The Sherwood table

The comparison of transceivers in the Sherwood table is a repeating discussion on our national repeater PI2NOS in the Netherlands. Soon after the word Sherwood tables is mentioned a sudden moment of silence occurs, probably because ham operators immediately start to look in that table whether or not they can spot their gear and compare it to other receivers. The table is intended for measuring the quality of the receiver. There is no unique way as far as I know on how to do this. From a technical point of view Sherwood engineering in Colorado has chosen for the field dynamic range narrow space. The table can be found here, I found the table via hamnieuws because the latest icom-7300 transceiver just appeared in the table. The next thing you hear on PI2NOS is that some icom-7300 already appeared on which is equivalent to ebay or the craigslist in other countries.Speculation starts, and it continues for a few days until someone mentions the word: yes, but the Sherwood table said. Well they do, but what does Sherwood not tell you.

Did the table make sense to you? In that case congratulations. Did you base your decision on buying a radio on the Sherwood tables? Well, most of us did not. When I brought my radios I looked at price and performance, I wanted to have a base-station that was affordable, and that performed both on HF and VHF/UHF. A few choices remain and you decide whatever the budget allows. I still think this is the best way to make a decision.

Is sensitivity of the receiver a serious selection criterion? Theoretically it is, in reality it is not. Depending on the IPO setting of my FT-991 the sensitivity varies between -123 db to -143 dbm. The theoretical limit is -174 dBm for 1 Hz bandwidth, so -174 dBm/hz which you can obtain with the equation P = k.B.T where P is represented in Watt, k is the Bolzmann constant 1.38064852 × 10-23 m2 kg s-2 K-1  the temperature in Kelvin is T and the bandwidth is B. Take a room temperature, T=273+20=293K and a bandwidth of 1 Hz, the value in dBm/Hz for P is 10*log10(k*T*B/1e-3) and we arrive at the -174 dBm/Hz. All receivers in the table are at least 30 dB larger than the room temperature noise floor because of the way Sherwood measured the noise floor. The crux is that the measurement is done with a 500 Hz CW filter, the theoretical noise floor is therefore P = 10*log10(k*293*500/1e-3) = -147 dBm. Better than this is not possible unless you do something against the noise of the amplifier for instance by cooling it down. We don’t, but radioastronomers do.

My Yeasu FT-991 has a noise floor of -143 dBm which is when AMP2 is selected rather than IPO or AMP1. IPO is oftentimes good enough for the stronger stations, and AMP1 is what you often use, but AMP2 is usually noisy so that you’re better off attenuating the signal. It depends on the frequency band you are working on, I’m talking here about 20m and 40m. If you are really worried about sensitivity then use a low noise amplifier LNA to counteract the cable losses between the antenna and the receiver. Usually we don’t use a LNA unless you’re working with frequencies for instance in the L-band. GPS receivers have for this reason a LNA directly near the patch antenna.

Next to sensitivity there is in any superhetrodyne receiver the noise inserted by the local oscillator. The LO phase noise is mostly determined by the noise of the (thermo-stabilized) quartz crystal and the noise of the phased locked loop of the circuit that generates the signal going into the mixer. It is measured as dBc/Hz, so dB’s relative to the carrier and then as a spectral density.  The measurement of the phase noise is performed over 10 kHz bandwidth if I understand the description of the Sherwood table correctly. I’m not sure how the measurement is done, probably you need to open the radio and measure the LO output directly.

I don’t think that sensitivity and local oscillator noise are anymore an issue there days, sensitivity is sort of a non issue in urban areas where every device on the block generates some QRM that may be picked up, but sure, if we would go to the field and run the receiver on batteries then sensitivity is the factor that counts.

What counts more for radios in general is the ability to separate signals, and for this most receivers have a seemingly infinite amount of filters that apply at the HF pre-amplifier, and the pass band filters that play a role at the intermediate frequencies. This is also the criterion that Sherwood takes for comparing different receivers. In the end they come up with the dynamic range narrow space in dB. If one would take two nearby transmitters, then how do they affect the  noise floor of the receiver? This is certainly an issue in particular on the crowded 40m band in the evening in Europe. And this is where you can try to compare receivers provided that the roofing filters are comparable.

