Antenna Tuners

Under Construction this page needs work because I modeled the L networks with a higher Q inductor than the T network.

Component Q

A typical #14 or 16 gauge B&W Airdux or Miniductor coil might have a Q of 300 in open air, and that Q might drop to 260 or so with sheet metal near the axis.

Here are some actual measurements of a small 3" diameter air core coil. This particular coil uses fairly thin wire and is not of optimum form factor, so basic open-air Q is only around 160. The following measurements are with the coil axis spaced around 1/4-inch from a large flat metal sheet. This coil is from a MFJ 300 watt tuner. Measurements are on 2 MHz. Measurements are made on an HP4191A impedance test set (the industry standard for measuring components):

Air for several inches around the coil = 160
Aluminum = 150
Copper = 152
steel = 142
stainless steel = 70
nichrome sheet= 62

We can see some materials greatly decrease Q, but most materials of reasonable or good conductivity have little overall effect on Q.

The following data was supplied by G3TXQ Steve. When I enter my data in a spread sheet, I will post it also. The trend of Steve's data agrees with my measurements, although there are differences in data because I measured different inductors. I measured less peak Q, but also significantly higher minimum Q in the inductors I tested. While the inductors I measured has less delta in Q, they also had a serious dip in Q up near 25-35 MHz when near half-inductance. This is because of series resoances.

Notice that shorted turns do NOT have the drastic effect sometimes predicted in amateur articles. For example, look at the 160 meter Q. It actually peaks with some turns shorted!

Do not worry about shorted turns reducing Q in a normal air-core inductor. There are cases where Q can be seriously altered, but they are not common cases. I will post measurements of shorted turns and Q as I measure new design projects.

T-network vs. L-network

One argument that never seems to end is which network is better, a T-network or an L-network. This argument goes on continuously because each network has advantages and disadvantages, and in actual fact loss differences are really fairly small.

If we budget similar amounts of money, the L-network generally has a little less loss. The saving grace of the L-network is it just won't match a bad load impedance. The L-network "self-protects", and generally won't match loads that greatly increase network heating.

The T-network has much wider matching range! The T-network also requires less switching and fewer components.

The circuits below assume the following component values:

L-Network = capacitance 15-2700 pF (Q=3000 overly optimistic at high capacitance), inductance .1 to 30 µH (Q=200). Note: I'm working to correct the Q's and make them more realistic.

T-Network = capacitance 10-800 pF (Q=3000), inductance 30 µH (Q=200).

The L-network's capacitor must vary from 15 pF all the way up to 2700 pF, and at one kilowatt the capacitor has to be over 2000 volts rating if the tuner is going to feed high impedance loads. This means we either use a large vacuum capacitor, or the design must switch various fixed capacitors in parallel with a lower-capacitance high-voltage variable capacitor.

The Q for the T-network is probably pretty close to normal values except for low inductance values.

The Q for the L-network is far too high for fixed capacitors (should be around 1000 or less), and too high for the inductor across the range (way too high at low inductance).

L-networks have a very wide range of impedances they just will not match, and they require extraordinarily high values of capacitance to do what they can do. To approach a T-network in matching range, they would require switching the network into four possible configurations plus the extraordinarily large values of capacitance. Because of these restrictions, L-networks have limited utility on a 1.8-30 MHz all purpose tuner. L-networks are great for matching known load impedances on a single band, but are cumbersome and expense to properly implement in a wide range general matching network application. If you are lucky with load impedances, you will swear by them. If you are not lucky, you will swear at them because they just won't match things a cheaper poorer quality T network will easily match.

Varying Network Q

Loaded or operating Q is important in any impedance matching network. Loaded Q is the actual working Q, not the individual component "static" Q or "quality". For example if we have a capacitor with a Q of 3000 and a reactance of 3000 ohms, it has a series resistance of 1 ohm (or a shunt resistance of 9-megohms). This is quite possible in an air variable capacitor, but few small ceramic single layer capacitors produce such high Q's. If we shunt that capacitor across a load of 300 ohms, the loaded Q becomes just slightly less than 10.

Phase shift and network loaded or operating Q are closely related. The greater Q for a given impedance transformation ratio, the greater network phase shift becomes. The table below assumes a 1:1 impedance ratio. Q will be higher for larger impedance ratios.

A common rumor is there is only one optimum setting for maximum efficiency in a T-network. Let's look at some numbers using realistic "amateur product" Q values of 200 in the inductor and 3000 in the capacitors. This produces numbers where a rule of thumb works. With 1000 watts in, the loss in watts is approximately equal to phase shift up to 90 degrees!!! The loss table for 1000 watts applied, when matching 50-ohms to 50-ohms, would be:

From 0 degrees to 130 degrees phase shift, a T-network with modest Q components has less than 0.1 dB loss!

The real problem with loss is power handling, not signal loss. We need to be very careful where a few watts of loss occurs, because even a few watts can heat something very small to a high temperature. We would like loss to be spread out over a wide area in physically large components so the heat "gets out". Most of the time it is, but unfortunately some of the time it is not. Careful tuning will minimize potential problems.

Proper Way to Adjust a T-network Tuner

There is little change in efficiency or performance over a wide range of Q's. This is because loss is very low, so even doubling loss makes very little difference in power. Doubling loss will of course double heat, but the larger problem is the increase in capacitor voltages. Arcing can cause a tuner to fail even while loss in signal level is totally undetectable at the other end of a QSO. Arcing is a fast failure and very destructive, because it produces localized heating. Besides increasing voltages and heat, increasing Q also decreases bandwidth. This means a tuner requires more frequent retuning when operating frequency is changed within a band.

In order to minimize the need to retune and maximize the power rating of a tuner, it is necessary to tune or match with the minimum possible operating Q. In a T-network, we want minimum inductance and somewhere around the maximum possible capacitance that allows a good match. The quickest way to damage a switch or capacitor in a tuner is to adjust the tuner with too much inductance.

The worse advice, when tuning a T network tuner, is to start with the capacitor mid-way and adjust the inductor for maximum noise or minimum SWR. On lower bands the capacitors should be set to maximum capacitance as a starting point. Normally halfway open is a good place for the higher bands, typically 15 meters and higher in a 160-10 meter tuner.

The best starting places for capacitors would be:

160-80 meters=fully meshed

60-20 meters=3/4 meshed

20-10 meters=1/2 meshed

Matching Range

This is the lower impedance limit of an L network with 2700 pF of capacitance. It will not match any positive (inductive) reactance above 22.9 ohms on 1.8 MHz.

This narrow matching range, even with large values of components, is one drawback of an L network.

This is the maximum capacitive load that can be matched at 15 ohms resistive. At 15 -j317 the network is out of inductance.

The L network has poor matching range for low impedance loads, and requires a very large capacitor for 1.8 MHz.

The L network has a good reactance range if we lower the resistance down from maximum, but the capacitor and switch must be able to handle 2150 volts at 1000 watts.