Neutralization of Class C Triode - Tank Circuit Design

[N8UH]

Some new parts acquisitions have helped me finalize the outline of my transmitter project. What I'll be sitting on is essentially a 5K BC transmitter without the cabinets. 4x1s modulating a 3CX2500. That should give me some headroom. 

Anyway, I've been studying the old handbooks to figure out a way of designing a frequency-agile PA stage that will not require constant neutralization. The handbooks point out that a big old-buzzard split-stator plate capacitor and balanced link-coupled tank circuit should only require an initial neutralization on the highest frequency, and neutralization will hold for the lower frequencies. Cool.

But, and correct me if I'm wrong, there's a few problems:

- Split-stator capacitors that will handle the high voltages I'll be using (3-4KV) are very difficult to find
- Coil swapping would be required for band changes
- The harmonic attenuation of the link-coupled output will be less than a pi-network

Again, I could be wrong on any or all of the above points though...

So, if my thinking is right, I just live with the above design limitations, or go with a standard pi-network and re-neutralize on big frequency changes, which kind of sounds dicey.

Is my thinking right here? What are my options?

Thanks in advance!

[W3SLK]

Tim,
Seek the Timtron's advice on this. He and I were just discussing this very thing. It had something to do with regards to how you neutralize the input when using a PI network. I intend to make a rig with an single 833 modulated by 810's but it will be for 160, 75, and 40. OFS prevents me from the whole gist of the conversation but I intend to really hit the books once the time comes.

[WD5JKO]

Tim,

  As usual I will suggest something out of the ordinary. 

Before discounting this, consider the simplicity of the approach versus the big plate split stator neutralizing issue with plug in coils.

That big triode is low Mu (22), so the drive is pretty high:

http://www.g8wrb.org/data/Eimac/3CX2500A3_and_F3_and_H3.pdf


   I'd do the following:

Use a Pi-Net output circuit
Run the tube in grounded grid (likely ~ 6 db gain)
Use a SB220 as the RF driver (might need a KW peak drive in G-G)
Plate modulate the final with the 4-1's
Cathode modulate the driver with the WA0ZHH circuit
http://kx5jt.net/myPictures/wa0zhh.jpg

Remember with G-G a lot of the drive stage power comes through to the output, so the high drive is not wasted. For grid driven, class C plate modulated, the spec sheet says the RF drive needs to be 125 watts. In G-G if the driver needs to put out 1KW (a guess), the RF grid drive is still 125 watts. That means 1000-125=875w of the drive feeds through to the RF plate circuit.

My 2 cents,
Jim
WD5JKO

[KM1H]

With a series feed HV the Tune cap needs to have a minimum 2X the DC spacing for AM. With shunt feed such as in a pi-net it can be just a "hair" over the DC; I use 1.4 as the increase and it seems to work. Just dont whistle into the mike when tuning, that can even cause arcs on SSB as off resonance voltage can soar.

Carl

[ke7trp]

Carl is right.  You need at least 2 times HV rating for Am on the cap.  You will need a min of 10KV Cap in there for 4K to 5K on plate..  I have a 9KV one here but its only 150mf so its probably not usefull to you.

I would not run that tube at Grounded Grid that is going to really take the Drive requirements up.  Running an SB220 to excite the transmitter is just nuts.  Talk about heat and having a weak link.

In grid Driven you need -400 volts for Class C at 4KV with that tube. But you are not going to run this thing with 5000 watts carrier out and if you did, You would be not have enough modulator anyways.  So being realistic here, You are probably going to run this thing down at a Grand around 2500 to 3000 volts. which will mean that you wont need anywhere neer the -400 volts.  I suspect the actualy Driving power from your exciter will be 100 watts or LESS CW in Grid Driven operation.

As for the neutralization question, I cant help there.. Sorry.

C

[KC2ZFA]

why would a pi output network require re-neutralization when you change bands ?
Isn't the grid-plate capacitance of a tube pretty much frequency-independent ?

Can someone clear this up ? I'm a newbie who is easily confused !