But this is also my critique on the Sherwood table, they do not look into the possibility of digital signal processing, noise blanking or notch filters which is what you often resort to in order to suppress noise of a neighboring transmitter which may cause some splatter. On the other hand, it is true that one receiver may perform better than another based on the filtering capabilities of the radio, and this mostly determines the standing in the Sherwood tables. At this same time I have my doubts whether filter response is the only thing that counts, mostly for two reasons.

Digital signal processing, noise blackers and contour filters on the FT-991 are able to take away a significant part of the noise to retain that part of the spectrum that you want to concentrate on. I use it quite a lot to isolate weak stations, thus DSP removes the noise and the audio suddenly becomes readable. The Sherwood tables don’t mention the word DSP, maybe the idea is that digital signal processing is not like a filter and that this is not what the classical analog receiver is supposed to do.

A second point is, if you operate a SDR or if take the audio spectrum from a standard SSB receiver and feed it to your PC then more can be done than just analog filtering. Any spectral analysis software can isolate very signals, this is for instance what PSK reception is based on. Also CW is filtered out in this way by software on my PC, a FIR filter is used to recognize weakest CW signal that are just a few 10’s of Hz apart. Analog filters don’t unless you specifically incorporate them in the design of the receiver, my 10 euro DCF77 time receiver that I once ordered from conrad or that you find in your alarm clock is based on this principle. The receiver is only able to listen to a time signal, and the bandwidth is more narrow than the CW filter.

An analogy: in GPS we have a processing gain that is unrelated to the performance of analog filters. The BPSK of the GPS satellite is such that the transmitted signal is spread over 10 MHz so that the PRN signal arrives below the noise floor of -174 dBm/Hz at the GPS antenna. The LNA in the antenna amplifies the PRN signal that is hidden in the noise, but it is code correlation in the receiver that finally reveals the signal. The processing gain by code correlation in any GPS receiver is 43 dB for the C/A code which is a 1MHz BPSK signal, you calculate it as 10 log10(2e6/100) = 43 dB, which is is enough to lift the GPS BPSK signal above the thermal noise floor of the receiver.

Lets return to PSK. It is also true that the quality of the receiver to suppress man made noise (QRM) from neighboring transmitters does affect the PSK. Oftentimes an automatic gain control is more than a nuisance than that it is a blessing, because the AGC in the the FT-991 does reduce the amplification and push a weak PSK station into the noise whereas it is not. In that case it helps that you can turn off or slow down the AGC, my FT-991 fortunately can. What often also works is to apply a CW filter rather than a full 3 kHz audio spectrum from the SSB receiver. However, in CW you can not transmit audio on the FT-991 so that I can only listen to the very weak PSK stations when the QRM is significant.

In the end what counts for reception is your local environment, the antenna, QRM caused by your household and the neighbors, the solar flux index (which we want high), the magnetic field index (which we want to be low), atmospheric noise in general which we want to be low, and maybe your luck in finding that weak QRP station before the pile-up starts. This is what you can not find in the Sherwood tables, it is also not a property of receivers, true, but maybe it was good to remind the buyer of a potentially too expensive radio.

Finally I could not resist to compare my airspy SDR receiver to the FT-991 and the websdr at the University of Twente. If you don’t own a good receiver then pay a visit to a websdr. I found that the airspy noise floor is certainly elevated relative to the FT-991, but also, any SDR can suddenly show signals at frequencies where the FT-991 and the websdr’s say that nothing is going on. Oftentimes this is a SDR problem of setting the gains correctly, the gain at the HF pre-amp, the mixer gain and the IF gain. What also helps is oversampling in the SDR and allowing a high resolutionin the waterfall plots. Just a waterfall plot on the PC is a remarkable asset, it does not need to be a receiver capability because oftentimes we have a PC running next to the transceiver. Just ground your PC to keep the noise floor down.

