[Amps] PASSIVE NEWORK CATHODE INPUT

[Jim Thomson]

Date: Sat, 15 Sep 2018 20:02:34 +0000
From: "Dr. Andres Goens" <YS1AG at hotmail.com>
To: "amps at contesting.com" <amps at contesting.com>
Subject: [Amps] PASSIVE NEWORK CATHODE INPUT


<Greetings guys. I am building an amplifier around a GS35B. Recently I came across the schematics of Heathkit SB230 using a passive network to the cathode , doing away with the ubiquitous PI and the need of a cumbersome switch.  The solution is very <tempting but I wonder if is there a catch? Any advice will be welcome
<Andy YS1AG

##  Passive networks  sorta work on  GG triodes that require very little drive power, or in the case of a SB-230, a single tube..with high gain.   The purpose of the PI  tuned input is to not only match  Zs,  but also provide for some flywheel action.   The  GG triode is only driven  210 degs out of  360.  Basicly its only driven on half cycles,  with the remaining   150 degs being a sky high Z..and no tube conduction.   The tube cant source anything, it can only sink current.  
Henry Radio  tried  a passive network on its  3K and  8 k ultra..and both are a dogs breakfast. Pssive inputs wont work on a GS35B..which requires a lot of drive to begin with. 

## On  a HB amp,  you  typ have more room to play with.   Instead of 9 x tuned inputs, and either band switched  or relay switched,  I opted for another approach, and that was to use a pair  of broadcast variables....and a tapped coil.   I used 4 x  section broadcast variables, each section is  17-540 pf....then strapped all 4 x sections in parallel.   So each variable ends up being  68- 2160 pf.    Both variables are padded on 160m  via the dual wafer  small bandswitch used to change taps.  The  160m position is not used on wafer #1, since the precise, exact amount of coil is installed.  Instead, the unused switch position is used to switch in  1500-2000  pf  worth of small doorknob caps  onto air variable #1.    Wafer  #2  is not used for anything.... except to switch in  1500-2000 pf  worth of padders onto  air variable  #2.   Small hb  4 uh coil used,  wound with either  6 gauge or  8 gauge  solid  copper wire.  17 turns onto a 1.5 inch OD  former.   Space between adjacent turns... then former  removed.   Most of use  GM3SEKs  PI   /  PI-L  software to determine the uh taps for each band.   4 uh was used for  160m, then  1/2 of that for higher bands. 

##  We also used a pair of  jackson brothers 6:1  ball drives....complete  with 0-100 calibrations marks across  180 degs of the skirt.   Then the bandswitch knob  in between the ball drives.  This all resides below the chassis.  Then you end up with
dead flat  1:1  swr across each ham band.   You can get the 6:1 jackson ball drives  with just the small  6:1  drives by themselves.... or complete with either  2 inch OD skirts,  3 inch skirts or  4 inch skirts...and all marked  0-100,  in increments of 1,  IE:  1-2-3-4-5....  100, across the  top 180 degs of the skirt.   Then its all silky smooth tuning, just make up a cheat sheet... and dial it up by the numbers.   The above was for a 50 ohm input Z...3CX-3000A7.   To test, we used bird wattmeters on either side of the tuned input..then into a 50 ohm dummy load.   Dead flat swr on all bands.  On 15M only,  with 200 watts  applied to the input, we only saw  160 watts on the output side. The 6 gauge solid cu wire ran warm.   15M tap was increased  by just .25 of a turn...and both caps had to be reduced a tiny bit in value  to obtain  1:1  swr.  Ok,  problem solved.  Power  shot up to  195 watts  on the output side.  On all the other bands, power was 200 watts on the output side.    In the final test,  800 watts of cxr was applied for  10 mins..on each band.  We also tested with 2 kw.   It only takes a few seconds to throw the bandswitch to correct band, then tweak both caps to their pre-described numbers.  High powered  PI  output  consisted of  2 x vac caps  and a tapped coil, with a 2nd,  smaller diameter coil used for  20+ meters.   To switch the entire mess from 15m  to  160m  takes exactly 45 secs....which is good enough for my application.  The above works superb, but eats up chassis  space, below the chassis, hence the  hb  5 inch high chassis..and RF deck fits into a 19 inch wide rack. 

