All posts by Martin Blustine

I studied physics and went on to work in infrared optics, millimeter wave and microwave engineering until retirement. My interests lie in teaching, music, radio astronomy, infrared systems and microwave and antenna engineering. I enjoy writing technical papers about ham radio topics. When I am not operating CW, I enjoy homebrewing ham gear and restoring vintage HP and Agilent test and measurement equipment.
Featured, General, Station Equipment

Restoration of an Agilent 53131A Frequency Counter

An enjoyable hobby is the restoration of vintage HP, Agilent, Tektronix and GenRad test equipment. In their day, these brands represented some of the finest U.S. engineering and manufacturing in the world, and two of them still do. What is common to all of them is the attention paid to ergonomics, product design and documentation.

Each time I restore one of these instruments, I gain insight into what the design engineers and the product designers had in mind.

Recently, I acquired an Agilent 53131A Frequency Counter found on eBay, one of my favorite sources for old test equipment. It is always a calculated risk because I never know for sure if whatever I have bought will arrive dead on arrival. Fortunately, eBay is very good about standing behind anything sold on their site that has been represented to be in good working order. They are unlike auction consolidators such as Allsurplus where everything is sold as is.

Naturally, when the counter arrived, I subjected it to the obligatory smoke test. The result is shown if Figure 1. The counter completed its self-test routine and rewarded me with a series of dashed lines in the display. That was a good sign, and it was off to a good start. When I purchased the unit, I made a best offer, and it was accepted. The reason that I did that is because I noticed some minor irregularities in the unit.

First of all, the display appeared to be dim in the photo, the likely result of running the counter for long durations for many years, maybe 30. The second thing that I noticed was the absence of the front and rear bumpers as well as a carrying handle. These, I am quite sure, were scavenged for resale on eBay. I have become quite used to that. Most of the instruments that I have restored were received minus their bottom and rear feet and in some cases their rack handles – a minor annoyance. I simply buy some and put them back on.

Most often these counters are purchased for HF use and do not feature prescaler options to extend the frequency range. This counter was no different. It was also shipped with the standard low-stability clock oscillator.

Figure 1. Agilent 53131A Frequency Counter, As Received. The display is dim but the unit passes self-test and enters ready mode. The unit was advertised and shipped without its front and rear bumpers as well as its carrying handle. The time base is a standard low-stability one, and there is no frequency extender option installed.

Upon seeing the condition of the display, I looked at the power supply schematic and found the test points for VFD display bias. The voltages checked okay. That test having been completed; I ordered a new display on eBay from a seller in China. The old display was marked Japan, but I could not find any other suitable displays for sale other than one in a non-working counter being sold for parts only.

I also looked for an inexpensive prescaler to install as an option. The one chosen works up to 3 GHz but others with higher division ratios were also available. This one provides accurate results for signals as low as -15 dBm, which is good enough for my purposes.

There were many sources for the front and rear bumpers and the carrying handle on eBay. I likely bought my own back.

All parts arrived in less than two weeks, which is about average for items ordered overseas that must pass through U.S. Customs.

An initial test was run on the counter to make sure that it would arm and run. I used whatever signal source was closest for this test, in this case, an HP 3314A Function Generator. The setup is shown in Figure 2. Both units are out of calibration so, it can’t be determined which is worse, the counter or the function generator.

Figure 2. Counter, As Received, Measuring 18 MHz. The most convenient signal source for initial testing was a bench-top HP 3314A Function Generator. Both the counter and the function generator are out of calibration. It’s difficult to know which is worse.

The first part to arrive was the frequency prescaler. The only disassembly required for installation was the instrument top cover.

I fastened the prescaler to the wall of the chassis on a pair of standoffs. The prescaler board was supplied with a coaxial cable for connection to a front panel SMA to BNC adapter and a ribbon cable for connection to the motherboard. The whole installation process was completed in an hour. The result is shown in Figure 3. The sensitivity of the prescaler was tested with a test signal of 2.55 GHz, as this particular synthesizer won’t tune much higher than that. The signal level was set to -10 dBm but the prescaler will provide stable counts for signals as low as -15 dBm.

Figure 3. Prescaler Test. The prescaler was tested with an HP 8663A Synthesized Signal Generator set to 2.55 GHz because it is the only microwave source available. A stable count is observed for power levels as low as -15 dBm. The front and rear bumpers as well as a carrying handle have been installed.

