Designing an ATU

Most ordinary field engineers will not need to design an ATU in the course of their normal duties. However, knowing the theory behind it can be very helpful when troubleshooting problems.  Also, fewer and fewer people understand RF these days, especially when it comes to AM.  Knowing a little bit can be an advantage.

We were working on an AM tower recently when several discrepancies were noted in the ATU:

WFAS ATU, 1230 KHz
WFAS ATU, 1230 KHz, 1 KW, N-DA

This was connected to a 202° tower. There were several complaints about seasonal shifts and narrow bandwidth. The VSWR meter would deflect slightly on high-frequency audio peaks, always a bad sign.  A little bit of backstory is in order.  WFAS signed on in 1930 using a four-legged self-supporting tower.  This tower was used until about 1986, when it was replaced with a series excited, guyed tower.  The ATU in use was initially designed for the replacement tower, which likely had a good bit of capacitive reactance.  I am speculating on that, as I cannot find the original paperwork for the replacement tower project.  At some point, somebody decided to ground the tower and put a skirt on it, likely to facilitate tower leasing.  The skirt was installed, but the ATU was never properly reconfigured for the high inductive reactance from the skirted tower.  The truth is, the Collins 820-D2 or Gates BC-1G tube-type transmitters probably didn’t care.  They were probably like; bad load, meh, WHATEVER!  Although the audio quality likely suffered.  That all changed when Broadcast Electronics AM1A was installed.  To fix the bad load problem, a BE 1 KW tuning unit was installed next to the transmitter.

Technically, there are several problems with the above circuit, starting with the capacitor on the wrong side of the base current meter.  This capacitor was installed outside of the ATU between the tower and ATU output.  Was the base current meter really measuring the base current?  I don’t know, maybe? The shunt leg was lifted but both of the inductors of the former T network were left in the circuit.

We reconnected the shunt leg and moved the capacitor inside the ATU and to the correct side of the base current meter.   After several hours of tuning and fooling around with it, the ATU is still narrow-banded, although now at least the input is 50Ω j0. I believe the current design has too much series inductance to be effective.

Thus, a redesign is needed.  I think, because of the inductive reactance of a skirted tower, a phase advance T network will lead to best bandwidth performance.  The basic design for a +90-degree phase advance looks like this:

WFAS -90 lagging ATU
WFAS +90 phase advance ATU, 1230 KHz, 1 KW, N-DA

To calculate the component values for the ATU, some basic arithmetic is required.  The impedance value for each leg in a +/- 90 degree T network can be calculated with the following formula:

Z = √(inputZ × outputZ)

Where Z = impedance per leg
Input Z = the ATU input impedance, 50Ω
Output Z = the antenna resistance, 58Ω

Thus:  Z =√ (50Ω × 58Ω)

Z = 53.85Ω

Formula for Capacitance: C = 1/(2Π × freq × XC)

Thus for the input leg: C = 1/(6.28 × 1.23MHz × 53.85Ω)

C (input)  = 0.0024 μF

Formula for Inductance:  L = XL/(2Π × freq)

Thus for the shut leg: L = 53.85Ω /(6.28 × 1.23 MHz)

L (shunt) = 6.97 μH

For the output leg, we must also consider the inductive reactance from the tower which needs to be cancelled out with capacitance.  Thus, the output capacitor needs to have a value of 53.85Ω + 580Ω = 633.85Ω

Thus for the output leg: C = 1/(6.28 × 1.23MHz × 633.85Ω)

C (output)  = 0.000204 μF

The amazing thing is, all of these components are available in the current ATU, they just needed to be rearranged.  The exception is the vacuum variable capacitor, which I salvaged from an MW-5 transmitter many years ago.  I donated that to the project, as I am tired of looking at it in my basement.  The reason for the vacuum variable capacitor will become evident in a moment.  The input capacitor will be slightly over value, which will require the inductor to tune out the excess capacitance.  A good design rule is to use minimum inductance to adjust the value of a fixed capacitor, thus the capacitor should be not more than 130% of the required value.

About the Vacuum variable output capacitor; in the existing ATU had a 0.0002 μF capacitor already.  With a +90° phase shift, this capacitor is likely adequate for the job.  The vacuum variable may be pressed into service if something other than a +90° phase shift is needed for optimum bandwidth.  That will be the topic of my next post.

Final consideration is the current and voltage ratings of the component.  As this is a re-build using existing components, chances are that they already meet the requirements.  On a new build or for replacing parts, one must consider the carrier power and modulation as well as any asymmetrical component to the modulation index.  For current and voltage each, the value is multiplied by 1.25 and then added to itself.  For a 1,000 watt carrier the input voltage on a 50 ohm line will be approximately 525 volts at 10 amps with 125% modulation.  A good design calls for a safety factor of two, thus the minimum rating for component in this ATU should be 1050 volts at 20 amps, rounded up to the next standard rating.   The capacitor on the output leg should be extra beefy to handle any lightning related surges.

The current rating for a capacitor is usually specified at 1 MHz.  To convert to the carrier frequency, the rating needs to be adjusted using the following equation:

IO = IR√ FO

Where:
IO: current rating on the operating frequency
IR: current rating at 1 MHz (given)
FO: operating frequency in MHz

The vacuum variable output capacitor is rated for 15,000 volts, 42 amps.  Adjusted for frequency, that changes to 46 amps.  The calculated base current is 4.18 amps carrier, 9.41 amps peak modulation.  Thus, the capacitor on hand is more than adequate for the application.

