Print

[VE7RF]

Here the tuned input is finally installed, and tested.   The tuned input consist of a 'L-PI' network.   The 3CX-6000A7
has a 50 ohm input Z, so the L-PI  tuned input was designed in software by myself, using the GM3SEK PI / PI-L
spreadsheet.

The 3x6 triode has 42 pf of internal capacitance between the chassis grounded control grid...and the  directly heated
 ('instant on') cathode.   That 42 pf has to be subtracted from the  calculated C2 value of the tuned input.  Problem is, the previously installed  fil choke assy  (5 x oval beads slid over a pair of double insulated 4 ga, fine stranded power cables),
measured 50 pf of stray C, between EACH fil lug..and chassis.  With tube then plugged in, total stray C rises to a whopping 142 pf...which of course screws things up badly.   

The issue is, the fil choke assy was right up against the chassis in it's previous configuration.  My fix was to remove 2 of the 5 x oval beads, and re-configure the final assy, so it hangs in mid air, between the pair of 100 amp rated Super-con connectors...and the socket. That reduced the stray C from EACH fil lug to chassis...down to 22 pf.  Ok, so 44 pf of stray C instead of 100 pf.   I was expecting less.

With the original fil assy installed, the L-PI would  not resonate any higher than 28 mhz.  With it removed, it would resonate  higher, like 40 mhz.  The 1/8" tubing coil between the pair of air variable caps was reduced in diameter, in small increments, and re-tested each time.  Finally, Scott got it to resonate at 50 mhz.  New fil choke assy re-installed, and
L-PI tested yet again.  Then again with bottom lid on.  Success at last.  Flat SWR from 49-56 mhz.    Also installed was the
bypass caps at cold end of new fil choke assy.  Also the 6 x .01uf @ 3 kv rated disc caps across the cathode. 3 of those caps are paralleled..and ditto for the next 3 x caps.  Both those assemblies are wired in series..and the entire mess is wired across the cathode socket connections.   Drive RF from the C2 air variable, is applied to the junction point of the 6 x caps.  This makes a better use of coupling caps, and divides the drive RF current equally 6 x ways, with 1/2 the total current being applied to each end of the cathode.   This is the correct method of driving any GG / cathode driven tube.

The input  RJ1A ceramic vac relay was also installed, and RG-400 teflon coax was used between vac input relay and  the C1 air variable cap.   More RG-400 used from the same vac relay..to the input SO-239 jack.  The cone shaped shroud over the SO-239 input jack eliminates any Z bumps.  A 26.5 vdc supply was temp used to power the  vac relay
for testing purposes.  A temp 50 ohm resistor was wired from one cathode socket connection..to the chassis...to simulate the tube's input Z. 

This entire process took several hours.  When you are dealing with miniscule amounts of uh (like .15 uh) between the 2 x
air variables....and  .015 uh  between the C2 cap.... and the junction of the 6 x discs.... it becomes a slow, tedious process.  The value of both those inductances greatly affects  the network.   If the tuned input is not done correctly, you are dead in the water. The BW is exactly where I wanted it...and jives with software.   Enjoy the 6 min video.

https://www.youtube.com/watch?v=Hr7BPxz0UBc

[SWMAN]

  Oh !

[VE7RF]

On a similar note, the smaller 3CX-3000A7 will drop right into the same socket.  Only the smaller chimney + fil V is changed.  Input C is 38 ohms on the 3x3  (vs 43 pf on the 3x6)... none issue. The tube's input C is simply absorbed by the C2 air variable cap.  Both tubes have the same 50 ohm input Z.  The tuned input for the smaller 3x3 tube is identical to the 3x6.   The high power pi net on top side of chassis uses the same pair of vac tune + load caps, same plate choke, etc.  Plate load Z on the smaller tube would be a bit higher, so  the value of both coils would be the only change there.  Only 1 RJ2B output relay would be required. Smaller blower used. Minor bias V changes  used.   No IPA required...just the 200 w xcvr.
  5 kw out with 200 watts of drive.

[KD6VXI]

The Q would be higher with the 3x3 but I seriously doubt you'd need to change coils.

The 3x3 I built for Dino covers 10, 12, 17 and 15 meters.  One amp, one coil.  Input and output networks.

I was able to load it to 14ke pep on 10, 14.5kw on 15.

It's not like we are talking a Q of 200.  I used a Q of 8 on ten meters to get this bandwidth.

--Shane
WP2ASS / ex KD6VXI

[VE7RF]

The Q would be higher with the 3x3 but I seriously doubt you'd need to change coils.

The 3x3 I built for Dino covers 10, 12, 17 and 15 meters.  One amp, one coil.  Input and output networks.

I was able to load it to 14ke pep on 10, 14.5kw on 15.

It's not like we are talking a Q of 200.  I used a Q of 8 on ten meters to get this bandwidth.

--Shane
WP2ASS / ex KD6VXI

Shane, it's a whole different ballgame on 6m.    This is a 50-54 mhz amp.  There is next to NO wiggle room.  And there is NO way in hell it can work, without the  extra .13 uh coil between plate block cap assy...and C1 vac cap.  And this is with a total network Q of 12.  (Input Q = 8.8     Output Q = 3.2 )  Plate load Z = 1019 ohms.  5500 vdc loaded to 3 amps....(9900 watts out, 60% eff).

