Driver Tank Circuit, Series or Shunt?

[KD1SH]

  Resonant parallel LC tank in the plate circuit of a driver or other intermediate stage; what's the best way to configure it?  I mean, one could place it in series with the B+ and let the plate current pass through it (the DC of course would only pass through the inductor), or one could place it in shunt - from plate circuit to ground, with appropriate DC blocking caps.
  My thinkage is that putting the tank in series with the plate has the disadvantage of requiring the inductor to handle the plate current.  But, with the AC impedance of the tank being maximum at resonance, the gain of the stage should be higher, since the plate current changes appear across a higher load impedance.
   In my 6AG7 oscillator I put the tank in series with the plate, and it gave it a beautifully clean sine wave output, and a lot more peak-to-peak to boot.  In my driver, though, I'm not so sure.
   I've seen it done both ways. Either configuration will "smooth" the output and fill in the missing negative half in a class C or B amp, but are there significant advantages or disadvantages to either? 

[PA0NVD]

The advantage of a blocking cap and  grounded resonance circuit is that the tuning cap can be grounded, no HV at the tuning cap and directly connected across the coil. . If you use a fixed tuning cap, that advantage disappears, but that you have to slug tune the coil. Disadvantage is that you need  an anode choke with possible losses and false resonances,
In a series circuit, the tuning cap has HV on it. Grounding the cold side of the tuning cap will introduce losses because the resonance current has now to pass the decoupling cap. This AC current is Q times higher than the AC component of the anode current wich can be considerable and puts a very high demand at he quality of the decoupling cap.
So technically, the best may be a series circuit where the tuning cap is connecting across the coil and not from anode to ground.

[DMOD]

...But, with the AC impedance of the tank being maximum at resonance, the gain of the stage should be higher, since the plate current changes appear across a higher load impedance...
 

i prefer the parallel resonant circuit for driver tube stages.

Here is a driver I suggested for another poster on AMPHONE. The .01 blocks any DC on the tuning capacitor and allows one side of the plate tuning capacitor to be at ground. CPT and LP1 form the parallel resonant circuit.

The problem with the series resonant circuit is proximity effects from body capacitance when attempting to tune either capacitor or inductor, so the capacitor rotor or the slug stem has to be isolated very well to prevent that proximity capacitance from affecting the tuning of the series circuit.


Phil - AC0OB

[PA0NVD]

I think you misunderstood Phil. Both are parallel resonance circuits, but he means  series when the parallel circuit is in series with the anode, so when the anode current passes through the coil like in your circuit. You circuit is definitely not preferred in my opinion. The resonance current Q x the AC anode current passes trough 2 blocking capacitors introducing losses  You should prevent that.  The tuning capacitor should be connected directly over the coil, to keep the high resonating current inside the parallel circuit.
He calls it parallel when the resonance circuit is grounded at one side, both coil and capacitor, a good situation. But that requires a blocking capacitor and a anode choke. The blocking cap is no problem, it carries loe AC current. The choke may introduce some losses and have false resonances, normally not to big a problem, but it is an extra component in the circuit.
Hand effect during tuning is the same for both circuits

[DMOD]

Guilty as charged for being a Literalist.  

In the PDF file below, Page 1 is a parallel resonant circuit in which the parallel resonant circuit is in series with the B+. As long as the bottom end of the circuit is RF shunted (referenced) to ground, it should work just fine. This could be what the OP was talking about. He can clarify if he chooses.

Page 2 is an equivalent parallel resonant circuit for my suggested driver. It completely decouples the DC from the tuning cap plates with the tuning capacitor's rotor at ground potential. The total RF current is circulating via ground through ~ 9 ohms. It functions very well with no capacitive proximity effect.

Page 3 is simply the equivalent of Page 2 with the B+ and RDecoupling added as in the 807 driver circuit.

Page 4 is truly a Series Resonant circuit with an added RF choke, which is the reason most designs use a || resonant circuit since it deletes the RF choke. Both tuning components must be isolated well from any body or chassis capacitance while tuning. It too has to be decoupled at the B+ end to avoid parasitic resonances.

Any tuning inductance or RFC should preferably have a Self-Resonant-Frequency (SRF) well below or well above the band of operation.


Phil - AC0OB

[KD1SH]

In the PDF file below, Page 1 is a parallel resonant circuit in which the parallel resonant circuit is in series with the B+. As long as the bottom end of the circuit is RF shunted (referenced) to ground, it should work just fine. This could be what the OP was talking about. He can clarify if he chooses.

Yes, that's what I meant.  I took the shunt cap as a given, so I didn't mention it.  Including it is good design practice, but I suspect the circuit would work fine without it since the tank at resonance is effectively a high impedance RF choke.

Page 2 is an equivalent parallel resonant circuit for my suggested driver. It completely decouples the DC from the tuning cap plates with the tuning capacitor's rotor at ground potential. The total RF current is circulating via ground through ~ 9 ohms. It functions very well with no capacitive proximity effect.

That makes sense.  But couldn't you decouple using just one cap, at the junction of XcX and LP1?  Since both of those caps would appear to be effectively in series with the circulating resonant currents, wouldn't they incur losses?  Maybe the losses would be inconsequential.