The comparison of a websdr to your own receiver is something I certainly recommend. Twente’s websdr tends to pick up more the German stations which are visible within its local horizon, I tend to pick up the stations that are further away and depend on signal propagation and the ionosphere to arrive in Rotterdam. Twente ‘s websdr oftentimes does not hear me while stations 1500 km away do hear my transmissions. Therefore I also watch websdr’s that are located furher away from Rotterdam, for instance in Sweden where you can oftentimes hear your own SSB transmission.

Last update : 1-May-2016, added some details on digital signal processing




Whispering airwaves

CW (continuous wave) or Morse code (named after its inventor Samuel Morse) is the oldest way to transfer text, we control the transmitter with an on/off key and the dashes and dots result in characters. With bi-phase shift keying (BPSK) we don’t, instead we leave the transmitter on while we transmit a short message. Mathematically you can show that the second technique works better. The Fourier transform of a square wave looks ugly, in particular the odd harmonics of the base period repeat almost forever. In order to prevent that this happens we shave off the edges to make the square wave behave better in the radio spectrum. If you don’t then key-clicking is the result, which could potentially lead to complaints from your neighbors or the government responsible for monitoring the spectrum. CW also has other ugly aspects, namely, that the power supply varies between low and full load so that the voltage may drop affecting for instance the frequency of the oscillator. Sometimes you hear this as chirping in the CW code.


With BPSK there is no key-clicking because we leave the transmitter on, but, in order to transmit dashes and dots we change the phase of the carrier as indicated in the above figure. An OFF in CW has a zero-phase in BPSK, and an ON in CW has a phase of 180 degrees. The BPSK acronym stands for Bi-phase Phase Shifted Key, it was introduced in space communication for instance with the introduction of GPS in 1980 where just 27 Watt is transmitted at 20000 km altitude with a directional helix antenna so that it appears for us on the ground as a 500 Watt antenna in the sky. The more generic name that covers BPSK is phase shifted keying or PSK because more than one phase setting may be used, also, the rate at which we send the ones and zeros in PSK may vary.

PSK also found its application in ham radio, near 14070 kHz you can receive PSK signals with any USB receiver. But the fun is that we can also transmit PSK signals by providing the an audio signal generated by your PC to your transceiver. There are several implementations of PSK, you can generate for instance BPSK at different speeds, best known are PSK-31 and PSK-63. Also you could generate more than two phases, like 4 in which case you get QPSK code. Finally you could also apply an amplitude variation together with phase modulation, an example is QAM. For sake of simplicity I will stick to what most people do on HF, and that is PSK-31 or PSK-63 modulation.


When you listen to BPSK on HF it sounds like a faint whisper, the demodulator in MMVARI (a windows program on your PC) picks it up and displays the message in the upper text box. What you reply is like in CW traffic, use your call sign and exchange reports etc etc. The efficiency of the information transfer is incredible, under mediocre HF propagation conditions I was able to establish contact with stations on the East coast of the Black Sea in the early morning, later in the morning with Sweden and Italy. There are also PSK monitoring networks like pskreport that generate reports of what you’ve sent. In the evening you will see that stations on the east coast of the US pick up the transmitted signal.

I used the MMVARI application on my PC together with the micro keyer 2 to generate PSK-31 and PSK-63 code. In the beginning nobody returns an answer to what you send because the transmitter really needs an audio in signal from the PC (Even I make these mistakes is what you say after it did not work). Plugging in the RJ-45 cable is what I avoided so far but there is no other way to make the Yaesu FT-991 process the sound generated by the MMVARI program.


Another tricky aspect with all PSK modulation is not to over modulate the SSB transmitter. There is a ALC meter on the FT-991 that shows the automatic level control of the end-stage and there is a transmitted power meter. It is however not clear to me what the ALC meter on the FT-991 really shows, I’m inclined to believe it shows the same thing as the PO meter, so equivalent to power delivered by the transmitter. PA0HPV Henk explained me that if the transmitter can output 100W then PSK modulation has to stay at about 1/4 of that maximum, roughly 25W. The best option is to put a dummy load on the transmitter PL-259 jack and to inspect with an oscilloscope via a T-connector between the TX and the dummy load to see what comes out. Poorly modulated PSK signals lead to splatter, and this is what you want to avoid. Occasionally you see PSK signals in the spectrum that show this behavior. I guessed at this point what the proper audio level was, at some point ham operators responded to my BPSK calls which is a sign that apparently something is transmitted and recognized. The optimal PSK settings are on my list, this is something I have to look into as soon as I have a dummy load and the required T-connector. 