'Pre coil' = .13 uh
C1 =        23.2 pf
L =           .13 uh
C2  =      234.0 pf

The plate load Z of  1019 ohms gets transformed  DOWN to just  241 ohms. The output PI net is designed around the 241 ohms.  The 45 pf of anode to  grid C, plus the extra .13 uh coil form a step down L network.  (shunt cap, series coil).
...


Ok, here is what happens when a 3CX-3000A7 tube is plugged in....and the SAME pair of .13 uh tank coils are used.

Same 5500 vdc loaded is used but only 2 amps of plate current. (6600 CCS watts out, 60 % eff)

Plate load Z increases to  1528 ohms.

Total network Q has to be INCREASED to a whopping  21.   (input Q of 16....and an output Q of 5).
Plate load Z gets transformed from  1528 ohms...down to  505 ohms.

'Pre coil'   =  .13 uh
C1      =      38.6 pf
L         =      .13 uh
C2      =    314.7 pf

So NO,  you can't just swap to the smaller tube.... and use the SAME  PAIR  of coils, won't work.   (It will 'work', but the circulating current skyrockets, and BW narrows by a bunch).

[KD6VXI]

Your last sentence says it all.  It will work.

The point is, a tube dies during a contest or whatever, and you have a 3x3 spare.

Pop it in, it works.

Is it ideal?  No.

It will work.

And if the tank is designed around this fact, you won't be burning anything up.

--Shane
WP2ASS / ex KD6VXI

[VE7RF]

Your last sentence says it all.  It will work.

The point is, a tube dies during a contest or whatever, and you have a 3x3 spare.

Pop it in, it works.

Is it ideal?  No.

It will work.

And if the tank is designed around this fact, you won't be burning anything up.

--Shane
WP2ASS / ex KD6VXI

Here's a more eloquent way to do it.  Shoulda thought of this before, since it's blatantly obvious..and I already use this trick on another amp.
 
Just drop the B+  down to  3668 vdc loaded at 2 amps.  Then the plate load Z doesn't change..and remains at the same 1019 ohms.

Then it becomes  7336 watts  DC input....and  4400 watts output.  (2936 watts of anode diss).  No big deal, since the plate xfmr is tapped for all sorts of voltages.  You can also achieve even more effective sec taps, by using various combinations of taps.  IE: my big dahl is tapped at  0-2300-3900-4800-5200.   By using the 5200 + 2300 taps, it now becomes 2900 vac. (5200-2300 = 2900).     By using the 4800 + 2300 taps, it now becomes 2500 vac (4800-2300 =2500).   Ok, so that's  2 x more voltages  that I now have, on top of the existing 4.  Primary tapped at 198-208-218-230-240-250. Between pri + sec taps.... I can get a bunch of different voltages.


In addition to lowering the B+ a bunch, the chimney would have to be swapped out.  The fil V would have to be raised from 7 vac...up to 7.5 vac.... non issue really.   Ideally, a 2nd brand new (or rebuilt) tube is on hand at all times..and has been  pre-tested.  If building a hb amp from scratch, I highly recommending having 2 x known good tubes on hand, before you even start the project.   WAY too much effort goes into the project.

If using russian surplus tubes, like throw away  GS35B's, and GU74B's.... make damned sure you have a bushel load of em...they have not been manufactured since the early 70's.   Ditto with throw away  Eimac  YC-156's MRI tubes.  Buddy bought 12 of those YC-156's....and only 3 are good. The rest of em develop cathode to grid shorts after 5 mins into the 8 min warmup. 

[VE7RF]

The Q would be higher with the 3x3 but I seriously doubt you'd need to change coils.

The 3x3 I built for Dino covers 10, 12, 17 and 15 meters.  One amp, one coil.  Input and output networks.

I was able to load it to 14ke pep on 10, 14.5kw on 15.

It's not like we are talking a Q of 200.  I used a Q of 8 on ten meters to get this bandwidth.

--Shane
WP2ASS / ex KD6VXI

How did u manage to achieve 14.5 kw out on 15M...with a tank circuit designed for 10m...and with just one coil being used ?  What was the loaded B+ and plate current ?  Knowing that, I can calculate the plate load Z.   Ok, just did it in software.  7 kv loaded @ 3.5 amps would work.  But that's pushing the tube more than I would like.

On my own 3x3  160-15 amp,  I had to use the L-PI trick, to get the Q down to 8 on 15m..... which rises to a Q = 13....when 15m main tank coil  re-used on 17m.    That's the one drawback of the L-PI trick.  When re-using the  main tank coil, (pre-coil is fixed..and always in the circuit on all bands), the Q will skyrocket on the next lower band.   Without the L ('pre-coil'), the Q only increases as to the ratio of the 2 x bands.

IE:  21 / 18 =   16.7% increase.    However, if a pre-coil is used, the Q  increases  by  62.5%.

On  SSB, not a huge deal.   Toss in FT-8, and you will soon find out ..fast like, how hot tank coils get.  With double the average plate current, when using FT-8,  the watts being dumped into the tank coil as heat now quadruples.  Then u start to see why ameritron amps crack the ceramic bandswitch, when FT-8 used on 17m band......(re-using the 15m tank coil).

10M  FM is another doozy.