Page 4 is truly a Series Resonant circuit with an added RF choke, which is the reason most designs use a || resonant circuit since it deletes the RF choke. Both tuning components must be isolated well from any body or chassis capacitance while tuning. It too has to be decoupled at the B+ end to avoid parasitic resonances.

Now that's an interesting concept.  Has anyone actually done this?  I wonder how the tuning would be, since a series resonant tank operates inversely to a parallel tank, with the RF impedance approaching a theoretical zero at resonance?

[KD1SH]

The advantage of a blocking cap and  grounded resonance circuit is that the tuning cap can be grounded, no HV at the tuning cap and directly connected across the coil. . If you use a fixed tuning cap, that advantage disappears, but that you have to slug tune the coil. Disadvantage is that you need  an anode choke with possible losses and false resonances,

   Yes indeed - I wasn't thinking about that, but being able to ground the tuning cap is a big advantage.  I like a common RF reference throughout, and from a mechanical standpoint having to isolate the cap is just another complication.
   I don't like messing around with slug-tuned coils. I've got a bunch of nice ceramic forms with threaded rod slugs, but with unknown core permeability I always have to go through much experimenting to get it right.  Besides, after decades of hamfests, my nose is running air-variables, so I'd much rather tune that way.
   False resonances and "should-have-known-better" resonances - I've learned (the hard way) to pay attention to the self-resonance frequency in the specs when I order chokes from Digi-Key and such.

[DMOD]

Yes, that's what I meant.  I took the shunt cap as a given, so I didn't mention it.  Including it is good design practice, but I suspect the circuit would work fine without it since the tank at resonance is effectively a high impedance RF choke.

A parallel resonant circuit has to consist of a capacitor in parallel with an inductance. The only way any inductance will resonant at the proper frequency by itself, is if the self (distributed) capacitance of the coil is at the correct value.

That makes sense.  But couldn't you decouple using just one cap, at the junction of XcX and LP1?  Since both of those caps would appear to be effectively in series with the circulating resonant currents, wouldn't they incur losses?  Maybe the losses would be inconsequential.

At the low reactance values of the capacitors, any losses through the caps are inconsequential. There is only one decoupling (ground referencing) capacitor, XC2RFGroundReference.  If you are referring to Page 1, capacitor XC2RFGroundReference ( a .005 uF) shunts RF to ground referencing the bottom junction of CPT-LP1 to ground.

Now that's an interesting concept.  Has anyone actually done this?  I wonder how the tuning would be, since a series resonant tank operates inversely to a parallel tank, with the RF impedance approaching a theoretical zero at resonance?

The tuning would be tricky in high impedance circuits unless strict isolation is observed. Series resonant circuits are best used where amplifier output impedances are low, such as in the output circuits of ClassE transmitters.


Phil - AC0OB  

 

[KD1SH]

A parallel resonant circuit has to consist of a capacitor in parallel with an inductance. The only way any inductance will resonant at the proper frequency by itself, is if the self (distributed) capacitance of the coil is at the correct value.

   Sorry; I think we're talking about different capacitors.  Shunt caps; decoupling caps; blocking caps; terms with enough interchangeability to lead to confusion.
   Indeed, you can hardly have a parallel LC circuit without the C.  Distributed capacitance means there's no such thing as a "universal" choke - all will resonate somewhere due to distributed capacitance.
   What I was referring to was the "ground reference" capacitor going from the B+ end of the tank circuit to ground in your page #1, establishing a common ground reference but not participating in the resonance.  Basically I was musing about whether the circuit would function without it, since the tank circuit at resonance essentially would act like an RF choke.  Now that I think further on it, it probably wouldn't, since the positive excursions of the resonating tank would have no return path to ground.

[DMOD]

A parallel resonant circuit has to consist of a capacitor in parallel with an inductance. The only way any inductance will resonant at the proper frequency by itself, is if the self (distributed) capacitance of the coil is at the correct value.

...Now that I think further on it, it probably wouldn't, since the positive excursions of the resonating tank would have no return path to ground.

Agreed and it can be confusing but the application of the cap is usually determined by the context.

The "ground reference" cap 'XC2' Page 3, provides a return path to ground for the RF currents and has no part in the resonance of the LC circuit, so the RF doesn't get coupled into the + side of the DC power supply, and acts to present a low impedance path to ground reference.

XC2 "shunts" RF to ground because of its low impedance.

A Decoupling or Blocking cap usually refers to the blocking of DC and the coupling of AC, such as XcX and CPC on Page 3, and has no part in resonance.

If you don't want DC on the cap rotor and want the stator to be both mechanically and electrically at ground reference, XcX blocks the DC but allows the RF currents to appear at CPT. XcX has such a low reactance that it does not materially affect overall tuning. For example, if CPT is set to 45 pF, the total tuning capacitance across LP1 is 44.798 pF.

The reason for the stator to be both mechanically and electrically at ground or chassis reference is mechanical rigidity and placing it electrically at ground reference.

Now XcX was put in the circuit in case one or more of the rotor plates short to a stator plate from accidentally being bent. Being a clutz at times, this has happened to me and not seeing the shorted plates, I had some smoke at power up.   Another reason for having an RDecoup.


Phil