Dec 30, 2009

Upgrade the current VHF station IV; MDS measurement EXPLANATION

General remarks
All rigs have been tested at 14 and 28.4Mhz and other adjacent frequencies to rule out bandfilter anomalies.
Pre-amps,RF gain, AGC and other possible settings which would affect the results have been optimised every time. During each measurement a verification was done with the reference K2 to rule out potential errors as much as possible. Many measurements, especially the hearing tests, were done by two persons.
The found ENB (Effective Noise Bandwidth, column H) allows for comparing apples to apples when talking filter bandwidth.
The filter curves were taken from the audio output using Spectrogram. The Elecraft Wide Band Noise Generator module was used to create a constant signal of about S2-S4.


Performance evaluation

Both column N and R are key in the evaluation of the measured results and hearing results on 144MHz.
Note:the measurement in column Q is totally different from the rest, where both the RF generator and transverter are using in- and outdoor-antennas:

As such there is a true radio spectrum in between as media. A real life performance test so to speak. Column Q represents the set RF output of the generator at which the MDS recognition threshold by human ear was found.
As mentioned before; I'm testing 28Mhz weak signal performance on a quiet band using a transverter, that is totally different from lowband and/or contest use

What really matters is about making that extra QSO yes/no.


Kenwood TS570SG

Frankly this rig was only tested during the preliminary test 'Method 1'; it is currently in use as an NCDXF HF beacon tracking receiver. The 570SG has the Inrad 400Hz CW filter and a temperature stabiliser for the LO crystal. Its internal DSP is an early 16 bit version; OK at 200Hz and above, but heavy ringing below.

28Mhz sensitivity is, next to the icom 7400, tops. However, for serious weak signal reception connected to a VHF transverter, this rig (and another 570DG) showed its weaknesses; the sound is rough, noisy and it just does not bring out the real weak ones. Is this due to its published high phase noise? Can't tell cause I have nothing to prove that. The

ICOM 746
This rig also has a 400Hz CW filter from Inrad. It's owner installed Individual Inrad filters for both IF stages, but previous experiences showed there is too much attenuation caused by bad internal gain distribution. Only the 400Hz in the 2nd IF was used as well as the original 2400Hz in the 3rd 455KHz IF stage. The additional AF CW peaking filters do work very well, but so does the free extra AF background noise. On 144 MHz its own internal VHF front end was used. In total it lacked some 5 dB in performance. When owning such a rig I would use an external preamp and bypass the internal.

ICOM 7400
The '746 Pro version' showed the best sensitivity on 28Mhz. Using its internal VHF front end, it does hold its own listening to carriers down in the noise (test method 1). But when trying to copy real CW signals the DSP offers a very rough sound and is ringing already from 200Hz and lower. I tried getting a better S/N result by adjusting its internal IF gain setting (input to the DSP unit), but apart from the gain difference in AF volume it did not offer any better result in the measurements, nor during the hearing results.


DSP filter curves at 300Hz and 200Hz settings
Please notice there 'brickstone' sharp shapes.
I have had another 7400 several years ago and used it intensively on 144MHz SSB and CW. The sound was very harsh and annoying over time. Then came the first K2 and a DEM 144/28 transverter; what a sonic relief!
Next, from the AB4OJ's Icom website I tried the following trick: choose a filter setting >500Hz and then reduce the bandwidth by offsetting the twin PBT controls. The final result is a 50Hz filter with a very bad shape factor, totally failing in selectivity but the sound is next to noisy much more mellow.

In practise this 50Hz 'low quality filter' offered the best results for the IC7400. Only feasible on a very quiet band. Maybe this will help me on my 50MHz set-up next season.


Microtelecom Perseus
This is not your typical rig where you turn the VFO knob and quickly use some dials & knobs to make a QSO. It is more a tool to monitor single frequencies or a whole frequency band and make that visual. But it does this very well. This high potential receiver has many capabilities (visual and audio) and has phase noise at such a low level(-140dBc @ 2KHz, -150dBc@10KHz), only seen at scientific equipment.
It offered the lowest measured result at 28MHz sensitivity. Due to its internal latency, a true S/N measurement using the Marconi analyser was not possible.
Interesting fact is that despite lacking 6dB at the bench test(column P), it does perform well during a real life hearing test, using an outdoor antenna (column Q). Might this be related to its extremely low phase noise?

Perseus at 416Hz and 202Hz settings


From this 141Hz setting and lower the ringing becomes worse; I mean ok for monitor carriers but incapable of decoding weak signal CW transmissions.


