a96a0668   
   From: spam@spam.com   
      
   On Sun, 6 Mar 2011 01:00:05 -0800 (PST), Patrick Turner   
    wrote:   
      
   >   
   >My experience with tube Hartleys is that the tank circuit gets its   
   >cathode current input to drive the grid via a very non sine wave   
   >current pulse on each cycle. I found there is more harmonic voltage   
   >trash at the cathode than at the grid.   
   >Quite a few circuits show the cathode input on one tap, with grid   
   >drive on another tap up the coil, and the high impedance at the top of   
   >the tank is then not loaded at all by driving the grid, so the   
   >harmonic purity due to high Q is maintained, and most old radio   
   >circuits show the Hartley output from the top of the coil of the  tank   
   >which is at 0V potential. To damp the input current pulses, one can   
   >put a cathode resistance, 470r to the coil, and a cap from cathode to   
   >0V so that F higher than range wanted are bypassed.   
   >   
   >The grid and its grid current or a j-fet gate and its diode should   
   >have a minimal C coupling to the tank. Usually 22k to 47k grid leak is   
   >used, but maybe it can be more to keep the C low.   
      
   I don't find the same thing, but maybe I design differently. My way is   
   to equalise the Q-setting loadings of the gate and the source. Each   
   appears as a parallel resistive load across part of the tuned circuit.   
   Each should have the same ratio of R to X, so that the loaded Q is   
   determined equally by input and output. The actual R to X ratio   
   determines the coupling, which sets the pre-oscillation loop gain.   
   This sets both the start-up condition and the degree of limiting in   
   the active stage during operation. Oscillator linearity is determined   
   this way, and if you get it right, there is almost no extra harmonic   
   content in the source/cathode branch.   
      
   d   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)