From: info@turneraudio.com.au   
      
   In any Wien bridge oscilator equipeed with a variable tuning capacitor, there    
   are limits on performance at higher H above about 200kHz unless the   
   capacitance input at oscilator amp input ports is very low, and the stray   
   capacitance is very low.   
   Consider a typical oscillator with a typical broadcast band dual gang tuning   
   cap with each gang having 400pF maximum and 20pF minimum.   
   Suppose you wanted 100kHz when C = 400pF, then Wien bridge R value = 3,975r.   
   In theory, if C = 20pF, then F = 2MHz, but you'll never get that in practice   
   because of the stray C and device input C and maybe you'd only get 250kHz   
   before the oscillations stopped, because the NFB input became larger than the   
   PFB input.    
   But one might add some trimmer C across both gangs so minimum C might become   
   say 40pF, and in theory you'd get 1 MHz with R = 3,975. But then the maximum C   
   becomes 420pF and lower Fo drops from 100kHz to 95kHz, and if you reduced R a   
   bit then you'd get    
   low Fo to 100kHz, but then that moves the theoretical high Fo to about   
   1.06MHz.    
   One can spend days guessing R and C values and twiddling trimmer cap screws   
   and never get a nice decade change of F.   
      
   So, how to set up what is needed?    
      
   If you are using solid state devices, try to used j-fets at the differential   
   inputs which have low C, say less than 10pF at each. With tubes, a 6DJ8 as a   
   diff input isn't bad and C is fairly low, because the common cathode signal   
   voltage is large    
   compared to the anode Vo of opposite phase, so Miller C is small.   
      
   If you are aiming to get highest Fo = 1MHz, then you can try using a Wien   
   bridge network with fixed R&C of say 100pF and 1,590r. If you find it   
   oscillates OK with a planned 6Vrms output, then you may find Fo is only   
   800kHz.    
   This may be due to 20pF stray and other C in parallel with 100pF, but usually   
   the stray C affects the Wien bridge top and bottom C values assymetrically and   
   it just won't oscillate properly. Ideally, if you get is working fine with RC   
   giving 1kHz network    
   with NFB FB adjusted just right, then it should work fine when you switch up   
   the F ranges to the very top Fo and the NFB setting should not need any   
   re-adjustment.    
      
   Some oscillators, even the HP200A, have 4 gang caps with each gang having   
   500pF max C and 20pF minimum. The actual C change in the gang for each network   
   C = 1,000pF down to 40pF.    
   But we really want C mimimum to be 1/10 of C maximum.   
   Consider the formula, Ct = ( C1 - [ 10 x C2 ] ) / 9,    
   Where Ct = trimmer C value + stray C + input C,   
   C1 = maximum tuning gang C,   
   C2 = minimum tuning gang C.   
      
   Consider the 4gang tuning C with C1 = 1,000pF, C2 = 40pF.   
   What is added Ct for decade F change?   
      
   Ct = ( 1,000 - [ 10 x 40 ] ) / 9 = 600 / 9 = 66.6pF = say 66pF.   
   In theory, you'd get 1,066 Pf max C down to 106pF minimum.   
      
   But the trimmer C you'd need to actually add might be a fixed 39pF plus a   
   4-60pF trimmer C, and hopefully, when you try for 1Mhz, you will get it, with   
   R = 1,500r.   
      
   But the circuit may still not want to allow you get such a high F without a Vo   
   level roll off of maybe -6dB unless you have sufficient open loop gain even   
   with the R&C compensation network in place to stop the circuit oscillating   
   parasitically at above    
   1Mhz, at maybe 5MHz, which can be typical.    
      
   in my recent experiments while trying to get a tubed RC coupled amp to work   
   with parallel tuning gangs, I tried a fixed RC network with 220pF and with R =   
   390r.   
   F max still varied while varying the R&C HF compenstation network of the amp,   
   and as F rises the whole caboodle becomes more interactive and problematical   
   and the only solution seems to be to increase the minimum C so you get   
   reliable oscillations at    
   1MHz without parasitics also present, or any squegging, with oscilations   
   stopping and starting. I found 220pF would be an absolute minimum C value.   
   I am using two 3gang caps each 1,200pF to 39pF and this gives 2,400pF to 78pF   
   for the change of tuning C. The range of actual C will be 2,580pF down to   
   258pF with added C = 180pF, including whatever stray C is there. R value   
   should be 616 ohms for 100kHz    
   to 1MHz.    
   It means also that for 1Hz to 10Hz I'd need 61.6Meg ohms, which is far too   
   high because of biasing problems with tubes, ie, the minute grid input   
   currents.    
   So I shall settle for 10Hz to 100Hz being the lowest F range with R = 6.16M   
   with the tuning cap F control. I can do 1Hz to 10Hz range with 50k log tuning   
   pot with added 5k6k in series and 2.86uF fixed C, approx, subject to final R   
   adjusrtment.   
   The resulting 2 dials should give readings within +/- 3%, and stable operation   
   if I fit reduction drives.   
   Patrick Turner.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)