Once the vacuum fluorescent display arrived, I began to tear down the counter to the level required to remove the front panel. This also permitted cleaning 30 years of grime from the enclosure and circuit cards. Figure 4 shows the teardown.

Figure 4. Counter Teardown. The counter was disassembled using the assembly-level service guide. The power supply and motherboard remain in place. Only the front panel required removal.

Figure 5 provides a closer view of the front panel PCB and the old display. The new display is in the foreground. While the two displays are functionally the same, they have completely different layouts and are of different manufacture.

Figure 5. Close-up of the Front Panel and Vacuum Fluorescent Displays. While the displays look alike and possess the same pin-outs, they are entirely different designs.

The old display was removed from the front panel display printed circuit board that also contains the soft contacts for the front panel controls. Utmost care was taken in unsoldering and removing the old display. I wanted to preserve the old display if the new display was dead on arrival.

The unsoldering process was the one that I always use. The solder joints were painted with non-corrosive soldering flux. Next, each joint was resoldered with compatible solder. This is an important step because old solder joints tend to degrade with time. The newly resoldered joints are more easily heated and solder more easily removed with a solder puller like the one shown in Figure 6. This is an inexpensive tool that, when mastered, can remove all of the solder from the solder joints. Sometimes it takes three or even four passes with the solder puller on each joint to remove all of the solder from the holes. Once all of the solder has been removed, a needle nose pliers is used to wiggle each of the leads to ensure that it is free of solder and free from each hole. Another technique can be used provided that the replacement part is known to be good. Simply snip the leads of the old part and remove it. Then, there are still leads that have to be removed and holes that have to be cleaned. Some find this technique easier. Avoid the use of solder wick if at all possible. It really isn’t needed, and it usually ruins the circuit pads.

Figure 6. A Generic Solder Puller. This is a simple device that is just a spring-loaded plunger that, when released, can vacuum solder from solder joints. It pays to master the use of this device. Avoid the use of solder wick.

The old VFD was removed from the front circuit board. The old part and the new part are shown in Figure 7 for comparison. The two parts are of different provenance.

Figure 7. Old and New Vacuum Fluorescent Displays. The designs are very different: (top) the old part, (bottom) the new part. The pins on the old part are intact.

The display circuit card is shown in Figure 8. Non-corrosive soldering flux was painted on each solder joint and each joint was resoldered. This step is essential for old solder joints. Four passes were required to remove all of the solder with the solder puller. The board is undamaged as is the old display part.

Figure 8. Display Printed Circuit Board. The display was removed from the board with a solder puller. Four passes were required to remove all of the solder without damaging the pads. Flux residue was removed from the PCB with isopropanol.

The new display was soldered into the display circuit card and the PCB was cleaned with isopropanol to remove residual soldering flux and debris. The board was examined with a magnifier to ensure that the board was free of solder blobs and solder bridges. The counter disassembly procedure was reversed to reinstall the display circuit board into the front panel. Next, the front panel was fastened to the chassis, and the cover was replaced. The result is shown in Figure 9. The display brightness is just like new.

Figure 9. New Display Brightness. The new display works as advertised. A microwave signal of 2.55 GHz is displayed. The new prescaler BNC connector is marked Channel 3. The new front and rear bumpers and carrying handle are also visible. Please note that the difference in brightness across the display is an artifact that is due to the refresh rates of the counter and the camera.

One final modification will consist of the addition of a stable source to replace the standard one that drifts. Once complete, the instrument will be calibrated.

 

Antennas, Featured, General

Matching to the Complex Load Impedance of a Shortened, Non-Resonant Antenna – Part I

3 Comments

Matching to the Complex Load Impedance of a Shortened, Non-Resonant Antenna – Part I

Introduction

A common impedance matching problem is that of matching a 50 ohm transmitter to a shortened non-resonant antenna. Examples of non-resonant antennas are the 23-foot (7.01m), and the 43-foot (13.1m) backyard vertical antennas. These antennas have something in common. They exhibit high capacitive reactance.

It is hoped that this multi-part article will provide the reader with the tools necessary to contend with this common antenna matching problem.