Elecraft K3 serial:173
We're in the upper class of this test now and the left differences are only minor. The K3 is third best at the bench test in column N. The final column R shows no result as the K3 was not available at that time. An 8-pole 400Hz roofing filter was installed. Interestingly, using the 250Hz roofer and tightening the DSP filter down from 250 to 50Hz the K3 showed a reverse result (worse S/N ratio) on the Marconi. Several actions have been tried: internal IF gain setting, AGC and finally switching from FIR to IIR at 100Hz and 50HZ. The final one was the only setting to show another single dB improvement, but the ringing is awful and makes a weak signal QSO impossible.
The K3 starts with slight ringing from 150Hz and down. From there it becomes worse whatever the settings.


K3 400Hz and 200Hz FIR filters


K3 100Hz FIR and 100Hz IIR, please note that the IIR filter actually is wider.
This was also noted at the chapter FIR versus IIR filters at Clifton Laboratories.


50Hz FIR and 50Hz IIR; being a better 'peaking filter' the latter one did show minor improved measurement results.
Officially, any result difference within 3 dB should be considered "measurement deviations". But the differences are clearly there every day and remain when doing the test at a different QTH. When switching from the Marconi analyser to the hearing test, these last few dB's are still very apparent. Even more the perceived sound quality. This is where the K3 differs from the last 3 below; it sounds to harsh for weak signal. I have put out some questions on mailing lists and contacted different K3 owners; no special setting(s) for weak signal listening on VHF and up were found. Both the original K3 owner and another 6M enthusiast reported the same perceived 'harshness' compared to what they are used to at analogue rigs.
I'm not satisfied yet and convinced that we were not able, to let the K3 show its full merits here. Based on its published specs and current results at major HF contest stations,
I trust the K3 will be a killer during 6m/VHF contests. Elecraft is continuously offering substantial firmware upgrades and the K3 shows increasing potential. I still fancy one..

Elecraft K2 serial:3323 original
This is a B version. It is a K2 in original shape and aligned according the written procedure. The internal crystal filters have been set-up for 1000, 700, 400 and 200Hz bandwidth.
It also has the additional KAF2 audio peaking filter (270Hz and 100Hz). This reduces the out of band noise:

The second picture shows the wideband performance curve using both the 200Hz crystal filter as well as the AF2 100Hz audio filter switched in. Unfortunately this AF2 100Hz filter introduces slight audible ringing.

The measured reduced sensitivity at 28MHz compared to 14Mhz is a normal behaviour.
The K2 does have a relatively low phase noise but at 7Mhz. But like most contenders, not particular that good at higher frequencies like 28Mhz. Nevertheless its performance is superb; hey we're in analogue heaven now! The noise-free AF section of the K2 makes the volume control even act like an additional RF gain.
The sound is sweet and offers plenty of details in the quiet background.


Ten-Tec ORION
I have often used this rig for performance comparisons during the last 4 years and it has always been a close draw with the finalist below. The ORION had been modified by exchanging the 1000Hz crystal (roofing)filter for an Inrad 4-pole 600Hz. This special filter #762 was initiated on request of Bill Tippet, W4ZV
as the original 500 and 300Hz filter upgrades created a worse IMD spec due to a distributed gain issue. This was solved in the later ORION II model.
During the sensitivity measurements it became very apparent that any RF gain setting above 92 was creating extra noise and upsetting the S/N measurement. I believe this is the reason why so many new users confirm so called noisy receiver behaviour. Setting the RF gain to an appropriate level is outside the general consumer perception. The manual does explain proper use of the RF gain, but for weak signal you are used to "dialing it up to the max". Thus I got results which are several dB's better than published elsewhere. But the verdict is in the hearing tests and the perceived audio performance. The ORION does very well in this area, being the best DSP rig in the test.


The above 200Hz and 100Hz DSP filter settings are 227Hz and 155Hz in reality.
Both filter curves represent a more analogue shape.
Not the slightest ringing noted. Even the 100Hz DSP filter sounds as clean as possible. Perhaps this is due to different chosen DSP filter algorithms by Ten-Tec designers. It does offer the most analogue sound and shows great receiving performance; signals could still be detected deep down in its calm noise floor. What a relief for your ears.
Do not forget the PLL design offers an extremely low close-in phase noise, not matched by any other but the Perseus.
So far the ORION has offered me the best overall results in 22 years of Hamradio.


Elecraft K2 serial:2036 modified
In terms of options it is similar to the other K2. Several small upgrades have been done for boosting general performance. However the following modifications have been done over the years for optimisation on 28Mhz transverter use:

1. Pre-amp and AGC
The original preamp is a compromise in terms of system noise figure vs IMD performance. Changing its gain boosts its total system noise figure. More preamp gain involves a lower intercept point however.
See the Elecraft application note


2. 28MHz Band Filter
This is now peaked at 28.2 MHz


3. AF filter

The AF2 filter is changed from 100 to 140Hz. This small change reduces ringing completely.
The audio gain was slightly optimised by changing R9 to 12K.
Spectran was used to do the adjustment on the KAF2 board, which now peaked a few dB higher.

