Thinking aloud on the VK3XDK 10GHz transverter receiver performance

*** DRAFT ***

10/03/2013 Correction: the article was based on incorrect statement of specifications for the FT-817ND published in the sales brochure by Yaesu. The service manual gives a much better sensitivity and more believable in terms of the FM sensitivity and the article is revised to use the service manual sensitivity... my apologies.

In his web page describing his 10GHz transverter, VK3XDK gives the following information.

The 10.368 GHz RX section features a NE32584 HJfet on the input, biased at 10mA and about 2V drain-source. The FET uses microstrip matching on the input and output for a gain of around 10-14db beam width and Noise figure of less than 1db beam width.

The second stage uses a Minicircuits MMIC era3 with 3.5V (35mA) fed to its output. These MMICS feature reasonable gain at 10368 GHz beam width (around 10db?) and a fairly low Noise figure.

Following the second amplification stage is a microstrip (edge coupled) filter with a band pass of less than 400 MHz (loses of less than 4db) and a Hittite hmc220MS8 Double balanced mixer. This particular mixer likes an LO drive of over 7dbm for low conversion loses (10-13dbm Mixer). [sic]

This seems to describe a transverter with four key elements:

• NE32584 LNA with gain around 13dB and device Noise Figure (NF) around 0.4dB;
• ERA-3 amplifier operated well above its datasheet range, claimed gain of 10dB and unlikely to have NF better than the datasheet's 3dB at 3GHz.
• a BPF with loss of 4dB;
• a DBM which can be initially considered approximately noiseless apart from its conversion loss.

Table 1 shows a workup of expected gain and noise contribution (referred to the input) of each of the four elements of the transverter, and the total equivalent block has a gain of 11dB and NF of 0.87dB.

The designer has stated on VKLOGGER [o]verall (conversion) gain on the receive section has been found to be over 10db (Result will vary board to board).  [sic] This observation reconciles well with the estimate in Table 1.

ZL2BKC reported on VKLOGGER: [l]ooking back at my notes the measured NF of mine is 1.6dB and conversion gain 12.8dB, and just over 2dB with the relay and feed coaxes. There is a little uncertainty as to whether the NF is for the transverter or system, again fairly similar to that in Table 1.

Table 2 shows a workup of expected system performance using the transverter with two popular IF transceivers, cabling loss is taken to be zero.

On VKLOGGER forums VK3XDK comments on (presumably) sensitivity:

...the boards (without preamp) should be good to at least -110dbm! typically -120 to-130dbm where it becomes difficult to tell due to sig-gen leakage. [sic]

Though this statement is rather loose, if these figures are taken to be sensitivity for 10dB S/N in 2kHz noise bandwidth, they can be translated to system NF and Teq.

Table 3 shows the equivalent system NF and Teq for the three sensitivity levels mentioned by VK3XDK. In an application where the external noise is likely to be in the range of 20-300K, Teq should be less than the external noise for reasonable exploitation, so less than perhaps 200K for terrestrial work, and perhaps 30K for EME work. Clearly a sensitivity figure of -120dBm for 10dB S/N in 2kHz noise bandwidth will warrant an LNA, even for terrestrial weak signal work.

This study suggests that the bare transverter gain and NF requires a low NF IF for full exploitation, more so if there is significant loss in the IF cabling.

Assessment of system performance

Some tests that will commonly be applied by hams with little test equipment are:

• Sun/ColdSky Y factor test;
• HotEarth/ColdSky Y factor test; and
• half power beam width.

Sun/ColdSky Y factor test

In this test, the ratio (Y) of linear receiver system output power due to quiet Sun and cold sky is found. The receive system performance indicator G/T can be calculated from Y and knowledge of the current solar flux at the test frequency.

Considering the example configuration:

• frequency 10,368MHz;
• antenna gain 35dB;
• solar flux density 340SFU (Sag Hill 01/03/2012);
• estimated spillover noise of 29K;
• estimated adjacent cold sky of 6K; and
• receive system NF 2dB at the antenna flange.

The effective aperture of the antenna can be calculated to be 0.21040m², and the captured solar noise to be 340*1e-22*0.241040=8.195e-21W.

This solar noise power can be converted to an equivalent noise temperature by dividing by Boltzman's constant (1.38e-23) to give 593K.

The receiver NF of 2dB can be converted to an equivalent noise temperature of 170K.

