Test of OCXO and Rubidium Standard. And Building a Bit.

I wanted to test some 10MHz OCXOs so I needed some stable reference. I have a surplus rubidium 10MHz standard, and tried it for the first time.

In the first instance I want to get my very old, but otherwise frequency counter up and running again. The counter did have a reference oscillator locked to the old transmitter frequency of the national long wave station of 245kHz, which moved to 243kHz in order to comply with the new frequency grid. The reference frequency of the counter is 1MHz, generated by a crystal oscillator, which cannot now be locked to the standard. A new reference oscillator is therefore needed. The counter does have an input for an external 1MHz reference.

When I last tested the counter, the frequency adjustment range of the oscillator had deteriorated, so it could not be adjusted to the correct frequency, so I decided  to generate an external frequency.

The intention is to use a oven controlled crystal oscillator (OCXO) which can then be calibrated occasionally with the Rubidium standard, or maybe later, with a GPSDO reference.

In order to test the OCXO an even better reference is needed. so I found my surplus rubidium  (Rb)standard and tested it for the first time. I do not know its exact frequency (well 10MHz), but I could see the classic frequency sweep for some minutes, and after some cycles it locked into place.

So now I have a stable 10MHz reference, which should probably be calibrated, but I will use it as it is for now.

The first OCXO needs a 5V supply, and the Rb but at the test bench I had only one variable supply and a fixed 13.8V supply. I then built a quick 5 and 9V supply, using two DC/DC buck converters. All went into a small plastic box, normally used for interconnections in electric installations. The voltage into the converters can be from 12 - 30V, so I connected it to the 13.8V supply, the variable lab Power supply was then available to provide power for the Rb standard-

In order to compare the frequency I connected the two oscillators to my 2 channel scope, triggered by the Rb standard (Ch1) and connected the OCXO to Ch2, after letting them warm up for an hour or two.

Of course, the signal from the OCXO will then move across the screen, because its frequency is different from the Rb standard. The trick is then to change the control voltage of the OCXO until the signal moves as slowly as possible across the screen.

I managed to adjust it to move across one period, taking 3min:40sec (220sec). This corresponds to a frequency difference of less than 5mHz (yes, *milli*Hertz), but it was still drifting a bit.

However, this is quite acceptable, because the OCXO was just lying on the test bench, and not even insulated. A well insulated OCXO should be sufficiently stable for my amateur use (when placed indoors), even up to 10GHz equipment.

What next? I think I should build one of the OCXOs into a box (w/insulation), then add some buffering for the 10MHz signal, so a few output signals can be generated, as well as a "converter" for generating a 25MHx signal for control of an LNB, and finally a divider for generating a 1MHz signal.

The intention for the 25MHz signal is using a divide-by-two (5MHz), then extract the 5th harmonic, and voilá ... 25MHz. A square-wave conversion circuit followed by a good filter is probably a good idea.

Oh!, and testing the other OCXOs need to be done, too

All in all this was a few productive days, getting more test equipment up and running. Always enough projects to do here.

Outdoor LNB for QO 100 Mounted, and Quick TX Up-Converter Ideas.

The LNB for the OSCAR 100 downlink receiver was sealed after the modification, and has replaced the temporary one mounted at the 60cm dish.
I still need to complete the indoor buffer circuit, but some preparations for other stuff had to be done first.

Next comes the transmitter up-converter to 2.4GHz. I am still waiting for a passive mixer module specified up to at least 2.5GHz, and they are on the way.

The configuration will likely be as follows:
432/439MHz transceiver frequency -->
attenuator and the LO *) signal -->
Passive Mixer -->
amplifier, probably made with the RF amplifier of a modified (filters) MMDS converter (when using 432/439MHz as IF, these filters should provide sufficient image rejection) -->
20 - 50mW amplifier -->
1W amplifier -->
cable to feed antenna system -->
"8W" amplifier, likely providing 4-5W at the feed -->
feed antenna --> dish antenna **)

*) LO signal provided by a ADF4350 synthesizer board controlled by an Arduino Uno board -->
about 20mW amplifier at 1962/1968MHz.

**) experiments with other antennas will be done. I already have a panel antenna for RHCP and a 1m long yagi, as well as a cheap, probably not very good, 17elm yagi about 40cm long, as well as a PCB LPDA antenna for test as a feed antenna.

This should provide a decent up-link signal to OSCAR 100, useable for SSB, and certainly for CW or digital modes.

The building activities will be halted for a few weeks, but mid april they should resume.
More than enough experimentation before the refinement of the system starts.