##  Most  setups using 9 x separate PI tuned inputs will use  a myriad  of relays to   switch the 9 x networks..typ all on one compact board.   Fixed caps can be used, but I have never had a lot of success  with fixed caps..and an adjustable slug tuned coil.   I ended up having to solder in several caps  till I got it correct.  ARCO compression  trimmers  work a lot easier.... 2 of em used on each band, plus a  small torroid.   On the lower bands, like  160 + perhaps 80m, the arco compression trimmers are padded.   A  T network, consisting of  2 x coils in series, with a shunt capacitor at junction point of caps..to chassis  will also work, BUT you end up having to carefully tweak the value of each coil  to get a perfect match... pita...and typ only used on monoband amps, like 10 + 6M.    Even then, you end up with the loss of 2  x coils.    

##  Typ bandswitched   tuned inputs, with  5-9  tuned inputs, and relay switched, or bandswitched,  imo,  use too low  a Q, and typ tuned for the center of each band,   so they  cover the entire band. However,  one board is fairly flat,and can be mounted below a shallow chassis.   On the setup I use  with the pair of air variables and tapped coil,  I can use a slightly higher  Q, which results in better IMD, and higher overall amp eff,   but it wont  cover an entire band..on the lower bands.   No big deal, since the high powered   PI net has to be retuned between ssb  and  cw anyway.   So the  cheat sheet is modified, with several tuned input settings  across each band...and ditto with the PI settings  for the high power PI net.   The turns counters I use for each vac cap, not only count each turn, but also the entire skirt  circumference  is marked  0-100.   Each revolution of  both vac caps is marked  0-100....which makes for precise tuning.   IF instead, air variables are used for both the high powered  tune and load caps, then  6:1  jackson bros ball drives are used..with their 0-100 calibration marks across the upper 180 degs..... then u still get to..dial it up by the numbers. 

##  If you decide to use  GS35B  tubes, make sure you have several of them, like half a dozen at least.   Too much effort  goes into any hb amp.... only to find that replacement tubes  down the road  become unobtainium, or prices have skyrocketed.  GS35B  is a good tube,  but some will arc between anode and grid  at varying  B+ levels.  Some are good to  4500 V..and some are only good to 3700 V.   They really need to be hi-pot  tested before use.    You also require a variable bias supply  to set the idle current.  Simplest is just a series string of either  1N5408  or  6A10 diodes.  Idle current can be reduced a bunch on CW, like 10-25 ma....but more drive required to overcome the higher bias voltage used. You cant  cut off the idle current on a GG triode when on CW mode, you will end up with key clicks.   You can cut it off on  FM.   I wire a big electrolytic between 1st and last diodes in the series string, like  10-40K UF.  V drop across each diode increases with more plate current drawn.  With the big cap in there, the bias V  does not budge.   A  simple  12 position rotary switch can be used to change taps on the bias diodes.  Typ the 1st bunch of didoes is always left in the circuit, and then the switch taps every  2nd or 3rd diode.   A small  spst  toggle can be used across the last diode, for  a ultra  fine bias adjust.   For an even simpler bias switching  setup, a spdt center OFF  toggle can be used, which will result in 3 x bias settings... IE:  all the diodes used..when in OFF  position, or 2 other bias settings.   5408 is rated for  3A..with a 200 A surge...and the  6A10 is  rated at  6A  with a 400A surge.   You can also parallel  duiodes for more current handling.   On my test bench, the 1N5408  runs  fairly warm with  1A  CCS.    The  6A10  runs fairly warm  with 2 A   CCS. 

Later.... Jim  VE7RF.