The Smith Chart

I have been fooling around with Smith Charts lately. They look complicated, but are really pretty easy to understand and use, once you get around all those lines and numbers and stuff. Smith charts offer a great way to visualize what is going on with a particular antenna or transmission line. They can be very useful for AM antenna broadbanding.

Smith chart
Smith chart

.pdf version available here: smith-chart.

The first thing to understand about a Smith chart is normalization. Impedance and reactance are expressed as ratios of value units like VSWR. A ratio of 1:1 is a perfect match. In the center of the Smith chart is point 1, which expresses a perfect match. To normalize, the load resistance and reactance are divided by the input resistance. Thus, if the input resistance is 50 ohms and the load impedance is 50 ohms j0, then the normalized Smith chart point would be 50/50 or 1. If the load impedance is 85 ohms and the reactance is +j60, then the normalized Smith chart point would be .58 1.7,1.2.

More basic Smitch chart usage information in this video:

I touched on the black art of AM antenna broadbanding before. It is a complex topic, especially where directional antenna systems are concerned, as there are several potential bottlenecks in a directional array. To explain this simply, I will use an example of a single-tower non-directional antenna.

Below is a chart of base impedance from a single tower AM antenna on 1430 KHz.  The tower is skirted, 125.6 degrees tall.  An AM tower that is expressed in electrical degrees is denoting wavelength.  A 1/4-wave tower (typical for AM) is 90 degrees tall. A 1/2 wave tower is 180 degrees tall.  Thus this tower is slightly taller than 1/4 wavelength.

Frequnecy(khz)ReactanceReactance (normalized)ResistanceResistance(normalized)
1390-j 139-2.784058.1
1395-j 143-2.864008.0
1400-j 147-2.943507.0
1405-j 146-2.923106.2
1410-j 142-2.842705.4
1415-j 132-2.642364.72
1420-j 125-2.502104.2
1425-j 118-2.361903.8
1430-j 112-2.241703.4
1435-j 106-2.121553.1
1440-j 100-2.001382.76
1445-j 93-1.861252.5
1450-j 86-1.721142.28
1455-j 79-1.581042.08
1460-j 75-1.50951.9
1465-j 70-1.4921.84
1470-j 65-1.3851.7

The base impedance is not too far out of line from what is expected for a tower this tall.  Plotted on a Smith Chart:

1430 base impedance plotted on a Smith chart
1430 base impedance plotted on a Smith chart

One of the first principles behind broadbanding an AM antenna is to distribute the sideband energy evenly and have a symmetrical VSWR.  The antenna tuning unit will match the line impedance to the load impedance and cancel out the reactance.  Having the proper phase advance or phase retard rotation will distribute the sideband energy symmetrically about the carrier.   To determine phase rotation, the cusp of the plotted graph is rotated to face either the 3 o’clock or 9 o’clock position (0° or 180°).  The cusp is where the direction of the line changes, which in this case is the carrier frequency, 1430 KHz.  The above example, the line is fairly shallow, which is typical of a skirted tower.  Thus, the best phase rotation to start with is +79°.  This will likely be close but will need to be tweaked a bit to find the optimum bandwidth.  After looking at the plotted Smith chart, my first inclination would be to reduce the rotation, more tower +75° as a first step in tweaking.

When working with AM systems, the bandwidth of the entire system needs to be examined.  That means that final bandwidth observations will need to be made at the transmitter output terminal or in some cases, the input to the matching network.  It varies on system design, but things like switches, contactors, mating connectors, ATU enclosures, etc can also add VSWR and asymmetry.  Broadbanding even a simple one-tower AM antenna can require quite a bit of time and some trial and error.

I will touch on ATU design in the next post.

Troubleshooting an AM array

Today, there will be a quiz.

Recently, we had an AM antenna array go out of tolerance by a good margin.  This has been repaired, however, I thought I’d post this information and see if anybody could identify the problem and the solution. Unfortunately, I don’t have prizes to give away, however, you can show off your AM engineering prowess.

All of the information is pertinent:

  1. The station has two directional arrays (DA-2) using the same towers; the nighttime array is out of tolerance, and the daytime array is not affected and is performing normally.
  2. There were no weather events connected with this event; no electrical storms, no major temperature changes, no rain events, no freezing or thawing, etc.
  3. The problem happened all at once, one day the array was performing normally, and the next day it was not.
  4. Station management reports that some listeners were complaining that they could no longer hear the station.
  5. The ATUs and phasor were inspected; all RF contactors were in the proper position, no damaged or burned finger stock and no evidence of damaged components (inductors or capacitors) was observed.  Several mouse nests were cleaned out of the ATUs, however, this did not change the out-of-tolerance antenna readings.
  6. The towers are 1/4 wave (90 electrical degrees) tall.

Readings:

TowerPhase angle as licensedCurrent ratio as licensedPhase angle as readCurrent ratio as read
1147.20.583149.50.396
2 (reference)01.0001.00
3-1370.493-125.80.798
4107.50.48192.70.355
5-38.10.737-60.20.623
6-178.70.382142.80.305

Licensed values for common point current is 13 amps, impedance is 50 ohms j0 and there is normally no reflected power on the transmitter.  On this day, the common point current readings were 8.9 amps, impedance 38.5 ohms +j5 the transmitter had 340 watts of reflected power.

This is the overall schematic of the phasor and ATU:

WDGJ overall RF schematic diagram
WGDJ overall RF schematic diagram, click for higher resolution

Aerial view of the transmitter site, oriented north:

WGDJ aerial view showing towers as identified in schematic diagram
WGDJ aerial view showing towers as identified in the schematic diagram

So, where would you begin?  Ask questions in the comments section.