channel amplified the signal and put out a nice clean sine wave. This is where it came to me. I remember reading about how feedback networks lower distortion, but also lower gain. I thought to myself, "I had better pull up a schematic and see if this thing has any feedback around this stage." It of course does have feedback around the stage, and not only that, it has a PEC with the feedback components inside! FURTHERMORE!!!! It doesn't just have phono-stage feedback components inside, the PEC contains the feedback components which taylor the RIAA and NAB tape equalization curves for the two inputs. An isolated section of the tape/phono switch changes the EQ setting for whichever input is selected by the other sections of the switch. They are ganged.

    I have a broken lead coming out of one of these PECs and therefore not providing feedback around the stage for one channel, resulting in higher gain, (not actually good) and distortion, as well as RIAA non compliance!

    I thought about this for a day or two, and decided what I want to do. The PEC is done for. I'm not going to try and replace it. I'm not going to try and rebuild it. I'm going to remove the NAB tape equalization setting and the switch, from the feedback loop, and replace the feedback components within the PEC with components for a dedicated RIAA stage. This is where I come to the end of my Knowledge and abilities and I ask the experts for help.

    I took the liberties of drawing up the schematic as it stands and another of how I think it should be wired "without" the NAB tape section. I would love it if someone could explain to me why I'm completely wrong or what would be vastly better than this and why.

    Here is the phono-stage drawn as wired. The PEC feedback components are drawn in a box toward the bottom, and the switch is shown in the RIAA position. I want to eliminate the switch and go RIAA on bot inputs.
    [​IMG]



    This drawing shows the PEC and the switch eliminated, and only the components for RIAA left in the network. Please check my work here, I could be full of crap. Basically I want someone who knows way more than I do to verify that this will produce reasonably close RIAA compliance, or to show me what I did wrong, and perhaps suggest an alternative course of action, or some reading material on the topic. I lack the skills to design and model ultra tight EQ tolerances and I've never encountered a problem quite like this.
    [​IMG]

    Thanks in advance for your help, and happy holidays.
     
    Leland47 and bluemooze like this.
  2. AdamAnt316

    AdamAnt316 Collector of heavy things Subscriber

    Messages:
    1,752
    Location:
    MA
    My personal opinion is, don't try and reinvent the wheel. Recreate the original PEC using discrete components on perfboard, and install it as the original would've been. Note the performance, and see what you think. Perhaps from there you can go ahead with your modification ideas, but this stuff was designed the way it was for a reason, and I think it's a good idea to establish a baseline before going ahead and making radical changes.
    -Adam
     
    Leland47 likes this.
  3. knockbill

    knockbill Addicted Member

    Messages:
    6,906
    Location:
    SE PA
    Baseline is always a good idea, you can determine which way you are headed... and if you keep it stock, you have the schematic to refer to if something goes wonky...

    I thought this was leading to the error I've made with unbalanced channels,,, forgetting to adjust the input level pots!!!!
     
    trainbuftony likes this.
  4. jaymanaa

    jaymanaa AK Subscriber Subscriber

    Messages:
    54,473
    Location:
    OZ
    I agree with the posters above. I'll add that if it's going to be used, I will always "build" new PECs with parts on a small RS board. Especially with something as critical as phono stage tolerances. There's a lot going on there between the sections of those 12AX7s. I figure as well as accuracy, it can't hurt the overall sound, or tone either to use decent parts instead of some junk buried inside that brown concrete. :) I think all those things were, were growing pains toward modern printed circuits. I think we would have seen a lot more tubes with internals capacitors and such too, if transistors hadn't taken off as fast as they did (area 61?).
     
  5. trainbuftony

    trainbuftony Electron Herder Subscriber

    I guess what I'm asking is can I eliminate the 10 meg, and 100k resistors attached to pin one (tape equalization) of the PEC, in favor of full time RIAA, via the 560k resistor and 560pf cap, bypassed by another 150pf cap? Eliminating the tape input, in favor of an additional phono input is the end goal here. Also, how would I go about establishing a baseline performance analysis of the phonostage as it stands? Does that mean testing at every frequency in the audio spectrum through a reverse riaa and graphing the results? I'm not sure the one PEC I have left is even working well enough to discern anything useful by analyzing it's performance.
     