The image below shows mostly receivers connected to the internet that skim the allocated frequency bands for any PSK activity. It shows that my signal was automatically picked up and reported to the webserver, these are the balloons in the plot. Reception reports are convenient for investigating HF propagation characteristics. There are similar websites like WSPR where radio amateurs cooperate by transmitting low power PSK signals and reporting the received signal strength.


BPSK reception report, made on the 28th of April the day after I worked on PSK. Not all receivers in the map saw my transmissions, but the signals were seen from the East Coast of the USA up to Kazachstan. The HF propagation on 20m was mostly East West apparently, the F10.7 flux was also low, around 90 but under 100 and the Kp index was slightly elevated which works against stable HF propagation. An earlier version of this web page showed a different version.

Morning report 29-april-2016

This morning stations I had several QSO’s and showed that the signal propagation did not reach further than Turkey and the Ukraine, until I looked again to see that R8OAJ saw the PSK-31 in Novosibersk 4932km to the East, about twice as far as the usual 2000 to 2500 km. I wonder whether it is real.



Last update: 29-April-2016

Interface to the transceiver and go CW

Connecting the transceiver to the computer was not easy, the transceiver has a USB interface but this is good for changing frequencies and settings, but not what I wanted. The microkeyer II (MK2) in the image below on top of the Yaesu does bring you somewhat further. That green wire below the power button on the FT-991 connects the microkeyer II to the CW key jack on the transceiver. The MK2 connects with one USB cable to the PC and I use a separate audio cable from the Yaesu to the PC sound card. Remove the Yaesu USB cable when you do that because it generates conflicts with the autotuner which is also on the CAT interface. The audio cable to the soundcard allows one to generate waterfall plots on the PC and to isolate that part of the spectrum that is the CW code.



To test the transceiver CW capabilities you are best off with a manual keyer. I retrieved my 45 year old keyer which I got as a 12 year old to practice morse code with one of the kids on the block. There is a keyer program in the Yaesu transceiver which is for these fancy double action keyers, but I had to turn that option off. Finally the Yaesu did produce CW code, option break-in has to be on, option keyer has to be off and it is only going to work when you put the transceiver in LSB-CW or USB-CW mode. No CW keying in USB LSB FM or other modes, sure, why should you?


My CW keying practice is somewhat shady, I can read CW but that is about it, transmitting CW  is even worse. Perhaps I could come to the point of reading and sending 10 words per minute, but it is far easier to tell your PC to do the work. Most CW you hear nowadays is made by the computer I thought until PA4M reminded me that this is not necessarily the case. Trained CW operators don’t use a PC, but are armed with handkey, paddle, bug or sideswipers. They all come with a distinct sound and the trained ear will recognize this according to PA4M. It is on my list, I will try to recognize it.

The speed of CW code varies between 15 and 40 words (like PARIS) per minute, and when the signal to noise ratio is not too bad your CW software (I used CWget among others) simply decodes it. Transmitting CW is more difficult, the USB router software that comes with MK2 needs to assign a COM port for winkey, most software does not use winkey and attempts to directly cast audio into the transceiver. Sending audio from the PC to the transceiver is something that I never got to work, maybe I will at some time, but so far winkey does the job.

The reverse beacon lookup map that I found via the dxwatch website was sort of a surprise to me, it shows where my CW’ing was heard on 20m. May I now legally brag that my keying was heard across the Atlantic?


The reason that CW is so efficient is that all the transmitted energy is compressed in a very narrow part of the spectrum, probably less than 400 Hz, so that the spectral density is large, I guess it to be 10 or more times as dense as SSB with a bandwidth of 3 kHz. Narrow FM at 12 kHz used by radio amateurs is even worse, broadcast FM is more like 200 kHz bandwidth because of different frequency deviation. It is the spectral density that makes CW so impressive. This week ionospheric propagation conditions were insufficient to cross the Atlantic with SSB, “you hear them talking in New York but never say something back (no pun)”.  Apparently with all means at my disposal the computer generated CW did make it across the Atlantic, I was impressed.