Part II of this article will discuss impedance matching to a 43-foot backyard vertical antenna using a 1:1 UNUN and an autotransformer[1].

The high voltages developed in these matching networks will be the subject of Part III of this article[2].

Example 1: 23-Foot Backyard Non-Resonant Vertical Antenna

A 23-foot backyard vertical with numerous radials exhibits an impedance of 19.79 – j631.9 ohms at 3.6 MHz at its base.

For this exercise, we don’t care about the number of radials, conductor losses, ground losses, and reflected power that all figure into efficiency. All we care about is matching to this complex impedance.

Unmatched VSWR

Let’s calculate the unmatched VSWR at the base of our 23-foot backyard vertical before we apply matching techniques. The load impedance that must be matched is assumed to be

The source impedance of the transmitter is given by

To calculate VSWR, we need to relate these two complex quantities to the magnitude of the reflection coefficient,

Those of us who own nanoVNAs have become used to the term S11, which is the input voltage reflection coefficient. By definition

This is a complex number that has to be rationalized before its magnitude can be found.

We should begin by combining terms where possible.

VSWR: Method I – Rectangular Form

Rationalize the denominator.

Like terms are combined to find the reflection coefficient.

The magnitude of the reflection coefficient is obtained from

Finally, the VSWR is computed from

The VSWR is 406:1 due to the high value of capacitive reactance.

VSWR: Method II – Polar Form

As before, and after combining terms, we begin with

and let

By dividing Z1 by Z2, we obtain

Notice that when the angle is moved from the denominator to the numerator, the sign changes.

We may stop here since we already have what we need

or, for the exercise, we may convert back to rectangular form using the Euler identity and arrive back at the same place.

Rectangular and polar forms lead to the same result.

The high VSWR is due to the high value of capacitive reactance of the unmatched load impedance.

Matching Techniques

When faced with problems like these, it is often easier to break the problems down into more manageable steps.

How to match the real part of the load impedance was the subject of an earlier paper[3] but let’s review the procedure for this case.

We begin by inspecting the impedance to see what we can learn about it. The real part is 19.79 ohms. This resistance, RL, is smaller than the real 50 ohm transmitter impedance, RS. If we were to use a simple LC matching network, an L-network, we can see from Figure 1 that there are four possible topologies: two low-pass topologies and two high-pass topologies[4]. Notice that the low-pass topology is capable of conducting DC from input to output. This is not possible with the high-pass topology that blocks DC with a series capacitor. We begin by picking the low-pass topology in Figure 1(b) for which RS > RL.

Figure 1. L-Matching Network Topologies. Source and load impedances are real. Reproduced under CC BY-NC by permission from Michael Steer, North Carolina State University.

It is instructive to work through low-pass and high-pass topologies to illustrate how these problems are solved. It is important to note that all of these solutions result in an impedance match for a narrow band of frequencies. If multi-band operation is required, the use of multiple matching networks or the use of an antenna tuner (preferably a remote one) will be necessary.

Other matching techniques are possible, such as center loading and top loading but we will limit the discussions in Part I, Part II, and Part III to base loading.

Example 1: Low-Pass Topology

We begin by writing down what we know,

Step 1

We set the imaginary part of the load impedance, ZL, to zero for Step 1 of the solution. We will revisit the reactive part in Step 2 of the solution.

Thus,

 

We make use of Figure 1(b) to compute the unloaded Q for the L-matching network

and

where

Substituting, we have

Also,

At 3.6 MHz, the matching network inductance is

and the matching network capacitance is

Step 2

Our matching solution is incomplete until we cancel the remaining part of the load impedance that we ignored earlier, i.e., the imaginary part of the load impedance,
-j631.9 ohms, must be canceled to achieve a match. The secret to achieving a match is in finding a value of inductance that resonates with the capacitive reactance. Once completed, the series combination of load capacitance and added resonant inductance will result in zero reactance at the resonant frequency. We remember that for series resonance, ignoring any losses in the inductor and capacitor, the LC resonant pair looks like a short circuit at the design frequency. That’s exactly what we want – we want the load capacitance and the additional inductor to look like zero ohms at resonance. Of course, real inductors have series resistance due to the wire and capacitors have dielectric losses but for this exercise, we assume that they do not.