The buffer for the RX reference LO should be ready and tested soon, though.

OSCAR 100 External Reference, update.

The external reference oscillator works fine in quiet weather.
See previous post for more info.

With rain and gusts of wind the simple solution showed its weakness.
The oscillator is not buffered, so any change of load will change the frequency enough to create chirping sounds on the OSCAR 100 beacon.

In short, a buffer stage is needed, and the outdoor part of the installation needs to be made a bit more rigid.
The outdoor part is relatively easily done, and some more outdoor work was necessary, anyway.
For the addition of the buffer stage a bit more work is needed, with some down time for the receiver.

So, when there is time, I will need to get to work on that.
... More solder smoke needed.

Receiving OSCAR 100, Stage 2: External Reference.

A few days ago I made the simplest possible modification for external reference of a LNB. The LNB was a Twin type with two independent outputs. It was purchased with a local low cost dealer, and turned out to be a PLL type, suitable for modification. They also have a single PLL LNB, of which I purchased a few, just for experimentation (I said they were cheap).
The 2 port LNB modification is done by sacrificing one of the outputs for use as a reference frequency input. This eliminates some SMD soldering. Here is the process:
1) The crystal was removed after looking with an oscilloscope which side was the reference input for the PLL IC.
2) The port nearest the input was disconnected from the output circuits by cutting the PCB tracks
3) The input of the PLL IC had a connection to ground with a capacitor. PCB track to that one was cut, too.
4) A relatively large (1006) SMD capacitor of 220pF or so was soldered directly from the (now) input terminal to the soldering pad for the crystal. Not very pretty, but it works nicely.

That's it ! (pictures will follow.)

Testing this with a signal generator from a transceiver test set failed miserably. I could not find the signal in SSB mode, so initially I thought I had destroyed the LNB.

Testing the signal generator with a SSB receiver on 25MHZ revealed the problem: The 25MHz signal had audible small frequency variations, sounding like something between a warble and a rumble. Multiplied by about 400 this made the SSB signal extremely hard to find, but with the receiver in wideband FM mode, I found it. The LNB was OK.

Now for building, in the simplest possible way, a 25MHz oscillator stable enough to receive signals. I found in my drawers a canned crystal oscillator marked 25.000 00 MHz, and soldered the circuit, including a 7805 voltage regulator and some decoupling capacitors (ceramic and electrolytic), all mouned in an old (used) die cast box with BNC connector and a feed-through capacitor, used for reference frequency output and supply voltage, respectively.

Everything connected, and voila! Test signal received.

Getting the LNB out to the dish, and - here we go. After settling in for a few hours, the OSCAR 100 beacon was about 15 kHz high on my receiver, corresponding to the 25MHz oscillator being about 40Hz too low, but with a clear tone, if drifting a bit. It settled quite nicely in the evening. What a relief having a more stable reception. Stable SSB signals were readable without re-tuning for longer than 5 minutes.

The result is quite satisfying, and should be quite useable for normal SSB/CW contacts via the satellite.

Opening the window next morning got the reference drifting down 5-10Hz, moving the LNB output signals a few kHz upwards.

I decided that it was foam insulation time. I found a piece with a cut-out almost fitting the die cast box I use. I had to make a small cut-out of a 1cm wall and for the cables (power ans 25MHz out), and a piece of flat foam to make contact. Simply taped them together with - yes, you probably guessed - duct tape. The frequency slowly settling, interesting to see where it ends up. 

The oscillator frequency was slowly drifting upwards, even with the window open, bringing the reference closer to the wanted frequency of 25 000.000kHz.

There seems to be just a bit of heat generated in the box, and I hope it would stabilize.

I will wait a few hours to see how it settles.

The LO seemed to settle around 8 - 9kHz too low on 10GHz.

This morning the offset had increased a bit, to just above 9kHz. Opening the window appears to increase that to about 10kHz. Much better than without the insulation.

I think I will try with some resistive heating inside the box, to see if I can bring the frequency closer to the wanted one. A bit of experimentation is probably necessary.

Very usable, but I will probably try to mount a, say 220ohm, heating resistor to see if I can get closer to the wanted LO.

Could be fun to get within 1 - 2kHz of the wanted frequency most of the time.

Long term I will have to make a better reference frequency generator (probably GPS disciplined or a Rubidium standard), but this will have to do for now.

Time to think about up-link transmit capability. Probably from 432/439MHz, because it is easier to filter out the image frequency. I have some of the stuff, and more is on the way from China.