• Re: srpp or cathode follower? Paul Joppa 15:31:49 03/31/00 (3)
  In Reply to: srpp or cathode follower? posted by YH CHOW on March 31, 2000 at 07:49:20:

  There seems to be an emerging consensus that if you use a cathode follower, it is preferable to use a CCS load on the cathode. Other than that, there seems to be no real consensus out there on this question. Fortunately, it is easy to try them all in basically the same circuit, and see which you like best.

  The differences are probably due to the different harmonic distortion spectra of the various arrangements. Consequently, your preference may well depend on the distortion spectra of the rest of your system, of the particular tube you choose, and the output level (preamp or driver). There have been a few analyses and measurements published, mostly in Glass Audio over the years.

  I know Thorsten among others has experimented extensively. Not everyone has the same tastes, for instance there are a few who really like cathode followers (which he does not!), but maybe we can get him to describe the differences in sound. Thorsten, are you reading this thread?

  Incidentally, the SRPP and mu-follower are two different circuits. If you add in the possibility of current sources or chokes for loads and the choice of whether to bypass the common-cathode stage bias resistor, there are quite a large number of possible circuits.

  
  
  

  ---

  • The Sound of Cathode Followers.... Thorsten 04:26:27 04/03/00 (2)
    In Reply to: Re: srpp or cathode follower? posted by Paul Joppa on March 31, 2000 at 15:31:49:

    Hi there,

    Yes Paul, I have experimented.

    Here first the Topologies and Valves used:

    +B in all cases 200V, generated via Solid State Rectification, a "Zener-Follower" Regulator and then a LC Filter of 10H/ 10,000uF per Channel with 15uF Film Capacitors as bypass, full starwiring in PSU and Ground, including phono-stage. Heaters DC with additional RC Filtering for each Valve. Hardwired construction.

    Each Circuit was operated into a IEC load (10k/1nF) in order to "increase" the sonic signature.

    Amplifier was a modded Marantz Solid State unit with battery PSU. Speakers where DIY Copies of Wilson Audio Watt/Puppy. Sources where a heavily modded Marantz CD-67 and an Oracle Delphi Turntable. Mostly Music from LP was employed as the sound was by general consensus of all testers "in a different class" for LP over CD.

    Some short tests with a "worst case" selection of really bad CD recordings where included for completeness. Testing was done "semi-blind", meaning that only I knew the topology used, but there was no attempt to disguise anything. Switching over between topologies took about 30 minutes.

    1) ECC88 in Common Cathode Circuit, both sections in parallel 9.75k Anode Resistance (39k X 4 in Parallel), -2.7V fixed Grid Bias Voltage via Battery.

    2) ECC88 Common Cathode Stage followed by a Cathode Follower, DC Coupled, Balanced Anode/Cathode loads of 19.5k (39k X 2 in Parallel), common Cathode Stage with -2.7 fixed Grid Bias Voltage via Battery.

    3) Like 2, but Cathode Follower with Current-Source in "tail" using IRF840

    4) Like 3, but Cathode Follower with added IRF840 Cascode as per Allen Wrights "Super Linear Cathode Follower"

    5) ECC88 SRPP, 390 Ohm Cathode Resistors.

    6) ECC "Mu-Stage" (not sure about exact values, I think as per Allens Book)

    Sound Comments:

    1) Sounded natural, open, excellent Detail and Soundstageing. The sound is slightly "warmed up" compared to a direct bypass connection from CD or Solid Stae Phono, but this is almost unnoticable. This was unanimously considered best by all participants.

    2) Sounds very veiled, muddy, soundstage collaps, detail gone. Sound was however preferred by some on "edgy" CD-Recordings, but allways quite distinctly disliked on Vinyl via either Phono-Stage. This was least liked of all topologies if results where averaged.

    3) Sounds a little better than 2), still very veiled and muddy.

    4) Sounds now quite open, but there is now a very noticable edge, the sound becomes strident and unpleasant. Hung Jury, the Naim/Solid State guy's liked the sound, I personally thought it overall worse than 2), but not much.

    5) This sounded initially not too bad, decent Detail and Soundstage as well as Detail, but there is something weired going on. The sound is often strident and unpleasant, there was little agreement as to actually how it sounded and assesment varied strongly with Music. I suspect that SRPP could work just fine in lower resolution CD only Systems, but on Vinyl it was generally ranked notable Lower than 1)

    6) This was very similar to 5), but lower in Detail and also overly strident and agressive.