Thus, at resonance we have

where

C is the capacitance equivalent to the complex part of the load impedance, not the matching network, C‘, in units of Farads.

Thus,

The value, L”, is added to L‘. This will result in a new value for the L-matching network inductor, L”’

We don’t have to add the two together, and it may be easier to think about what each of the inductances does if we leave them as separate components.

This matching circuit may be simulated using RFSim99[5]. The circuit model is shown in Figure 2. The inductors L‘ and L” are drawn separately for emphasis. (Two inductors in series may be added.)

Figure 2. Low-Pass L-Matching Circuit. The inductor, L‘=1.08 μH, and added resonating inductor, L”=27.94 μH, are shown separately for clarity.

Return Loss and VSWR

In the VSWR section, above, we calculated the VSWR from

We may also calculate VSWR from the return loss, RL, directly from

where we have divided the return loss by 20 because the return loss is the voltage return loss.

Let’s calculate the VSWR for our simulation, where the return loss is 50 dB.

The resulting return loss for the low-pass matching circuit is shown in Figure 3. The 2:1 bandwidth of the matching network is ~78 kHz. The return loss is better than 50 dB at 3.601 MHz, or better than 1.01:1.

Figure 3. Return Loss for Low-Pass Matching Circuit. The 2:1 bandwidth of the matching network is ~78 kHz. The return loss is better than 50 dB at 3.6 MHz, or better than 1.01:1.

It may be concluded that this two-step matching technique for a low-pass matching network works quite well for our shortened antenna on the 80m band.

Example 2: High-Pass Topology with Series to Parallel Load Conversion

For the high-pass L-matching network Figure 1 shows that two topologies are possible. The one chosen would depend on how the capacitive reactance of the load is to be canceled.

There are two ways to accomplish this depending on where the inductance is in the matching network. The inductance in the high-pass configuration is always parallel at the input or parallel at the output. The position of the inductor depends on which is larger, RS or RL. The matching L-network inductor is always closest to the larger of the two as is seen in Figure 1.

The capacitance in the load is normally thought to be in series with the load resistance. This capacitive reactance could be canceled with a series inductor added to the L-matching network but there is a more interesting way to do it.

If the series RC load combination was converted to parallel form, as is often done for us on our VNAs, it would be observed that the new parallel resistance has a value that is much higher than the source impedance. By necessity, that would place the inductor in the high-pass L-matching network in parallel with the parallel capacitance of the load.

If we were to think about parallel resonance, and neglecting any losses in the load capacitor and resonating inductor, at resonance the pair looks like an open circuit. Then, the L-matching network just sees the real part of the load impedance.

The solution begins by converting the load from series form to parallel form.

Step 1

As it turns out, there is a transformation between series and parallel circuits that works at a single frequency. As was the case for the low-pass L-matching network, we ignore the reactive part of the load, initially. and incorporate it into the solution later. The series to parallel transformation works when

and

and

Then,

and

where

The Q-value is higher than we would like, but let’s proceed to see what happens.

There is enough information to derive the values

At 3.6 MHz the parallel load capacitance becomes

We observe that while the load resistance changes a great deal, the capacitance value hardly changes at all.

Step 2

Next, the real source impedance of the transmitter, 50 ohms, must be matched to the real part of the parallel load impedance. For this case we have determined by series to parallel transformation that

To keep our notation understandable, please note that for this section, RP is substituted for RL.

Since RP is greater than RS, we must use the correct high-pass equations for unloaded Q given in Figure 1(c).

Substituting, we find that

We have the values of capacitance and inductance that will match the pure 50 ohm source impedance to a 20158 ohm load resistance.

Step 3

Now it is time to resonate the parallel load capacitance that we calculated with a parallel inductor that will be added to our matching circuit. The value of this inductor will have a value that is similar to the one that we computed for the low-pass topology. We use the same equation for resonance as before

Figure 4 shows the model for the high-pass topology. A parallel inductor is introduced at the output of the L-matching network to resonate out the parallel capacitor in the load.

We don’t have to do it, but for practice, let’s go through the steps to combine the matching network inductor in parallel with the resonating inductor

Figure 4. High-Pass L-Matching Network Topology. (Top) L-matching network with a separate resonating inductor as marked. (Bottom) L-matching network, where the L-matching network and resonating inductor have been combined in parallel to produce a single 17.16 μH inductance.

