Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

Revision History
08 November 2009. Original
09 November 2009. Corrected typo in reference to frequency; added Elecraft response to data; added clarifying note re KAT3 tuner in bypass mode

Table of Contents
Introduction
Termination_Characteristics
Coupler_Return_Loss
K3_Data_with_50_Ohm_Termination
K3_With_25_Ohm_Termination
Analysis_and_Comments

Introduction

A report recently was posted on the Elecraft reflector noting anomalous SWR values displayed on a K3 transceiver when a Telepostinc LP-100 vector wattmeter sensor was installed on the K3's output connector. 

In order to independently assess this report, I ran a series of tests with my K3 and LP-100 wattmeter:

1. K3 reported SWR with a high accuracy 50 ohm termination directly connected to the K3's output connector.
2. K3 reported SWR with an LP-100 sensor connected between the 50 ohm termination and the K3's output connector
3. K3 reported SWR with a high accuracy 25 ohm termination connected directly to the K3's ouptut connector
4. K3 reported SWR with an LP-100 sensor connected between the 25 ohm termination and the K3's output connector
5. LP-100 reported SWR under the test conditions described in (4).

In all cases, the connections were made without coaxial cables; direct connections were made using only between-series adapters as may be necessary. My K3 was purchased "factory assembled" and, in any event, I do not believe there is a user-adjustable SWR calibration.

My K3 is equipped with the optional KAT3 automatic antenna tuner, which was set to BYPASS for the tests discussed on this page.

I wrote a computer program in EZGPIB to control the K3 and also read the LP-100's output. For each of the 11 frequency bands between 1.8 MHz and 50 MHz, the program commanded the K3 to transmit with 15 power steps between 1 and 100 watts. The program read the K3's reported SWR and, for those tests where the LP-100 was connected, the LP-100's data output including SWR. Before taking data, I updated the LP-100 to the most recent firmware release and recalibrated it.

The K3 was powered by a HP 6652A digitally controllable power supply, set for 13.8V with remote voltage sensing enabled.

The total collected data sets comprise several thousand data elements, and presenting the information in a compact but informative fashion is not simple. I've decided to show three sets of column graphs, representing 10, 50 and 100 watts. Intermediate power levels were similar to these values.
 

Before collecting data with the K3, I first characterized the terminations and the LP-100 coupler using an HP 8752B vector network analyzer. The "calibration standard" I use with the 8752B is a Mini-Circuits KARN-50-18 type N termination, with a minimum return loss over the frequency range 0-500 MHz of 35 dB. 35 dB return loss corresponds to a VSWR of 1.04. The accuracy of VNA measurements is tied to the accuracy of the calibration standards and hence the KARN-50-18 accuracy should be kept in mind in interpreting the results on this page.

The 50 ohm termination I used for the tests is a model RFT 100 NFE 100 watt termination. The return loss over the frequency range 1.8 - 54 MHz is 41 dB or better. (SWR of 1.02.)

Assuming a change in the K3's reported SWR changes when an LP-100 coupler is connected between the termination and the K3's output port, at least two possible explanations come to mind:

• The coupler represents in "impedance bump" and hence alters the impedance (and, of course, the SWR) seen by the K3's internal reflectometer.
• The coupler's return loss is negligible, but the electrical extension represented by the coupler causes a shift of impedance (where the termination is not 50 ohms) and hence alters the voltage sampled by the K3's coupler. Although, neglecting loss, the SWR is independent of line length, the impedance seen is different and thus diode non-linearity may account for a small shift in displayed SWR.

To assess the size of the "impedance bump" presented by the LP-100 coupler, I measured the reflection coefficient of four wattmeter coupler sections:

1. Bird 43 with type N connectors on the line section
2. Bird 43 with UHF connectors on the line section
3. P3000-D coupler for a VFD digital wattmeter
4. LP-100 coupler

In all cases, the coupler was  terminated with the KARN-50-18 termination used for calibration. Adapters were used as necessary.

The Bird 43 with N connectors has an exceptionally low return loss, 56.7 dB at 54 MHz. In fact, the plot line is off the scale of the plot below. (The scale is the same for all four plots.)

Connecting the 43's line section to the VNA requires a series of adapters. At 1.8 MHz, the return loss difference between the N and UHF versions is negligible, amounting to 2 dB or so out of 62 dB. However, by 54 MHz, the UHF connector version has nearly 9 dB worse return loss. Of course, even the 50 dB return loss of the UHF connector 43 amounts to an SWR of less than 1.01.

The P3000-D coupler is rated at a maximum usable frequency of 29.7 MHz, at which the return loss is 25 dB, or VSWR of 1.12. This is borderline significance, at least in terms of measurements, although of less importance in practical amateur radio operations.

If we set the threshold of acceptability for impedance bumps at 25 dB, as discussed above, the LP-100 coupler is below the limit up to 54 MHz. Up to 28 MHz,  the return loss of 30 dB or better corresponds to an SWR of 1.07, which is of no significance for practical amateur radio operations.

Rather than plotting many essentially identical numbers, I'm presenting the 50 ohm data in three tables.

The K3 reports SWR in 0.1 steps, and within this resolution, there is no difference in reported SWR whether the 50 ohm termination is connected directly to the K3's antenna port or after the LP-100 coupler.
 

The three plots below compare the K3's SWR reading with the 25 ohm load directly connected (red columns) and connected with the LP-100 coupler between the K3 and the termination (green column). In addition, the LP-100's SWR reading is presented (blue column.) I've also added black lines identifying the theoretical SWR and the SWR as measured by the 8752B VNA.

Several points are apparent from the data:

1. With 50 ohm termination, the K3 has no reported difference with and without the LP-100 coupler.
2. With 25 ohm termination, the difference in K3 reported SWR between direct connection and with the LP-100 coupler in line is negligible, amounting to at most the K3's 0.1 resolution step, except at 50 MHz, where the LP-100 coupler causes an increase in reported SWR, which, probably by happenstance, brings the K3's reported value closer to what it should  be.
3. The K3's SWR circuit is reasonably good up to 5.3 MHz, but has a noticeable under-run at 7 MHz and above, although 10.1 MHz is not as bad as 7.1 or 14.1 MHz.
4. The data shows reasonable consistency at all three power levels.
5. The LP-100 is remarkably close to the VNA measured value.

This data is, of course, based upon a single K3 and different K3's may react differently.

The good news is that Elecraft is aware of this problem and is looking at a firmware fix, as summarized in the following Elecraft Reflector post made by Wayne, N6KR:

Great data, Jack. We've been testing this, too, and we know why the  
K3's readings differ.

A firmware change will be made to improve the accuracy in general. But  
part of the difference on the higher bands is due to strays in the  
KAT3 module. These strays can be tuned out by the KAT3 itself, thus  
presenting the correct load to the K3 internally. But an external  
instrument is on the other side of the KAT3 and thus sees a slightly  
different match.

We'll post additional details soon. Meanwhile, no one need worry about  
this apparent discrepancy. The ATU is cancelling its own strays, and  
the SWR bridge is correctly reporting the load presented to the  
transceiver internally, which is the important thing when matching the  
rig to its load.

73,
Wayne
N6KR