400Hz crystal +270 Hz audio filter. And 200Hz crystal +140Hz audio filter.

This modified K2 has proven its superiority during these tests. It also functioned as the reference rig during the tests. Switching back from any set-up to this little analogue wonder always immediately showed its advantage.

The verdict
Best reception: modified K2
Best overall performer: ORION
Best alternative 'second receiver': Perseus (visual performance)
Great future potential: K3

Analogue beats digital, but only just.
Razor sharp DSP filters offer bad audio for VHF weak signal CW message detection.
K2 and ORION offer sonic details like you are diving into the abyss of the RF noise floor.
What is the true 'supporting performance' of low phase noise?


Mission completed?
I wanted to measure the noise figure of my K2 and, in the process decided to do some more testing..
Got a bit carried away.... But then again, I never got to measure known VHF performers like the TS850, FT1000MP etc.. And then there's that superb Javornik transverter. And Down East Microwave is (re-)designing a new transverter.
Oh well... lot's of promises for the future ;)


Wish you all a Happy New Year !!!

73 Mark, PA5MW


Dec 24, 2009

Upgrade the current VHF station III; MDS measurement

This follows the previous blogs on optimizing my Elecraft XV144 144/28MHz transverter + 28MHz IF transceiver

Purpose of MDS measurement

Find factual differences in 28MHz sensitivity of HF transceivers, using measuring equipment as well as ones own ears. Create repeatable and, as much as possible, reliable results.

Final decision criterion: Can I make that extra QSO Y/N ?

Prerequisites
Since my goal is to evaluate my local possible transverter+HF combo's, the actual measurements must reflect real life situations. There is plenty of good data available from ARRL, Sherwood, G3SJX etc.. But I need factual data on 28MHz. Every rig will be set-up for transverter use. If that is via other antenna entries, using extra relay routing, different internal circuitry etc... so be it. Every rig will be fine tuned for best performance on weak signal detection; being able to copy and read CW transmissions. That rules out filter settings which create ringing.

Getting a feel for measuring MDS

I have limited experience in doing this, I follow a different protocol and my equipment calibrations are outdated (2003). Below data is not scientific proof whatsoever. Nevertheless I have done my best to do make sure to deliver repeatable results. I tried a few different methods of determining MDS, just to get a feel and, more important, see how it relates to reality.

Test set-up



Headphones: David Clark Model 10/DC Stereo, with passive noise cancelling.
Picture shows a temporally test setup at a friends place.
The 'wires in the garden' is a K9AY low band receiving antenna.

Determining the MDS threshold reference level; method 1
This was determined by human ear recognition only:
Output RF generator: continuous carrier signal fixed set at -100dBm on 28.4 MHz and 144.4 MHz. It is capable of reducing its output down to -140dBm, but to rule out internal deviations and external possible load differences I used the step attenuator. A second identical step attenuator from a different brand was used from time to time for comparison and verification.
Generator connected via step attenuator to the 'Device Under Test', or to the Elecraft XV144 transverter, which in turn was connected to the DUT.
Output receiver: connected to headphones
Using the 10dB and 1dB switches of the step attenuator, a minimum threshold was found at which the RF signal can just be recognized and while turning the VFO the varying beat note can still be heard. This method delivers a more accurate and detectable threshold vs just a faint phantom signal xx dB buried in the noise.



The penultimate column shows the single receiver performance at 28.4MHz, the final column refers to the transverter combo result. Found 28MHz MDS values showed an average delta of 8dB compared to official published data which is 'promising'.


Determining the MDS threshold reference level; method 2
Taking it one step further I started doing the S/N measurement function on the Marconi communication analyser.


The second drawing shows the set-up with the 144/28 transverter 'combo'.
In both set-ups the audio is routed back to the analyser which determines the S/N figure.
At first I have done all tests using a 20dB S/N level as measuring reference, but since that is not resembling a true weak signal I cut it back to 10dB S/N. At that level all signals are already very near the noise level.
This test method quickly showed reliable results. I have repeated all measurements a minimum of 5 times at different days. To be sure I got the same repeatable results,
each time the modded K2 was used as measuring reference. I spent 3 weeks doing this almost every evening. There was one occasion where all test results shifted 1dB on a single day only. But apart from that everything stayed rock solid. I got enthusiastic and started dragging in other rigs as well.