Peter

[WD5JKO]

I would not run that tube at Grounded Grid that is going to really take the Drive requirements up.  Running an SB220 to excite the transmitter is just nuts.  Talk about heat and having a weak link.

Well maybe so, but consider this. The SB220 with the WA0ZHH cathode modulator would change everything. Why? Well the 3-500's would be run class C, and the plate efficiency of a grid + cathode modulated stage could come out to around 50%. So running 600W DC input, might give you 300W carrier. The 3-500Z's will be cold, and the power supply only warm. Then turn up the audio to the 6336's in the modulator until you have the peak drive you need. Then consider that about 80% of that drive goes straight to the antenna as feed through power. The G-G final also solves the neutralization problem. The week link would be the class A modulator with those 6336 triodes....Maybe ripe for a PDM solution.

I may be nuts, but I did think about this a bit.

Jim
WD5JKO

[N8UH]

I think your idea would work Jim, but since I have the original oscillator/Driver/IPA section and the full audio section from the original transmitter, I'd like to modify and keep those in service. And, I'd like to avoid purchasing more gear... 

Clark, actually, I have plenty of modulator, it's the original section from the 5K transmitter. Probably good for about 3KW of audio But you are right. I won't need a whole lot of drive. In fact, if I modify the osc/driver/IPA sections I can probably get away with using a nice synthesized VFO. 

[K1JJ]

http://www.amwindow.org/tech/htm/813/813.htm

How about using this simple neutralization circuit in this 813 rig?   It involves just floating the grid tuning RF circuit cap and coil for DC.  You can mount a 4"X 4" piece of aluminum (or whatever works out)  next to the tube plate for sampling.

Just run a conventional pi-network in the final, or whatever you have now.

I sometimes bring out a neutralzation trimmer cap shaft to the front panel (see schematic) to touch up neutralization on the higher bands, but find I never use it. It stays pretty close on all bands once set on the highest band.

T


Here's the neutralization circuit used with a 4-1000A modulated by a pair. The neutralization knob is near the center of the front panel:

[N8UH]

That's where I start getting a bit cornfused Tom... Does your Tetrode neutralization scheme work the same with a Triode? It was my understanding that Tetrodes don't have the same neutralization problems that Triodes do.

[K1JJ]

Hi Tim,

Neutralization is neutralization... :-)

A triode will need more feedback because it has a larger grid to plate capacitance to cancel out. That's why I suggested a larger plate sampling tab than I usually use with a tetrode.  Sometimes guys don't bother to neutralize tetrodes cuz the G-P capacitance is so small and the final remains stable. However, I always do mine anyway.

Because the tetrode's screen acts as a shield between the grid and plate, there is less G-P capacitance to cancel out, thus less sampling signal needed. So the difference between a triode's and tetrode's requirements is simply the level of neutralization signal needed to do the job.

I've used this same circuit on triodes too.

When you test it out, just turn on the fils and put in a drive signal. (No HV applied)  When you get the sampling plate size and positioning correct, you will see the trimmer cap adjust the amount of feedthru signal that appears in the plate circuit. I usually tap the X10 scope probe onto the C2 loading cap high side for best isolation from the plate circuit. Just dip that feedthru signal for minimum and you are golden to fire the final up.

T

[W3SLK]

Peter said:

why would a pi output network require re-neutralization when you change bands ?
Isn't the grid-plate capacitance of a tube pretty much frequency-independent ?

Can someone clear this up ? I'm a newbie who is easily confused !

Let me see if I can make this clear as mud Peter. You are correct, the Cg-p is constant and would be frequency independent BUT, high level plate modulation, (Class C final) is extremely rich in harmonics. Depending how you build your grid and plate circuit can have a serious effect on whether your triode works as a Class C amplifier or oscillator. As opposed to tetrodes, there is an extra element stuffed between the G1 and the plate. This acts as a capacitor in series which raises the neutralizing frequency. Now depending how you terminate the input could make it appear as capacitance in parallel, which is not good.
Also, going the route of plug-in coils, you face yourself with the same dillema as that of the BC-610: Bleeding all the HV off before you change the coils.
If others care to correct where I am mis-leading, have at it. I'll learn something from my mistakes. IIRC Tim said something to the effect of grounding(or not grounding) the un-used turns on the input coil with respects to band changing.
I just hope I wasn't too simplistic.