Once the simulation is run, Figure 5 shows that the high circuit Q results in a narrow 2:1 bandwidth, ~ 48 kHz. The 20158 ohm resistor is responsible for this.

Figure 5. High-Pass L-Matching Network Topology Return Loss. The return loss at 3.6 MHz is better than 50 dB demonstrating a VSWR of better than 1.01:1 over a narrow 48 kHz 2:1 VSWR bandwidth.

We conclude that the conversion to the parallel load configuration has resulted in an unloaded circuit Q that is high. This results in a narrower 2:1 bandwidth. The calculation is now repeated for the original series load to which will be added a series resonating inductor. Let’s see if the bandwidth can be improved.

Example 3: High-Pass Topology with Series RC Load

We return to the original series RC load impedance and choose the high-pass topology for the L-network.

Step 1

The high-pass solution for the series RC load begins with the following assumptions

The solution will ignore the imaginary reactance of the load impedance, initially. It will be used later.

The defining equations from Figure 1(d) for a high-pass L-matching network where the load resistance, RL, is smaller than the source resistance, RS, and where Q is the unloaded Q are given by

and

where

Substituting, we have

Also,

At 3.6 MHz, the matching network inductance is

and the matching network capacitance is

The value of unloaded Q is low, and the matching capacitance is high.

Step 2

The capacitive reactance of the load impedance has not yet been canceled as it was ignored in Step 1. A way must be found to cancel this reactance, but we observe in Figure 6 that the high-pass topology separates the load capacitance from the L-matching network shunt inductance, which is at the input.

Let’s try adding a series resonating inductance at the output of the L-matching network. That should work even though this inductance may not be combined with the L-matching network inductor.

The series load capacitance is resonated with a series inductor that will be added to our matching circuit. The value of this inductor is given by

The check for this topology is provided by the circuit model of Figure 6. A resonating inductance has been added in series with the L-matching network. It is not convenient to combine this series inductance with the parallel inductance at the input.

Figure 6. High-Pass L-Matching Network with Series Resonating Inductance Circuit Model. A resonating inductance is added in series with the L-matching network. It may not be combined with the parallel inductor at the input.

The results of this topology are shown in Figure 7. The return loss for this circuit model at 3.6 MHz is better than 49 dB for a VSWR or 1.01:1. The 2:1 VSWR is 78 kHz just as it was for the low-pass solution.

Figure 7. High-Pass L-Matching Network with Series Resonating Inductance Return Loss. The proof of this topology is evident from the plot. The return loss is better than 49 dB for a VSWR of better than 1.01:1 at 3.6 MHz. The 2:1 bandwidth is ~ 78 kHz which is the same as it was for the low-pass configuration.

The series RC load configuration is preferable over the parallel RC load configuration.

It may be concluded that this two-step matching technique for a high-pass matching network works quite well for the 80m band.

Conclusions

In Part I of this article we have used simple L-matching networks with additional resonating elements to match complex loads that possess capacitive reactance. This technique works for antennas that are too short. For antennas that are too long, the antenna load will also be complex but the resonating element required is a capacitor that cancels the inductive reactance of the complex load.

In Part II we will explore a different technique for matching to the complex load presented by a 43-foot backyard vertical antenna. The matching network will consist of a 1:1 UNUN followed by an autotransformer.

In Part III we will discuss the high voltages encountered in highly reactive loads. This occurs when antennas are far too long, or far too short.

References

[1] Salas, Phil, 160 and 80 Meter Matching Network for Your 43 foot Vertical — Part 2, QST, January 2010, pp. 34-37. http://www.arrl.org/files/file/QST%2520Binaries/QS0110Salas.pdf

[2] Ibid.

[3] Blustine, Martin, Highly Efficient L-Matching Networks for End-Fed Half-Wave Antennas, June 11, 2022. https://www.n1fd.org Add Contact Form /2022/06/11/l-matching-networks/

[4] Reproduced under CC BY-NC by permission from Michael Steer, North Carolina State University, LibreTexts™, 6.4: The L-Matching Network, https://eng.libretexts.org/Bookshelves/Electrical_Engineering/Electronics/Microwave_and_RF_Design_III_-_Networks_(Steer)/06%3A_Chapter_6/6.4%3A_The_L_Matching_Network

[5] RFSim99, Stewart Hyde, author, 1999. https://www.ad5gg.com/2017/04/06/free-rf-simulation-software/

 

Antennas, Featured, General

Temporary 20m EFHW Vertical Installation

I’ve been off the air since moving back to NH in 2020. Since the landscaping has not been completed on the property, it has been impossible to install the radials for a  6-BTV vertical. Radials don’t fare well under the treads of a Bobcat.