Table 4 shows the components of noise and totals for the hot and cold measurements. The Y factor is the ratio Hot/Cold=798/205 which can be expressed as 5.9dB. If in this case Y was actually measured at 3dB due to lower antenna gain alone, it would imply antenna gain 4.6dB lower than expectation. Small shortfall in Y may be caused by a larger shortfall in antenna gain.

People often report lower Y figures for similar configurations. Possible explanations include solar flux variation, non-linear receiver operation, inaccurate noise power measurement, low antenna gain, high spillover noise, high receiver NF.

Whilst this test can, if properly performed and based on valid data, give a good confirmation that system performance is as expected, it does not provide much information to localise contributions to poor performance.

HotEarth/ColdSky Y factor test

In this test, the ratio (Y) of linear receiver system output power due to hot earth and cold sky is found.

The test outcome is independent of antenna gain.

Considering the example configuration:

• frequency 10,368MHz;
• estimated spillover noise of 29K;
• estimated cold sky of 6K; and
• receive system NF 2dB at the antenna flange.

The receiver NF of 2dB can be converted to an equivalent noise temperature of 170K.

Table 5 shows the components of noise and totals for the hot and cold measurements. The Y factor is the ratio Hot/Cold=461/205 which can be expressed as 3.5dB.

People report lower Y figures for similar configurations. Possible explanations include non-linear receiver operation, inaccurate noise power measurement, high spillover noise, high receiver NF.

There are two significant issues with this test, it is independent of antenna gain so no conclusions can be drawn on antenna gain, and spillover noise is a large component of total noise in the hot and cold case at some frequencies (as it is here) and the test outcome may be very sensitive to accuracy of the spillover estimate.

Whilst this test can, if properly performed and based on valid data, give a limited indication that system performance is as expected, it does not provide much information to localise contributions to poor performance.

Half power beam width

Half power beam width for a efficient high gain antenna is tightly related to antenna gain. Careful measurement of the half power beam width can provide a good confirmation of antenna gain. Transit of the Sun through a fixed antenna can be used as a signal source, but reduction of the measurement data to a pattern is not trivial, see Plotting antenna pattern from Sun noise.

This test is not a good test for exploring spillover noise issues, but it can provide reassurance that the basic geometry of the antenna results in the expected main beam and would expose errors in location of feed unit, large errors in dish surface, gross over / under illumination.

Bottom line is that if the lobe is wider than expected, gain is lower than expected. There are ways to artificially force a narrow beam, so a beam narrower than expected beam does not necessarily imply higher gain.

Test summary

It is possible to predict the expected outcome of the tests with good knowledge of the system under test (as done in the examples above), but this seems rarely done, certainly rarely written up online. Combine this with a dearth of useful experimental data and most tests are conducted absent any real performance benchmark applicable to the system under test.

The tests are often applied poorly and the results are often invalid, for example:

•  an online poster described a Sun noise test at 23cm based on 70SFU which turned out to be the 10cm solar flux, not the flux at 23cm;
• a correspondent was observing very small Y factor on a Sun noise test due to AGC in the IF transceiver adjusting gain for constant audio power out... the receiver did not qualify as a "linear receiver"; and
• another correspondent faced with the same problem had modified the transceiver to disable AGC, but that exposed the transceiver to detector overload and non-linearity was proven not much above the AGC onset threshold.

Half power beam width measurement can provide a fairly good confirmation of antenna main lobe gain.

A most useful system parameter is the receiver system NF or Teq, and measurement (outside the scope of this article) is worth pursuing. With that knowledge, the ColdSky/HotEarth test can provide an estimate of spillover noise, and the Sun/ColdSky can provide an estimate of antenna gain.

Conclusions

Building a kit instead of purchasing a "built and tested" article is a great way to learn and a source of great satisfaction (and perhaps some frustration).

The key thing is that building the kit is only half of the "built and tested" article, completing the job means testing that the built kit meets specifications (sometimes absent though). It is the testing phase that is the real learning opportunity!

Nothing in this article is to imply that the subject kit is good or bad for any particular purpose or in concert with any particular IF transceiver. Indeed the outcome will be to some extent dependent on the skills and knowledge of the individual constructor and their operating environment.

Links

• Quiet sun radio flux interpolations
• Antenna Effective Aperture
• Receiver sensitivity / noise figure / noise factor / noise floor converter
• Plotting antenna pattern from Sun noise
• NFM Home Page

Changes

© Copyright: Owen Duffy 1995, 2021. All rights reserved. Disclaimer.