[W2PFY]

3CX2500

Is this an A3 or F3 tube. F means flying leads. Do you have two?

if it's an A type and you have room, you can put another tube in there to act as the neutralizing cap. The tube would run with no filaments hooked up or attached. You would merely use the tube as a neutralizing cap. I heard of it being done but I never tried it myself.

[N8UH]

They are F3s, and I do have a pair of them just in case

Interesting possible solution. Perhaps I could wire both up and switch between them to keep them in the rotation... Hmmmm.

[W2PFY]

Perhaps I could wire both up and switch between them to keep them in the rotation... Hmmmm.

I don't think the F types would not be a good candidate. The long leads would get you in trouble.  The filament needs to be floating as you would only use the grid & plate in the tube.The tube could have a blown filament as long as it wasn't shorted to the grid and still work In my Westinghouse, They use a 1-35 pf vacuum variable cap to do the job and it's adjustable from the front panel. They are found cheap on ebay because no one knows what to do with them. That would be the way I would go with a vernier drive on it. Once you set it for the highest freq that your going to use, you can pretty much forget about it.

I knew a guy that ran one for years. He used a GG 813 to drive it and he said it was real easy to neutralize. He also used 4 each 3-500 in the mod. I don't know what  you are going to use to drive it but if your going to use a solid state rig to drive the grid, your a braver man than I. I would want to put  a buffer like the GG 813 in there ahead of the 3CX2500. It that thing goes out of naturalization, you will get some extremely high RF voltages on the input coil and it will find it's way into a transceiver and kill it. How do I know, I did it. Bad news.

[k4kyv]

But, and correct me if I'm wrong, there's a few problems:

- Split-stator capacitors that will handle the high voltages I'll be using (3-4KV) are very difficult to find
- Coil swapping would be required for band changes
- The harmonic attenuation of the link-coupled output will be less than a pi-network

Split stator caps may be harder to find than single-section, but I regularly see them at hamfests. 

You could bandswitch the balanced tank circuit, but plug-in coils are much easier to do.

If the circuit is designed properly, with appropriate Q to the tuned circuit, there should be little or no difference in harmonic attenuation. I have used link coupled balanced tanks for decades, combined with an outboard antenna tuner, and have never had harmonic problems.

If the circuit is designed so that the modulated DC doesn't appear across the tuning cap, the plate spacing requirement for the tank capacitor is the same as that of a single-ended tank or PI network.  You can either use an rf plate choke and DC blocking cap to get he DC completely off the coil and capacitor, or run the entire plate capacitor and coil at full modulated DC plate potential so that only rf appears across the capacitor. The latter requires more care in the design, but the rf choke requirement is much less severe.  This is the way it is done in the BC-610.  This requires extreme care when changing plug-in coils, since they are at full +HV DC while the plate supply is on.

[N8UH]

Thanks for the great info everyone. Some of this stuff still stymies me from time to time!

Don, good info to know. I find link coupling intriguing. Maybe not with this TX, but eventually I'd like to build a classic TX using link coupled stages and outputs, like in the old handbooks.

Tom, so it sounds like if I use this scheme with a bigger sampling plate, things should be pretty trouble free. That's a relief! Thanks!!

So neutralization of this big Triode doesn't seem to be as bad as I had initially thought. If I use Mr. Vu's neutralization scheme, things should be fine!

Now, on to the calculations and design of this beast! Like I said above, I'm going to try to keep the original stages intact. The Osc/Driver/IPA section consists of a 12BY7 > 6146 > 4-400. With some minor modifications, I think I'll be able to make this section very conservative and frequency-agile, perhaps driving it with one of those synthesized VFO kits.

