Some time ago I purchased a few surplus 2GHz oscillator modules fro RF-Microwaves in Italy, designated SU-03. They are now sold out, but I finally got to start testing a few of them.
Those are synthesized modules operating on 2009MHz, with a reference crystal of 8MHz. From the description it is indicated that with a modified reference frequency the unit could be modified to operate in the 1960 - 2035MHz band.
The SU-03 requires a dual power supply of +/- 12V. For the test I used my standard variable lab PS with 12V, and a set of 3 Li Ion cells to provide the negative voltage. The negative voltage is needed, as the unit has 2 GaAsFETs that need a negative gate bias.
Power output is specified to 10dBm (10mW), so to be sure not to over-load (and destroy) the probe of my old HP432 I connected a 10dB attenuator at the SU-03 output. That was good, as the measured output turned out to be +15dBm (30mW, well above the spec of the thermistor probe). Better safe than sorry. I now have 3 units tested and in-spec for the +10dBm output. This should be quite sufficient for use with a passive (diode) mixer.
Next step: Look at the spectrum coming out of the SU-03. Here is where I think that the limitations of the low cost Chinese spectrum analyzer, covering 35 - 4400MHz comes in. This is a device costing less than $100, so how can we expect miracles? Well, we can't.
First of all, the maximum scan bandwidth of the spec-an is 350MHz with a 500kHz "IF" bandwidth, so in order to "see" the full spectrum I need to look at 300MHz at a time, then switching to the next segment, etc. It takes some time, and it is tedious, but it can be done.
Second limitation I see is, as I suspect, the local oscillator - I guess an ADF4351 (or a clone) - has a square wave output, so subharmonics of the original input frequency show up on the display (1GHz, 666MHz, 500MHz) quite strongly. Well, knowing the limitation is half of getting more reliable results.
Between 2GHz and 4GHz I see no spurious coming out of the oscillator, at least they are about 60dB or more down from the wanted signal. Also, between 1 and 2 GHz I see no spurious signals. This indicates that the output from the oscillator is rather spurious free.
When looking at a more narrow frequency span, the limitations of the low cost spectrum analyzer really shows. The curve is no longer a curve, but has steps of about 5dB in the in the spectrum, and further it is possible to see the effect of the direct conversion design in the analyzer, in that the center null is somewhat visible.
When looking at a moderate bandwidth, some asymmetric sideband noise is visible. I can not se if this comes from the oscillator block or the local oscillator in the spectrum analyzer, but I suspect it comes from the low cost local oscillator in the analyzer.
Clearly, the low cost solution, while not useless, is not very efficient, but for now it will have to do, as I do not have a (more expensive) analyzer covering up to 2GHz or above. What I do have is a rather old (analog) model covering up to 1.5GHz. With that I can test for spurious output(s) below 1.5GHz, that's all.
Next test of the oscillator will be a frequency test. While I do have a frequency counter covering up to 2.4GHz it is not locked to a frequency standard, neither does it have an input for doing so. So high precision frequency measurements are not possible.
The frequency counter that does have such an input just covers up to 1.3GHz. What will be necessary to use my GPS controlled 10MHz source is a divide-by-10 (counter), because this particular counter requires a 1MHz external reference frequency. Some soldering work needed.
There is, however, the possibility of adding a pre-scaler to it so all is not lost. It looks like I should get the soldering iron going again, so I can get a divide-by-4 pre-scaler up and running, with that one I should be able to measure frequencies up to 3.5GHz (with the use of an added calculator) with good precision. .More soldering work. Also, this would be my first (fully) microwave construction, not just using modules or adding modulators or doing small modifications. Now it gets interesting (hopefully not in the Chinese sense of the word ;) )
What is this (modified) oscillator brick useful for? Let me see, with modifications:
- 1968MHz is within the range of operation of the VCO, so a local oscillator for a 432 < - > 2400MHz transverter (or QO100 up converter) is possible
- possibly a base oscillator for a 10GHz transverter (followed by a x5 frequency multiplier, e.g. 1987.2 x5 -> 9936MHz - LO for a 10368/432MHz conversion)
- a precise 2000MHz oscillator for down conversion of 2400MHz to frequencies that a better spectrum analyzer or frequency counter. Possibly with a divider to 1GHz and/or a frequency multiplier for higher frequencies. We shall see. what I do
Small update:
I tested the oscillator with the old spectrum analyzer (up to 1500MHz) No subharmonic signals could be seen. The oscillator is free of spurious signals from 0 - 1.5GHz on this one. Nothing could be seen from 1-2GHz and 2-4GHz. My conclusion is that this oscillator module is very clean, at least with regard to spurious signals. I cannot measure phase noise, but I can probably get that done at a ham radio friend with better instruments.
This looks like an excellent candidate for 2.4 and 10GHz transverters with 432MHz IF. It is fair to say that a ham radio friend recommended this at a small ham meeting, so I mostly expected this.
The interesting part comes when I try a modification for an external reference frequency signal, how much it depends on the purity of the reference signal.
More on that later.
No comments:
Post a Comment