    The general consensu was that overall the simple common Cathode Stage described as 1) was the best, by a notable stretch.

    I later carried out tests (without pannel) where I compared the parallel ECC88 Stage against using a single section of the 5687 all other parameters unchanged. This produced an even better, more open, detailed but coherent sound with much better Bass.

    If one where to compare the single 5687 Linestage against 2 - 6 the distance would have been even greater.

    On a ranking from 10 to 1 points, with 10 Points being best I'd personally rank as follows:

    5687 Single Section Linestage - 9 Point
    ECC88 Parallel Linestage (1) - 7 Points
    ECC88 SRPP and Mu-Stage (5 & 6) - 5 Points
    ECC88 Gainstage/CCS CF (3) - 4.5 Points
    ECC88 Gainstage/CF (2) - 4 Points
    ECC88 Super Linear Cathode Follower (4) - 3 Points

    Further notes:

    It is noteworthy that Allen has since the above test (of which was aware if not in detail) made changes to the SLCF Circuit and that one Can use a Valve as upper Cascode Sections, as well as Current Source. This will (from experience with similar Circuits) raise the score from the SLCF to 6 maybe even 6.5, bringing it close to the classic Common Cathode Stage but with better Drive Capability.

    I have myself designed a variation of the Mu-Stage which offers the (optional) ability to dispense with ALL Solid State components AND with the Mu-Stage Coupling Cap, as well as incorporating all the SLCF Improvements to the Cathode Follower Section.

    This not only measures so good that it frightens all the Solid Stage guys (> 50V RMS into the IEC Load with < 0.05% THD, all 2nd Harmonics, < 100 Ohm Z-Out - beat dis) without Negative Feedback and uses only a 5842 and a 5687 as well as a 30H Audiograde Inductor (or a J-Fet as Current Source). In addition, as far as I can tell, it sounds as good as the normal Common Cathode stage using the same Valve.

    Still, I no longer use this circuit myself, having never really found an application where I really needed that kind of Drive Capability and normal common cathode stages using suitable valves have fewer parts, less issues around Heater/Cathode Voltages (the Super Mu-Stage needed two seperate Heater circuits and still run the upper valve near it's static limits) and sounds no worse....

    Later Thorsten
    
    

    ---

    • Thanks! was: The Sound of Cathode Followers.... Paul Joppa 12:51:10 04/03/00 (1)
      In Reply to: The Sound of Cathode Followers.... posted by Thorsten on April 03, 2000 at 04:26:27:

      Thanks, Thorsten. I hoped you would post these results - very valuable contribution to our knowledge base. I'm going to save this one.

      We had a VALVE meeting yesterday, and listened to Richard Riley's 2A3 amp which uses a 6SN7 in a slightly simpler arrangement - I assume you are doing basically an SRPP with a choke for the upper bias resistor. This is his improvement on my original circuit which had a conventional mu-follower. It's simpler and it sounds better, even though he now has this extra choke (actually the primary of the cheap Allied interstage) hanging by its wires outside the chassis. I believe he is taking the output from the lower tube plate, which has less drive capability but seems to sound better. I don't know if the lower cathode resistor is bypassed - Richard, are you reading this?
      
      

      ---

      • The Super Mu Stage (No Bovine Jokes please).... Thorsten 13:22:25 04/03/00 (0)
        In Reply to: Thanks! was: The Sound of Cathode Followers.... posted by Paul Joppa on April 03, 2000 at 12:51:10:

        Hi all,

        Now, I can't upload pictures and at that time of the night I sure ain't gonna ASCII Draw this circuit, but no Paul, it's a little more complex. The Circuits is like this:

        Starting from Bottom to Top (The Bottom Up approach):

        At the bottom is a 5842/417A or similar, operated with unbypassed Cathode Resistor and some Positive Feedback from the whole output of the Circuit to Cathode (Remember negative Feedback = bad; Positive Feedback = good) usually only 2, perhaps 3db PF though.

        The Anode Leads to the Grid of the first 5687 and also to either our Choke or to the Cathode of our J-Fet CCS (2SK147 works great BTW). The Choke is for Purists, sounds slightly more natural, the FET has a bit more Detail and a taughter bass (more like being hit by a hardwood 2X4 rather than a Spruce one), overall, I like both.

        Anyway, either we have a Resistor in series with our Audio-Choke to set the Current of the 5687 or we have a solid State CCS taking care of it, both work well. In either case the free end connects to the Cathode and no, C4S need not apply, FET's please.

