LCL Networks in place of Pi networks -- much better.

[IN3IEX]

Let's experiment something NEW  .

After deep  studies it was found that the Pi network cannot match every load. ( )
With reactive antennas most of us use the common Pi net in the TX for matching to 50 ohm (with good Q and low pass functionality) followed by the very versatile T (CLC) matching network (that is high pass).

Here we have a matching solution (T - LCL)  that behaves like the combination of the above:
http://www.ing.unitn.it/~fontana/LCL%20Matching%20Networks.pdf

The paper is a convincing example. The full theory is nice but diffcult to apply.

Here we have similar Z-plots for a Pi network. http://www.ing.unitn.it/~fontana/SimpleP.pdf

Anybody interested in testing the above simulations?

[KM1H]

There is nothing new there, its been used to feed into VHF/UHF GG amps for many decades. And the common output Pi-L has been around even longer and is used in several commercial amps.

Carl

[IN3IEX]

Anyway it looks very interesting for HF for many reasons.

[DMOD]

Hi Giorgio,

The "T" network has been used for many years in AM Broadcast, both as a primary ATU network and as a phase shifter for Directional Antenna systems.

In many cases, a series inductor in inserted on top of the shunt capacitor to increase tuning range, in lieu of switching parallel caps across the main variable cap.

I do find it more useful at the lower HF frequencies of 80 and 160m for tuning verticals.

I have seen some linear amp GG circuits use the T network in the cathode circuit rather than PI networks.

Phil - AC0OB

[IN3IEX]

Thank you for the interest.

Obviously all those networks are known and used since the origin of radio transmissions, but how they comparatively perform, within a frequency range, on the full complex plane is still a matter of discussion, including how to present the results of the analysis (for enabling practical testing and applications).

I attach a recent paper on a Q design method for T (LCL) networks, with some frequency response plots.

If you look at my recent set of simulations, you see that the 60uH +600pF+60uH (all variable) may match a power tube to almost any antenna on the full HF spectrum. This is a remarkable result. 600pF is quite less than the thousands of pF required for the Pi network load capacitor (at 1.8MHz), that has also a quite limited matching ability.

The cost of two roller inductors is high, but we do not have the band switch, no HV RF fixed capacitors in the matching network, no Pi coil and only one variable C, that is why I believe that T (LCL - low pass) networks can be advantageously adopted in place of common Pi networks in tube HF class C transmitters and tube linear amplifiers.

Possible drawbacks come from the voltage on the capacitor. In the 7 MHz example with highest Q (second example) the RF voltage on the capacitor is 2.6 times the RF voltage on the plate.  In the last 7 MHz example the RF voltage on the capacitor is almost equal to RF plate voltage.  

[K5UJ]

I had a T network last year, a Heathkit SA-2040.  I got it all cleaned up and nice and new looking then I read through the manual and dug up an article on T networks in an old QST issue and found out they are high pass networks.     High pass = harmonics  .  Especially with an old link coupled output network as found on the BC610, Eldico TR-1 and other rigs.   I already had Matchboxes for balanced line, and a semi HB L network for my coax lines so I sold it at the next hamfest.   I'd like to see a T network used in the output of a RF PA matching the anodes to a feedline going to a discone and measure the field strength of the second harmonic.   I have a hunch we'd find out through that experiment why T networks are not used in commercially manufactured vacuum tube RF amplifiers.

[IN3IEX]

Hi K5UJ.  Most if not all "T network" tuners are high-pass filters. They are C series - L to ground - C series.
I am proposing a L series - C to ground - L series direct Anode -> SO239 tuner/tank circuit. It is low-pass "more" than the Pi network.
I am doing some analysis during spare time.
It seems that the most serious drawback is high voltage on the capacitor under high Q tuning. The Q and the voltage will be limited by losses, but I do not consider losses in this preliminary phase. I consider only the ability of matching complex impedances and compare to the Pi network.
Please consider that I have not built such circuit. The tuning procedure is still obscure. For instance if I tune for maximum output does it mean that losses are minimized, therefore the voltage on the capacitor in minimum? What do you suggest?

Giorgio

[Steve - K4HX]

For a C-L-C topology, the lowest loss occurs when the L is minimized and the output C is maximized. The inductors will generally be lower Q than the caps, so more loss occurs. On the face of it, a circuit with two inductors is will likely have more loss than a circuit with only one. Just how much difference is the question and more often than not, the difference will be small enough to not be of concern.