6567d689   
   From: info@turneraudio.com.au   
      
   On Sunday, 6 January 2013 05:08:30 UTC+11, Andre Jute wrote:   
   > A positively thermionic New Year to all RATs. Andre Jute   
      
   Fetch me my heart pills, for we have a visit by Andre.   
      
   All the best for 2013.   
      
   Meanwhile, as time allows, I continue to solder and cycle my way towards ETA   
   of 2042, with help of a good Hubot, and yet more invented medical tricks if   
   allowed by Unkel Fate, and the passing traffic does not squash me to a pulp   
   while I am cycling about    
   in a state of pleasantly transposed conciousness.    
      
   In a previous post I mentioned using a single EL34 in pentode mode to power 2   
   complementary pair source followers in class A, 2SK176, 2SJ56. I've now got   
   bugs ironed and produced an ULTRA SIMPLESTICA amp with following character,   
   Noise < 0.5mV, DC offset < 10mVdc, Rout, class A, 0.07 ohms for DF = 85 with   
   RL = 6 ohms, THD > 0.1% at 10Vrms, BW = 4Hz to 300kHz, unconditionally stable   
   at any power and with any RL type of value or no RL at all. Square wave =   
   excellent, with low    
   ringing with pure C loads. Amp makes 15W max, 6 ohms, Pdd at idle = 32W total   
   using +/- 16Vdc rails at 1A each. Excellent overload and high tube voltages   
   are excluded with gate-source zener diode clamps and with use of 2uF cap   
   coupling to each gate with    
   560r series gate R.   
   At gross overload, caps develop negative charge like a tube amp to cause   
   mosfets to be become biased for class C. But for normal use, bias is self   
   regulating with 0.47ohms between each source and Vo.    
      
   Prototype amp has 2 chassis, with Vdc applied to EL34 heater. +/- 16Vdc rails   
   have CLC with 10,000uF - 10mH - 15,000uF for Vr < 4 mV. B+ = 350Vdc, well   
   smoothed. Ia = 25mA and is delivered from B+ to anode via 40H plus 2k7, with   
   Ea at 265V. So RLdc is    
   extremely high value. Ek = 20Vdc, and cathode bias network = 660 ohms plus   
   470k. Screen is fed 4mA via 27k with Eg2 at 250Vdc, and screen is bypassed to   
   Vo using 8uF el-cap.    
      
   EL34 is tricked into thinking its a triode with low THD. OP stage gain = 0.9,   
   so if 9Vrms appears at output, 10Vrms is needed at gates and provided by EL34   
   anode. Screen signal voltage = Vo = 9Vrms, slightly less than anode voltage.   
   Therefore the EL34    
   has gain just slightly higher than triode µ of 9. In fact EL34 gain averages   
   10.    
   So EL34 is acting similarly to a triode strapped tube where all anode voltage   
   is applied as FB to the screen. In this case the mosfet OP stage is included   
   in the simple loop and because RLdc plus gate bias resistance totals about   
   45k, then the gain of    
   screen is screen gm x ( Ra // RL ) which in this case is about 14x. Hence any   
   THD produced by mosfet stage is reduced by factor of about 0.07, and source   
   follower Rout without FB of 1.3 ohms is reduced to 0.07 ohms.    
      
   I know Andre likes simple amps, well, here is such a one, just 3 devices in   
   total, and yet measuring better than just about any other 15W amp with only 3   
   devices. Screen FB was used in 1955 in tube amps to avoid trying to apply GNFB   
   from OPT sec to    
   cathode of an input tube. It can be applied to an input pentode if you want,   
   but here in my case I have JUST ONE TUBE that does everything by producing   
   high current low Z drive to mosfet gates with high C, and while accepting FB   
   in a simple easy loop    
   that is similar to simple triode strapping.    
      
   I have to complete web page about it all and will post a notice when its ready   
   for ppl to view.    
      
   Included will be another version using EL34 as pure triode strapped, and a   
   paralleled 6CG7 normal input tube with FB applied conventionally. This way the   
   very slight THD of 0.1% per 10Vrms of EL34 in triode PLUS the 0.4% of mosfet   
   OP stage is ALL reduced    
   by whatever factor I want; if Vin is allowed to be about 1.5Vrms, then THD   
   should be < 0.05% at 10Vrms output with say a 6 ohm load.    
      
   For those wanting higher power, then use 4 mosfets each dissipating 20W, and   
   use +/- 25Vdc rails from 18.5Vac-0-18.5Vac PT winding. Expect extremely fine   
   sound. PO max would be 35W, class A, with load of 6.8 ohms. In such a case,   
   each mosfet has RL = 27    
   ohms, very easy for them, so in follower mode the gain reduction is high, and   
   THD even lower than I am getting now in this present experiment.    
      
   In 1990s there was a maker in West Oz, Micrex, who made a commercial hybrid   
   amp with mosfet output and one EL34. But they had bjts used for a differential   
   input stage and mosfet OP stage was rated for 90W in low bias class Ab. I once   
   was asked to bring    
   an 8585 amp ( with 4 x 6550 in class A in each channel ) to the owner of the   
   Micrex and allow a comparison. I sat in silence for 3 hours while the owner   
   changed CDs and amps. Some good Maggies with ribbon HF were used. Piano was   
   especially good.   
   But after 3 hours, the guy says, "Well, your pure tube amp is vastly superior   
   to the Micrex". I left him with a quote and returned home with my amp. A week   
   later he said he wanted to get a full tube amp, and for more then what I   
   quoted, but from his    
   regular guy who had fixed and modded his other gear. OK, I was used, but a   
   point was sort of proved. My present amp with class A mosfets and not a single   
   damn bjt seen anywhere may sound as well as a fully well done tubed amp.   
      
   Patrick Turner.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)