So what do we have here?
Column D: extras which do not come standard with the product
Column E: selected crystal filter in the 1st IF being the standard or an optional filter
Column F: set DSP filter. IIR and PBT refer to receiver custom settings
Column G: set audio 'peaking' filter, only applicable for K2 and the IC746
Column H: measured Effective Noise Bandwidth using info from Owen at: http://vk1od.net/measurement/enb/MeasureIfBw.htm
Column I: calculated using the metric converter at http://vk1od.net/calc/RxSensitivityCalc.htm
Column J: published MDS results from ARRL at 14 MHz (for comparison only)
Column K: measured S/N using the Marconi generator/analyser at the 10dB S/N level reference
Column L: same
Column M: same but now using the 144/28 transverter combo
Column N: calculated delta on results found in column M
Column O: same as K but using human ear for MDS recognition
Column P: same as O but using the 144/28 transverter combo
Column Q: same as column P but now the RF generator is connected to a small antenna loop and the transverter is connected to the outdoors 144 MHz yagi. The results shows the delta to the found "best in class" receiver.
Column R: remarks will be discussed in detail in the next blog.

"I'm not happy with your results....."
What does this all mean? Why is X doing this and Y doing that? My Z-rig at home does better than your measured yours etc.......
Please note the whole measurement concentrates on reading the signal at 28MHz under weak signal circumstances.
That is a zillion light years away from performance during a contest, let alone on low bands.
The hearing measurements were done with two persons and done as serious as possible. It even turned out my friend was able to dig another dB or two in the noise but he created an extremely small brain filter which was already pre-synced on the signal :). Fun but not the agreed threshold level where we could both detect and read the CW transmission.

So ????????
Benchmark results are nice for comparison and I'm pleased with the found noise figure results.
This will allow me to do careful calculations using the mentioned VK3UM application, see my earlier blog.

I appreciate the MDS test results 'by human ear' the most.
In the end the QSO is made using that exact instrument, so by upgrading my whole 144Mhz set-up I want to use that 'measurement tool' as much as possible. I'm very reluctant it turned out to be a reliable tool.
The S/N test function on the Marconi generator produces a modulated signal which sounds like separated dots (e-e-e-e-e) at some 15WPM. That is the close enough to reality and certainly much better than any stable continuous carrier.

During the tests there were large differences in sound quality perceived. DSP filters sometimes sounded harsh and extremely small filters introduced ringing. Ringing does not need to be a problem if there is plenty of signal, or you need to filter out the adjacent station during contest etc. However at the weak signal level on a quiet band ringing is not making that extra QSO.

Conclusion
The modded K2 sounded best and proved the winner for decoding weak signals, followed closely by the Orion and K3. The Perseus not only shows a great picture but offers good readability too.

What's next?
In the next blog I will comment on all tested receivers individually and explain in detail how they were set up and modified for best results. Their measured filter responses will be shown as well.


Dec 2, 2009

Upgrade current VHF station II; the plan

What is my goal?
Pure and simple: have more fun in hearing (seeing?) weak signals, participate in a few contests and/or work new dx, on 144MHz.

Throw in some new hardware !!??
Now there's a lot you can do to upgrade your VHF station boosting the performance in theory, like:

- adding a pre-amp
- additional (band-)filtering
- mount the transverter near the antenna
- install extreme low-loss coax
- raising the antenna height
- buy new high performance equipment with DSP, add SDR receiver etc.
- low noise power supply

However, none of these will offer gain by principle.
All the above need to be seriously evaluated before implementation.


Hardware performance simulation
I do like VK3UM's EME system performance calculator a lot.
This great application allows you to evaluate your set-up and simulate any future upgrades.
See free software at: http://www.vk3bez.org/vk3um_software.htm

A typical view shot looks like this:



Mind you: there is RX Noise Figure which is determined by hardware only. And there is System Noise Temp and Noise Figure, which show the real life situation when terrestrial or sky noise is added to the equasion.

The hardware setup is clearly cut into pieces:
Antenna
Coax to the 1st RF stage (LNA)
Coax the the second RF stage (receiver)
Settings for coax type, connector and relay contact losses.

The program calculates both a total receiver noise figure based on pure hardware only, and a total system noise figure/noise temp, based on the environmental terrestrial sky noise (not EME).
Toggling the LNA on/off switch the total system noise figure gains 3dB at receiver performance.

The plan
Let's start evaluating the current total receiving performance.
There's only 2 building blocks in VK3UM's performance calculator. So I need to cut my 144MHz receiver chain in separate parts and evaluate the individual blocks first, before starting to calculate the total system performance.

my current blocks:

1. Antenna (Tonna 17el @12m AGL)
2. Coax into shack (2 pieces of EcoFlex-10 and -15 +5 connectors)
3. 144/28MHz transverter (Elecraft XV144)
4. 28MHz IF transceiver (Elecraft K2, modified for bettter 28MHz performance)


First step: evaluate transverter +IF receiver performance

I will need to do a MDS test on the K2 IF receiver