From: info@turneraudio.com.au   
      
   Phil mentioned, as I expected, about yet another piece of old junk made in   
   Autrsalia when far better stuff could have been made........   
      
   Although capable of battery operation, a mains PSU came installed in the    
   battery compartment. The vast majority of units made had this option fitted.    
      
   The custom made, wire wound, reverse log, 15kohm dual potentiometer was the    
   best thing about it and normally outlasts everything else. I still have two    
   perfect examples, one in daily use.   
      
   **You are lucky. The pot on the one I have IS SHIT, intermittent, and you   
   can't pull it apart to clean it and I guess it was worn out. There were other   
   faults in PS, fucked bjts from having recieved stray excessive voltages from   
   something it was hooked    
   up to.     
      
   Because of using such a beautiful pot, BWD 141s have exceptional frequency    
   stability.    
      
   **I have found all pot operated wien bridge oscilators are far less stable   
   than somethig operated by a radio tuning gang and fixed resistances. The HP   
   33A gear was SO MUCH better, with a beautiful 4 gang C you can get 100pF to   
   1,000pF. But the amount of    
   C isn't enough to allow sensible R values at 1Hz. There is only room for a   
   suitable replacement for the Comptom pot, or you have switched F where the R   
   says absolutely stable, and when you switch to a selected F, that's what you   
   get.   
      
   The typical THD is 0.05% between 100 Hz and 100kHz.   
      
   **Yes, not bad, but excessively good for just response testing I have other   
   low THD oscilators for low THD.     
      
   Replacing the original BWD discreet electronics with a pair of  NE5532    
   op-amps and a "phase shifting oscillator" topology reduced the THD to    
   vanishing levels (circa 0.002% ) at all audio frequencies.    
      
   **No doubt you could use opamps but They don't go very well at 1MHz. I have   
   not explored using special much wider BW devices. I've just used what everyone   
   can buy at wescomponents.com.au   
      
   **My circuit gives no DC offsets, and excellent square waves from 1Hz to   
   100kHz, quite good enough from 100kHz to 400kHz. Square waves above 400kHz are   
   acceptable for limited use and could be better but I really should put in   
   another emitter follower    
   comp pair buffer after the attenuator switch and pot to create Rout of always   
   50 ohms to preserved the HF content to maybe 20MHz. I have not explored making   
   a better Schmitt trigger sine to square wave converter. But I do have quite   
   good all round    
   performance and better than the crummy original junk that BEGGED to be   
   BUTCHERED, as it was never ever going to get used as it was, and I came close   
   to putting the junk in the bin.   
      
   Shame then Turneroid had to butcher his one so badly.    
      
   **Shame Phil likes to stand up for such terrible old junk.    
      
   **The fet I use for NFB control is not a perfect solution. Its action creates   
   considerable THD as anyone will find if they put a CRO on the drain terminal   
   of the MPF102, the typical j-fet used in such a situation. What puzzles me is   
   that you have a fet    
   with source at 0V, and gate has negative bias applied, like a vacuum tube, but   
   drain is also at 0Vdc bias and the applied Vac to drain swings above and below   
   0V, and there is considerable THD created by the action so the Rd looking into   
   the biased fet d    
   terminal is a non linear resistance. I think there should be about +10Vdc at   
   the MPF102 drain, and that this should be supplied by a CCS, using a   
   u-follower cuit with additional MPF02 with fixed 10Vdc appied to its gate,   
   470k bias R and its gate cap    
   coupled to drain of bottom j-jet. Then the drain to 0V circuit of bottom fet   
   can be cap coupled to other R of NFB network, 2k7, and maybe the Rd with DC   
   bias across the fet may be more linear.    
   I am also thinking of a light dependant resistance and LED, but that all seems   
   to have circuit impedance that is is too high for MHz operation.   
   There is actually rather a small amount online abut the real issues involved   
   with fets used in NFB regulation. BUT, I find RS component have type 327 lamps   
   which are rated for just over 1W at 28V and 40mA, giving 700r, so 150r at low   
   levels of an    
   oscillator, and the other R then can be 300r, giving FB R total of 450r, and   
   OK at HF, and not too low to load the amp too much. 327 would be   better than   
   12V "grain of wheat lamps" which are 27r cold, and ideally 60r with a volt.   
   Using a few in series    
   gives starnge results because then you find any batch you buy are all   
   unmatched, and some run with much varying T. Having said that, I have a 1kHz   
   oscillator that gives 8Vrms at THD 0.004% using NE5534, and with just the one   
   12V lamp which has about 2.   
   67V across it.    
   Patrick Turner.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)