A collection of single-band, matched, end-fed half-wave (EFHW) antennas was constructed while I was living in FL. All of these antennas underwent testing on an antenna range consisting of three 10m tall masts spaced 70 ft apart.  These antennas were matched with L-networks. The test results were reported in a separate article[1].

Seeing that July 4th weekend was approaching, I was eager to get on the air for a few days before the landscaper arrived. I decided on a 20m EFHW vertical that makes use of some of the guy ropes that were prepared for FL antenna testing. Figure 1 shows the installation of the 12.5m high telescoping fiberglass mast.  The mast is anchored with a tilt-over base mounting plate described in a separate article[2]. Guying is provided at two levels. The guying radius is 25 ft. Guy anchoring is accomplished with polycarbonate Orange Screws[3]. While these anchors work well in FL sand, they do not work quite as well in rocky New England soil. I managed to snap one of them off in the process of screwing it into the ground.

My favorite knot for adjusting the guy rope tension is the taut-line hitch. I used the taut-line hitch on the FL antenna range for three weeks, and the anchor screws came loose before any of the taut-line hitches did.

20m end-fed half-wave (EFHW) vertical

Figure 1. 20m L-Network Matched EFHW Vertical. The wire antenna and matching network is fastened to the fiberglass mast with rubber bongo ties. The mast height is 12.5m (41 ft). Base anchoring is accomplished with a hinged, tilt-over base mounting plate that was described in another article. Please click on the photo to enlarge it.

The antenna counterpoise consists of a 3 ft (~ 1m) section of outer coax shield, Figure 2. A line choke is inserted after this 3 ft section of coax to terminate the counterpoise. The remainder of coax to the shack is made up of a 40 ft long section of RG-8X.

Figure 2. Matching Network, Coaxial Shield Counterpoise and Line Choke. The matching network was designed for 14.1 MHz. Since the matching network has a wide bandwidth, the antenna wire was cut slightly longer to resonate at the very bottom of the CW band. Please click on the photo to enlarge it.

A Smith Chart is plotted in Figure 3. It shows that the antenna match over the entire band is well within the 2:1 VSWR circle.

Figure 3. Smith Chart for 20m L-Matched EFHW Antenna. A match better than 2:1 match is achieved over the entire 20m band. The antenna wire was cut longer to provide the best match at 14.025 MHz. Please click on the photo to enlarge it.

The VSWR performance is plotted in Figure 4. The matching network consists of a lowpass L-network consisting of a series inductor followed by a shunt coaxial capacitor. The antenna wire has been cut to resonate at 14.025 MHz since I enjoy operating in the bottom 50 kHz of the 20m CW band. It’s not that the VSWR performance was that bad but I just could not understand why the antenna wasn’t achieving a near-perfect 1:1 match. I turns out that the residual mismatch is in the Polyphaser lightning arrestor located in the service entrance panel.

Figure 4. 20m VSWR Plot. The L-matching network exhibits wide bandwidth and good efficiency. The antenna wire is cut to resonate at the very bottom of the CW band where I like to operate. The match is very good but not perfect. This was due to the residual VSWR in the lightning arrestor located in the service entrance panel. Please click on the photo to enlarge it.

I operated a simple station consisting of an ICOM 718 at 100W to make three consecutive CW contacts with French stations. The next three days should produce some interesting DX.

References

[1] Blustine, Martin, Highly Efficient L-Matching Networks for End-Fed Half-Wave Antennas, June 11, 2022. https://www.n1fd.org/2022/06/11/l-matching-networks/

[2] Blustine, Martin, Tilt-Over Bases for Antenna Masts That You Can Build, June 30, 2022. https://www.n1fd.org/2022/06/30/tilt-over-bases/

[3] https://www.orangescrew.com/

Radio Amateurs Developing Skills Worldwide