Minimising the tuning drift effects due to external temperature variations in the Titanium Satellite C1W-PLL Wideband LNBF Although the Titanium LNB is named in the header, the comments which follow obviously would be applicable to any other brand LNB, providing those devices use a 25 MHz LO injection frequency and the injection point is readily accessible. The primary cause of frequency drift in the LNB is due to the 25.00 MHz crystal changing frequency as temperatures change. This is an intrinsic property of quartz crystals, since the physical area of the quartz wafer changes dimensions as temperature changes. Since a crystal s frequency is determined by it physical area, it follows that these changes affect frequency. Because the LNB we are using was designed for use in TV reception where very wide-band signals are used, there is no need to use crystals with very high temperature stability. Depending on the type of cut, a common general purpose crystal could drift around 12 Hz per MHz. This means that the 25MHz crystal used could vary 25*12 = 300Hz. You might think that this is not such a huge variation, but keep in mind that the LO (Local Oscillator) in the LNB operates at 5.150 GHz. This means that the internal circuitry in the LNB multiplies the 25MHz from the crystal by 206 to generate the 5,150MHz of the LO. As we have seen earlier, the 25 MHz crystal can vary up to 300 Hz. Multiply this by 206 and the frequency of the LO can vary by as much as 6.180 KHz! In circuits where frequency stability is critical, RF engineers have some choices. Because the frequency drift is mainly caused by temperature variations, they can be minimised by placing the crystal in a thermostatically controlled oven, which maintains the crystal at a specific temperature. Crystal ovens are expensive and require a fair bit of power to operate the oven. Another solution would be to use a VCXO (Voltage Controlled Crystal Oscillator). The circuits used in these crystals measure the off-set frequency, which is then translated into a control voltage, which in turn changes the capacitance of a varactor in the oscillator circuit. Such devices are complex and expensive an outside the scope of a mere amateur. Another solution is to use a TCXO (Temperature Controlled Crystal Oscillator). A TCXO would appear to be the most cost effective solution for an amateur. Tolerances between 0.1 0.5 ppm are available for 25 MHz crystals. Even the 0.1 ppm TCXO could show a 2.5 Hz frequency variation, or once multiplied to the LO frequency 2.5*206 = 515 Hz (either up and/or down). I tested a 0.3 ppm accuracy TCXO and found that after the device stabilised the frequency was 25,000,007 Hz. The actual frequency is of course irrelevant, as you will have to search for the signal on your SDR anyway. After about 3 days the frequency varied between 25,000,007 and 25,000,010 Hz, measured on a frequency counter using a GPS disciplined 10 MHz frequency standard. This would be adequate for decoding 10,500 bps burst signals, using a reasonably wide receiver band-width, but the variation is too much for 1200 bps burst signals. Initially I thought I could solve the problem by using a 25 MHz signal produced from a 10 MHz GPS (Caesium Beam Clock) disciplined frequency standard. This would produce a frequency
accuracy of Hz 10-6. This proved to be unworkable, because the internal SMD components (Surface Mount Devices) in the LNB also change their physical dimensions as temperatures vary and consequently their values. Again at >5GHz this will result in considerable frequency variations. In my case I finished up using a PLL signal generator using DDS (Direct Digital Synthesiser) technology with a resolution of 0.01 Hz. It does mean that I do have to periodically manually adjust the frequency of the generator, sometimes by as little as 0.3Hz. Modifying the LNB Before we can use any of the suggested solutions, the LNB will have to be modified to allow external 25 MHz frequency injection. Remove the screws which secure the lid. Once opened up the LNB looks like this: The 25 MHz crystal is clearly visible and we have to remove it. Carefully apply heat from a fine tipped soldering iron to one side of the crystal while gently lifting the crystal from the pad using a small screwdriver. It will now look like this:
The solder side of the crystal we just removed is clearly visible on the left. Next we have to drill a hole to accommodate a suitable socket. I used an SMA socket, because they are small in size. For the location I drilled a hole above the existing LNB F socket. On the outside this is surrounded by a ridge, which I ground level with a Dremel tool and should now look like this: Next we mount the SMA socket thus: All that remains to be done is to connect the SMA socket to the live pad on which the crystal was mounted as shown in the image below:
We can now replace the lid and we should be ready to roll. Because I needed to carry out extensive tests and eventually finished up using an external signal generator, I chose to mount a socket on the LNB. The TCXO is quite small and it should be possible to mount it inside the LNB and not use a socket. The TCXO requires 5V to drive it and this can be taken from the voltage regulator IC visible in the right hand bottom corner of the above image. The TCXO I used a reasonably priced TCXO on ebay for my project ($ 3.50 for 2) https://www.ebay.com/itm/25mhz-tcxo-qty- 2/222682451705?hash=item33d8e88ef9:g:bQIAAOSw4DJYkukr These devices require 3.3 5.0V to operate. Because during my tests I had determined that the LNB likes to see around 6.5V pp on the 25 MHz injection, it is preferable to operate the TCXO at the maximum voltage of 5.0V
To make supplying power to the unit simple, I built a 7805 IC into the project, which means that the unit can be fed from a plug pack with an output anywhere between 6.0 >12.0V. Plug packs are often switch mode power supplies, and they can introduce a lot of annoying noise in an RF circuit. For that reason I used a 470µF tantalum electrolytic capacitor at the output of the 7805, to smooth out any power supply noise. You will also notice a small LED just to the right of the voltage regulator. This is just to show that the unit is receiving power and is not really necessary. Power from the plug pack is taken from an appropriate DC socket which fits the plug pack. This TCXO is rather small and has to be surface mounted. This requires a bit of dexterity and some experience with soldering, but it is not by any means rocket science. I used a small universal mounting pcb I happened to have handy. A small square of Vero board would work equally well. Mount the TCXO as close as possible to the output socket, for which I once again used a SMA type. You will notice that the SMA socket has a short length of semi-rigid coax attached to it. The shield goes to the earth connection of the TCXO and the centre conductor connects direct to the TCXO output. Again the semi-rigid coax is not really necessary and you can use short lengths of wire. My finished unit looks like this:
When fed with 5.0V, this TCXO produces a wave form which looks as follows in the image below: You will note that the output is not a perfect sinusoidal wave form, but it is not bad and a check on the spectrum analyzer confirms that it is free from harmonics. The output level is just below 6.0V pp. I noted earlier that the LNB likes to see 6.5V pp, however it operates quite happily with this < 6.0V pp. The next step is to simply mount or re-mount the LNB on the dish and connect the 25MHz external oscillator source. (and of course the LNB to receiver coax). Usual disclaimer: The procedures outlined above are a record of how I went about modifying my LNB. I cannot and will not guarantee that they will work for you, as there are many variables which are outside my control. Any modifications you undertake are completely at your own risk. I have no association with or financial interest in the suppliers I used for my TCXO or LNB.
My completed installation looks like as shown in the image below. The 25MHz external signal is fed through the white coax This is what the inside of the TCXO looks like. Note the variable capacitor in the top left corner, which allows for fine tuning the output frequency. The top of the crystal is in thermal contact with the case by means of a double sided silicone pad. There is no voltage regulator on board. The measurement units are in mm. i.e approximately 4.5 X actual size. Enjoy! pf