April 25th is Marconi day, born in 1874 he was the first to send a wireless message over sea in 1897. In 1901 he was able to transmit the first radio message across the Atlantic, it was a remarkable invention that changed the world. [link]

End fed HF wire antenna

For a HF antenna you can make a half wavelength dipole, but the problem is that it is only resonant at the wavelength you design it for unless you insert some loading coils. Moreover the center feeding on a half lambda dipole is not an easy option unless you have two masts and a balun at the center point. In my case I just wanted something simple to start with to see what the transceiver (a yaesu FT-991) does on the HF bands. So I installed an end-fed antenna that consists of an impedance matching circuit, essentially it is an impedance transformer called an unun that converts the high impedance of the end-fed antenna towards a lower impedance that matches your coax cable. Furthermore you need 10.1 meter of wire with a loading coil 1.85 meter away from the impedance transformer that you see in the first image.


Looking into the garden


And from the tree to the house

The core of the unun is a ferrit ring that transforms the high impedance of the antenna to a lower impedance, typically around 50 Ohm which is what the coax cable has. The transceiver also performs impedance matching, about half the space inside the transceiver consists of coils and relays that do the job, but I also got an external autotuner which can perform impedance matching over a slightly wider range.


This is what inside in the box, the transmitter input goes to the black coil that consists of two turns inside shrink tube, it has a 100 pf high voltage capacitor in parallel (I mistakenly wrote series in a previous version). Next there are 7 tight turns, you cross over, and there are 7 wide turns. Wire straps keep the turns in place. Some people prefer glue or wax to keep everything together. The ferrit ring in a material that is electrically not conductive, but magnetic field lines really like to stay in there so that it increases the inductance of any coil. This is precisely the function that it has, magnetic field lines are transferring the energy from one coil to another so that in the end the impedance of the transmitter (it feeds with 50 Ohms) is transformed to a high impedance of the end fed antenna. To avoid (or better, suppress) eddy currents in the mantel of the coaxial cable you clamp some ferrit tubes around the coax cable. For low power transmissions up to 25W this is enough to avoid complaints from you family members that they can alien voices through they headsets etc.

The autotuner  (an LDG YT-1200) gets the end-fed antenna in resonance between 7 and 30 MHz. It results in SWR’s typically around 1.5, all together this is a relatively simple solution, you don’t need a structure on the roof, and the components are relatively cheap. This shop provides you with all the required components. So far I’ve been able to establish contacts within Europe on 40m, 20m with no more than 25Watt. See also my logbook on QRZ and in particular the logbook on that site.



This is the yaesu FT-991 transceiver, it needs a bit of current (up to 25 Amps) at 13.8 Volt and this is what the power supply on the right side does. On top of the power supply is the autotuner that talks with a cat interface cable to the transceiver.

When I showed the first image to my mother of 89 she said, tell me, what do you see in the bushes? I explained her that it was an HF antenna and that you could talk all over Europe with a wattage not enough to light up your living room, but apparently enough to talk with radio amateurs 1000’s of km away. She was raised on Java Indonesia which was a Dutch colony up to 1948. The radio receiver in those days was a box with glowing radio tubes in the corner of the room, and she recalled transmissions from a radio station in San Francisco but also the propaganda transmitter radio Zeesen in Germany. In those days the frequencies were below 300 kHz and perhaps already in the lower end of the current MW. Radio in those days was the only way to find out what was going on in a world at war where disinformation and information had to be separated. Radio San Francisco did a great job for the Dutch on Java in Indonesia, the radio system was in those days also a few meters of wire to a one of the trees in the garden. In Europe the possession of radios was prohibited, but my fathers family (like many other families) has the receivers stored somewhere deep away in closets etc. The free word, this is what you heard in the bushes.

Last update: 14-April-2016