From: info@turneraudio.com.au   
      
   In a previous post about my radical surgery performed on an BWD Electronics   
   141 oscilator, I mentioned I had troubles with the original Compton pot used    
   to alter the F within each range. The Compton is a wire wound item with   
   adequate power ratings, and it reputed to work smoothly without Vo jitter or   
   bounce, and with linearlity between DC and at least 1MHz. But I found it   
   performed with jittery Vo for    
   the parts of the ranges where it did work, and for part of each range the   
   oscillations stopped or went wild. I decided the pot was terminally sick, so I   
   replaced it with 12 pos make before break switch.    
      
   I re-examined the 2 gang pot, and although it looks impossible to separate the   
   two sealed pots, but the outermost has a metal cover which can be easily   
   prised off. I plotted the R values for each track and sonn found a region   
   where outermost pot had an    
   open connection between wiper and resistance wire track. Both pots measured   
   similar R where the wipers of both did connect to the their tracks.    
      
   I was lucky the sick pot was the only one that seemed accessible. I found some   
   grease had been used for lubrication, but no sign of any corrosion or fault in   
   the wiper operation and all mechanical aspects looked brand new. BUT, sure   
   enough, wiper contact    
   wasn't being made at a small length of the track, so I cleaned the track +   
   wiper without damaging the hair thin wire. I thought a bit more use might   
   dislodge the tiny spot of hard oxide that may exist on the resistance wire   
   track. No luck, and I gave up    
   after an hour or two, and felt vindicated that I'd replaced the darn pot with   
   the best thing I had laying around.   
   So when Phil says the Compton Pot is a blessed marvel that enabled superlative    
   oscillators to be made, it could never have been alwats true, because pots can   
   become defective, or maybe even be defective when they are sold.   
   Wire wound pots can be notoriously jerky with Vo levels. But OK when not being   
   moved, when they are then usually noisless.   
      
   I'm currently making a tubed version of WB oscillator.   
      
   Included in my pile of junk acquired over the last 20 years was a metal box   
   about 400mm x 250mm x 300mm which was amoungst stuff froma retiring ham radio   
   guy. Seems he only wired 3 of 16 tube sockets, maybe some amp for a long lost   
   cause. Anyway, a very    
   suiable box for what I now wanted.     
      
   I'm to use FOUR 3 gang pots with each gang having max C = 458pF, and with   
   trommer C the range of C can be 2,770pF to 277pF so the R value for 2MHz   
   operation in a WB oscillator will not be too high, C not too low, giving good   
   HF operation.   
   For 1Hz to 10Hz, I will use a dual 50k log pot with fixed C about 3.2uF   
   So 2 dials are needed, but I should easily get 1Hz to 1Mz. So that when the   
   lowest F range is selected, the switch changes to the variable R and fixed C.    
      
   The tube amp has 6DJ8+EF80 input diff pair, EL86 gain pentode, then 2 more   
   EL86 used as a white follower. Bandwidth of amp = 3Hz to 3.3Mhz, dead flat   
   between 1Hz and 1MHz. Input C to each input port is low enough. Open loop gain   
   has shelving of gain HF &    
   LF outside the audio band of 20Hz to 20kHz. So far, 1kHz THD of amp <0.03% at   
   7Vrms output, a bit high, considering the is so much gain reduced from about   
   1,400 to 3.0 with the FB tried. I think the 6DJ8 CF input is causing excessive   
   and I'm working on a    
   method to apply the NFB to input EF80 cathode, with resistive NFB network   
   allowed to float between 1,100uF output cap after the EL86 buffer and a   
   8,200uF cap to ground. It means no DC flows in any part of the NFB R divider   
   network, because I have a 13mA    
   CCS sinking cathode Idc to 6BX6. But the bottom R will be about 360r with   
   lamps used, and this acts as local current FB to 6BX6, halvingt its gain, but   
   giving some slight local linearity. Thus the grid may remain at 0V and only   
   terminated by the WB    
   network R values.    
      
   I bet nobody in the world followed what I just said, but then my post is aimed   
   at those wanting to making their own Oskar Laytor because they have the   
   patience. Its only when ppl try to make something real that any descriptions   
   around what they do become    
   interesting.    
      
   I might find room for a square wave Schmitt Trigger and an additional output   
   buffer. But maybe not with tubes, because getting a decent 1MHz square wave   
   means highest F present should be 10MHz. Maybe I cheat, and use some BJTs,   
   actually easier to do, and    
   for an additional OP buffer after the attenuators, so that Rout of the unit is   
   50 ohms.   
      
   I look back at the HP200A oscillator with a 4 gang cap and I think, "how   
   quaint", and it was said to be a good tool to test audio amps, with THD < 1%,   
   and an unneccessarily high 20Vrms+ output. But it only went 30Hz to 35kHz, so   
   how could you test any    
   audio amp? What is wanted is something to look at amp behaviour below 30Hz and   
   above 35kHz, so for me the HP200A would be quite useless!!!!   
      
   Patrick Turner.   
       
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)