  6. Dandy

    Dandy Super Member

    Messages:
    4,088
    Location:
    North West, UK
    They are pretty small circuits, so I agree that you should rebuild them as in the schematic. I'm about to rebuild the busted PECs in my Dynaco SCA35. The advice there is match the caps exactly.
     
  7. nj pheonix

    nj pheonix AK Subscriber Subscriber

    Messages:
    8,395
    Location:
    New Jersey
    Scott presents it slightly differently in the schematic (than Fisher or Eico). I have to agree with the group. Tiny piece of breadboard, build it discreetly (make 2 , leave good side in until you're satisfied replacement is working properly. I use really big shrink to cover the ones i've done and i color code the leads so i don't screw up.
     
    trainbuftony likes this.
  8. trainbuftony

    trainbuftony Electron Herder Subscriber

    I ordered some parts from mouser to build these out. I already have breadboard, I already have 100K 1% resistors.
    Resistors:

    CMF6010M000FKEK112 10 meg
    RN65D5603FB14 560Kohm
    100K, I already have but here is the number anyway.
    CMF55100K00DHEB 100Kohm

    Capacitors:
    FKP3J001502B00JSSD 150pf
    PHE450PK3560JR05 560pf

    I spent 8 bones to ship $5 worth of parts. I'll keep you all posted as to how it works out. If you see any obvious mistakes point them out now that I've already made the order...
     
  9. Dandy

    Dandy Super Member

    Messages:
    4,088
    Location:
    North West, UK
    Hope you ordered two of each?
     
  10. trainbuftony

    trainbuftony Electron Herder Subscriber

    Yes 2 of each. I was tempted to order more to make the shipping worth it, but that gets out of hand, and turns into a $350 parts order in no time... Even though I only needed $5 worth of parts...
     
    drtool likes this.
  11. dcgillespie

    dcgillespie Fisher SA-100 Clone Subscriber

    Messages:
    9,028
    Location:
    Ball Ground, GA
    Train -- Sorry I'm late to your party, but you will in fact need to retain the 10M and 100K resistors, which are effectively connected in series and then collectively placed across the 560 pF cap in the Scott design. This resistive element in the NFB network basically controls the gain at 20 Hz in the RIAA playback response curve, and works in conjunction with the 560 pF cap to establish the 3180 uS turnover point in the curve as well.

    Those familiar with RIAA NFB networks from other manufacturers may be more used to seeing resistor values of around 4.7M (typical for Dynaco) or 2.2M (Fisher) in this application of the network, but Scott employed a significant amount of positive feedback around the active gain stages of their phono preamp design, which ups the Open Loop Gain (relative to the NFB EQ network) significantly. Therefore, the value of this resistor is also raised appropriately over those in other designs as well which then allows the Scott design to enjoy the benefit of the enhanced overall gain it provides over the other offerings. It's value becomes so much higher that it would be easy to think that it plays only a very minor roll (if any) in the RIAA EQ function, but it most surely does.

    I should also add that in my work with the Scott design, I found that by raising the value of the 150 pF cap in the EQ network to 180 pF, and increasing the value of the 560 pf to 635 pf, produced very high compliance with the published RIAA curve.

    I hope this helps!

    Dave
     
  12. trainbuftony

    trainbuftony Electron Herder Subscriber

    I nailed it! I built a replacement PEC on a piece of perf board and i installed it as the feedback loop of the distorted, high gain section. I hit it with a half dozen frequencies between 20 and 20khz and the amplitude after the phono-stage was close enough to the original PEC that I am confident that I nailed it. (measured on my Tektronix scope) I'm going to eat something and I'll post some photos after I get the other channel done.
     
    tonyk likes this.
  13. Dandy

    Dandy Super Member

    Messages:
    4,088
    Location:
    North West, UK
    Nice work!
     
  14. trainbuftony

    trainbuftony Electron Herder Subscriber

    Here are the photos as promised.
    The board:
    [​IMG]

    Both boards installed:
    [​IMG]



    Gain is da same channel to channel:
    [​IMG]
    [​IMG]
    It's about time: I have had an unreasonable fear of feedback loops and this diagnostic procedure has helped me to start to overcome that fear.
     
    AdamAnt316, crispycircuit and drtool like this.

Share This Page