        The output is indeed taken from the Cathode of this 5687, the whole stack runs at around 160V per Valve and around 12 - 15mA (sounds better than higher currents, tried it).

        So far so good and so boring and done so often. It sure sounds decent, but there is one more step to go. We still have another 5687 Halve looking for a Job. And here is waht we make it do:

        We CASCODE it with the other 5687, using a apir of 1M Resistors and couple one of the Resistors between +B and the Grid of the second 5687, the other between Grid of the second 5687 and Cathode of the First. This Resistor is also bypassed by any decent old Foil Film or Oit Cap, 0.1uF or larger.

        Now we need around 500V +B for the Whole stack, the Current through our first 5687 is being kept pretty constant for AC by our Inductor, the Inductor is "amplified" by the Mu of the 5687 (around 17), so the 5842 will see an anode Load Inductor of around 500H, well out of the Way even with an unbypassed Cathode Resistor.

        The Cathode Impedance of the 5687 ( < < 200 Ohm) Sees a 30H Choke in parallel with the load, again, well outside of any area of concern. Best of all, the Modulation of the Anode/Cathode Voltage of the upper Valve is also being significantly reduced, instead of being equal to the Output Voltage, we now have it equal to around 0.06 of the Output Volatage, significantly reducing the remaining non-linearities.

        BTW, the Problem Allen Wright had in the original Version of this Circuit with the ECC88 was the Capacitance of the Hexfet (several nF) loading the Cathode Follower and (obviously) being severely modulated by the Signal, hence the unpleasant agressive sound.

        Anyway, I hope this little gem Helps.... It's a brilliant Driverstage for a 300B. It's almost as good as a simple EL84 operated as Penthode, but it has more ooompf than the EL84 truth be told. The Super-Mu Stage will drive any Grid not only into Class A2, but probably into Class A3, if there was such a thing.... ;-) The EL84 is good for Class A1 only.

        Later Thorsten
        
        

        ---
• Re: srpp or cathode follower-differences obesrved- Quest 08:29:12 03/31/00 (1)
  In Reply to: srpp or cathode follower? posted by YH CHOW on March 31, 2000 at 07:49:20:

  Hi,

  First off I would like to correct you for the several notions:

  (1) SRPP (aka "Shunt Regulated Push-Pull") is not a Mu-stage. They operate in a slightly different manner. Although they seems to operate in a somewhat similar topologies, current-control-wise Mu stage is somewhat closer to a constant current mode.

  (2) While some people see the upper half of an SRPP/Mu-stage as a current regulator, others like Thorsten looks at it as a cathode follower stage. While such topologies has it's merits of high input impedance and extremely low output impedance, high PSRR, high gain and so on. There are things such as high-order distortion yet to be dealt with.

  
  (3) A CF(aka "Cathode Follower") offers < 1 gain. In other words it will require an additional gain stage before it in order to make it at least a unity gain. By itself CF stage offers no gain but loss.

  (2) A CF stage can sometimes be perceived with feedbacks.

  (3) CF stage offers high current, low output impedance driving capability.

  (4) CF stage may not be linear throughout it's useable bandwidth.

  Look: To me low output Z may seems appealing to many one of us but realizing that there's so much compromise involving when using these topologies leaves me little to be desired. While CF stage can be "tricked" into better linearity over it's usable bandwidth, SRPP/Mu stage is more difficult to obtain satisfying results (IMHO). They usually sounded diffused(not focused) and so on, despite their test figures looks good on the paper.

  If you have the guts to do so try using conventional plate-out (Anode load) topology with a radical approach. You'll be surprised how much better they sound in most applications if applied properly.

  Good luck!

  Quest
  
  

  ---

  • Re: srpp or cathode follower-differences obesrved- YH CHOW 20:12:05 04/02/00 (0)
    In Reply to: Re: srpp or cathode follower-differences obesrved- posted by Quest on March 31, 2000 at 08:29:12:

    What would be a "radical approach" to plate/anode loading?
    I can think of 3 variations:
    -resistor,
    -constant current source using
    tube or "real" current source using say, pnp/p-channel mosfets,
    -transformer/choke.

    Now that you mentioned it, the tube constant current load
    is not really a constant current load in that the plate of the
    lower amplifying tube sees the cathode of the top tube load. The
    cathode is really a low impedance input, is it not? It seems to
    me that a true constant current load is possible only with
    a pnp or p-channel device at the plate of the amplifying tube.

    YH
    current load would really be possible
    
    

    ---

This post is made possible by the generous support of people like you and our sponsors:

Sonic Craft

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---