|PA8W Amateur Radio||
|Soft vs hard antenna commutation|
This page describes the soft switching approach as implemented in the first PA8W conventional doppler.
The same basic approach was used to design the antenna drivers of all later models.
The antenna driver of the doppler RDF is a very important part of the total system, and therefore it may need some extra explaining.
Initially, trying to keep things as simple as possible, I started off with the most simple version of an antenna driver,
a hard switching device that selected every antenna for 1/4 of the cycle time.
It turned out that it made the RDF pretty intolerant to strong signals outside the passband because all switching signal harmonics mix with the RF signals,
causing these undesired effects.
You have to realize that the antenna switching diodes act like diode mixers, so higher harmonics on the control signal would produce a lot of undesired mixing products.
So it didn't take long before I took a closer look to see what could be improved there.
Testing some intermediate driver with reduced switching speed I noticed a big improvement but this driver wasn't flexible enough to get the best possible results,
so I designed a new one, which I still use in all later versions of my RDF's.
This new driver was based on the knowledge that reduced switching speed means reduced reciproke mixing problems.
Also, during my experiments I had found that some overlap between the antennas seemed to be beneficial.
Although I found complex driver designs that used this technology, I held on to the minimalistic approach I generally follow;
just keep it as simple as possible..
So I came up with a pretty straightforward but flexible design.
The new driver was tested and tweaked using a few potentiometers which could set the most important parameters over a wide range,
so I could happily experiment with all kinds of driving waveforms.
After extensive testing, all unnecessary components were removed from the driver circuitry, and the potentiometers were replaced by fixed resistors.
Still the driver was maintaining 99% of its qualities.
The soft drive method produces much less noise and less additional mixing products than the commonly used hard switching.
Additionally, I found that a substantial overlap of time in which two sequential antenna elements are active, results in even better reduction of noise and reciproke mixing.
(The actual switching is done with PIN-diodes or 1N4148 switching diodes. See the bottom of this page for comments on that.)
The next screendumps illustrate the timing of the antennas:
We are looking directly at the voltage on the antenna driver output pins:
The red wave in the picture represents the control voltage of the first antenna.
The blue wave represents the control voltage of the second antenna.
The yellow wave represents the control voltage of the third antenna.
The second blue wave represents the control voltage of the fourth antenna.
The horizontal centerline is 0V, and one vertical division is 5 Volts.
So, the maximum ON voltage is +10 volts, which means that the current through 3 PIN diodes in series will be around 8mA.
The maximum OFF voltage is about -6 volts, which is good enough to ensure excellent isolation of the non-active antennas.
Also obvious is the serious overlap between two antennas, at the 2V mark, where the PIN diodes will start to conduct.
The next screendump will give you an even better impression of the overlap.
Note that there is some difference in rise and fall time of the control pulse. This is due to the simple circuitry.
I did perform a test with additional hardware to match rise and fall time but this revealed no further improvement.
So I decided to stick to the simple design.
Some argue about a real sinusoidal switching signal to be even better as it contains no harmonics.
But this is only partly valid: The most important point is the on/off transition point of the diodes.
If the control signal is passing this point relatively slowly, it will virtually be as good as a sinusoidal control signal.
And using our circuitry, it is much simpler to realize.
Now the same control voltages but measured after the 1k series resistors, so in fact directly on the first PIN diode:
The colours are the same for antenna 1,2,3, and 4.
The horizontal centerline is 0V, and one vertical division is 2 Volts.
Just above 2V you recognize the threshold of the 3 PIN diodes in series, so just above 2V the antenna ON status begins.
Also obvious is the 1,2 division (= 2,4millisecond = almost 1/8 array cycle) of overlap between two antennas.
This means that with every transition from one antenna to the next, there's a short period of time where two antennas are active, simulating a pseudo 8-element array.
Improvement compared to hard switching:
Below spectrograms show the frequency spectrum from zero to 100kHz on the antenna control lines:
First: "Normal" hard switching, harmonics at 100kHz only 41.6dB down compared to the 501Hz fundamental frequency,
and the harmonics attenuation will slow down going up in frequency...
Second: The PA8W soft switcher, harmonics suppression at 6kHz already exceeds the 100kHz value of the hard switcher!
Above 12kHz, all you see is the noise floor of my simple FFT analyzer...
PIN diodes: are there alternatives?
Yes, there are.
In some designs found on the internet the 1N4004 is suggested, but I really can't support that idea.
That's a pity, because it would be a solid diode which would not be zapped so easily if you happen to use a transceiver for the RDF and squeeze the PTT by mistake...
However, for VHF the cheap 1N4148 is a good alternative, according to my measurements:
I took a normal switching circuit as we use in the Doppler RDF antenna's, and put it in the coax between a vertical antenna and a IC746pro receiver.
Then I tuned to a solid carrier on 145MHz and made the following measurements:
It is important to realize that the dB indications are derived from the IC746pro's S-meter, for what it's worth.
No absolute figures!
Another thing to remember is that in this test case, only one diode switch was measured. (In a normal RDF array there are 2 or 3 diodes per element, so overall isolation is much higher, probably more than 80dB.
Also, no other antenna was connected and loading the coax with a 50 ohm impedance like in a real RDF array.
This would further improve off-state isolation.
However, the comparison shows that the cheap 1N4148 is a very good alternative for the extremely small PIN diode.
Another remarkable thing is proven by the test; in contrast with common knowledge;
negative biasing of the 1N4148 does not really improve isolation in the off state, but it does slightly improve the PIN diode performance.
Whether you use PIN diodes or the 1N4148, the performance of the RDF will hardly be degraded.
Isolation will be much higher than necessary, so other factors such as mutual coupling of the elements will be amongst the more important issues.
I now use 1N4148 only, because they are easy to handle, sturdy and very affordable.
More details in the pages describing the mobile and homebase arrays.