The modulator section is 12BY7s > EL34s > 4-1000s. It's intact, so just adjusting the biases for lower power operation should be all I really need to do.

So, the biggest part will be designing and implementing the PA section. Shouldn't be too bad.

I'm thinking I'd also like to maintain the cutback functionality of the original transmitter, since it's already there. Of course, instead of 5KW/500W, I'll need to go with something a bit lower.  Or, I could just use a big Variac for the plate supply and call it a day.

Thanks for the help guys, I'm sure I'll have many more questions during the design and execution stages. What a great learning experience!

[KM1H]

They use a 1-35 pf vacuum variable cap to do the job and it's adjustable from the front panel. They are found cheap on ebay because no one knows what to do with them.

I'll take all you want to send this way

[KC2ZFA]

Mike said:

Depending how you build your grid and plate circuit can have a serious effect on whether your triode works as a Class C amplifier or oscillator. As opposed to tetrodes, there is an extra element stuffed between the G1 and the plate. This acts as a capacitor in series which raises the neutralizing frequency. Now depending how you terminate the input could make it appear as capacitance in parallel, which is not good.

OK, so you want to neutralize triodes (and tetrodes just to be sure). But
what got me worried I have misunderstood something is UH's statement

"go with a standard pi-network and re-neutralize on big frequency changes"

Tell me this ain't so since in what I'm dabbling with here the neutralizing
caps will not be accessible after the thing is neutralized at the highest
band and then buttoned up !

Confused,
Peter

[K1JJ]

Peter,

Just like the older 6146 finals in 32S3 Collins or Yaesu rigs, once the neutralizing cap is set on a higher band, like 10M or 20M, it should be good across the rest of the lower bands too. It's a matter of degree, and I've found from experience an average layout can stay stable across the bands without constant retweaking.

Perhaps a problem occurs when extra stray in/out capacitance is added in when switching the bandswitch or adding in whatever L/C components in the normal bandswitching process. Big rigs have big physical components that may affect consistency in layout. But the G-P requirements of the tube itself should remain pretty constant across the bands.

I have noticed a slight tweak needed on occassion for my own 4-1000A plate modulated rig to get it perfect, but it wasn't really enuff to cause any instability. I have an insulated shaft coming out to the front panel from the neut cap, but never really use it. Notice that most of the commercial ham AM rigs never had them.

I wouldn't worry about it even without an external tweaking shaft - triode, tetrode or pentode.


T

[W3SLK]

Tom, et al, I don't have near the building experience of 'big rigs' as you do. But I think this is what Tim was eluding to when we were discussing a single triode. I wish he would jump in here and then all of us would understand the point he was trying to make to me. He did reccomend a PI network in the final but using a balanced input with an extra Cn(?) to be switched in when changing bands. This isn't an issue with BC rigs since its like Ron Popeil: "Set it and forget it!"

[K1JJ]

Mike,

Yes, I'm sure you're correct. It's just a matter of degree to how perfect the adjustment needs to be across the bands. I think a big rig triode can take a little neut misalignment as it goes thru the bands and still be OK for stability.

That said, sure, adding in some external neutralizing padding caps where needed via the bandswitch would be an excellent way to perfection...

I also like using a pi-network. I guess my 813 neut circuit would be called single-ended, but never had a problem with it not being balanced.

Maybe Tron will clarify what he was saying - I've never known the Tron to be wrong when it comes to RF, so would not bet against him...

T

[w1vtp]

<snip>
I sometimes bring out a neutralzation trimmer cap shaft to the front panel (see schematic) to touch up neutralization on the higher bands, but find I never use it. It stays pretty close on all bands once set on the highest band.

T


<snip>

With proper layout it shouldn't be necessary to "tune" the neut..  

Tom - maybe as you indicated on the high freq's it might be necessary but I wonder about trying to put a BC xmtr up much higher than 40m in the first place.  Your circuit should do the trick or just use a balanced grid ckt and neut.. the classic way.  There's plenty of info on that approach.

Al