It turns out that the newer (version 4) of the QRP ProgRock board has pads for a TCXO. See below for a link to the ProgRock web page.
The "ProgRock" synthesizer from QRP Labs is an inexpensive device based on the Si5131 "any frequency" synthesizer that may be used to produce up to three frequencies simultaneously - typically from around 8 kHz to around 200 MHz (with some limitations) but it may be coaxed to go down about 3.5 kHz as high as 290 MHz.
Typically programmable using a pushbutton and a DIP switch, newer versions of firmware may be programmed via a serial port as well. These devices also have an input from a 1PPS (1 pulse per second) source, such as a GPS receiver, to allow precise setting/control of the frequency.
Unlike a VFO, the ProgRock produces only a set of fixed, pre-programmed frequencies: Up to 8 "banks" of frequencies may be selected via three digital select lines.
What sort of things might this be used for?
- Arbitrary frequency sources for the workbench.
- Providing clocks for digital circuits.
- The local oscillator of a fixed-frequency receiver or transmitter.
- An internal local oscillator for a radio - such as a a frequency converter or BFO.
For casual use, the supplied crystal - a typical computer-grade unit - is adequate, but if you need the frequency to be held to fairly tight tolerance - say, a couple of parts-per-million - over a wide temperature range you will probably want something else. QRP Labs does sell an "OCXO" version of the synthesizer which works well, but it is more complicated to build and adjust, it consumes several watts of power and produces extra heat.
You might ask: "Why not just use the 1PPS input for frequency control?"
Such step changes would probably go unnoticed on SSB or CW, but with certain narrow-band digital modes there might be a problem. While modes like WSPR or JT-65 can deal with frequency drift, this would normally occur very gradually over the period of several symbols giving the decoder enough time to track, but if the frequency shift were very sudden, a few symbols would probably be lost. While the occasional loss of data is normal, any loss caused intrinsic to the receive system - perhaps due to frequency steps of the local oscillator - would degrade the remaining error-correcting capability overall.
In other words: If phase or very fine frequency changes will affect your communications, you might not want to use the 1PPS input.
Using a TCXO:
Another option is to replace the crystal with a TCXO (Temperature Controlled Crystal Oscillator). These small, self-contained oscillators have on-board circuitry that counteracts the temperature-related drift, holding the frequency relatively constant over their design range.
A suitable device is a part made by Taitien and is readily available, being DigiKey part number 1664-1269-1-ND (Mfg. P/N TXETBLSANF-27.000000). This device is tiny - only 3.2x2.5mm square so soldering to it is a bit of a challenge - but still manageable with a fine-tipped iron and some magnification.
As pointed out in the QRL-Labs documentation, some TCXOs may have "stepped" frequency adjustments as part of their temperature compensation due to a built-in temperature sensor and D/A converter referencing a look-up table. If sufficiently large (e.g. results in more than a few 10ths of Hz "step") these frequency discontinuities can disrupt/degrade modes such as WSPR that operate over very narrow bandwidths. If a TXCO does this, the synthesizer being controlled by it will also exhibit the same frequency steps, proportional to the output frequency.
The Taitien TCXO units noted above were observed at 432 MHz (the 16th harmonic of the 27 MHz TCXO) using signal analysis software to magnify possible frequency steps: If such "step" behavior was happening, it was smaller than 0.34 Hz at 432 MHz (e.g. 0.02Hz at 27 MHz.)
The power requirements of this device are very low - only 1-2mA, far less than the 100-200mA of a warming crystal oven - and it may be powered directly from the existing 3.3 volt supply of the synthesizer board. This TCXO produces about a volt pk-peak output which is in line with what the data sheets for the Si5351A suggest for a capacitively-coupled external signal being fed into the crystal input. It is possible that the Si5351A would work just fine if this TCXO were directly-coupled, but I included the capacitor just to be safe.
Wiring the TCXO:
As noted above, later version of the ProgRock have pads for a TCXO, albeit one with a different footprint than above. The device suggested by QRP Labs is the FOX924B-27.000 (Digi-Key P/N: 631-1075-1-ND) which is quite a bit larger than the Taitien device and has a rated stability of 1.5ppm. This device has a higher output voltage swing, which allows the omission of the coupling capacitor used with the Taitien device noted above.
I first removed the crystal and cleaned the holes of solder. The TCXO module was then glued using cyanoacrylate adhesive (a.k.a. "Super Glue") "belly up" to the circuit board (after it was cleaned with denatured alcohol) at a location on the bottom side of the board between the synthesizer chip and the crystal position as shown with pin "1" in the lower right corner. With the oscillator firmly in place, a small piece of 30 AWG wire was used to solder pin "4" (V+ - upper-right) to the nearby connection of C3, one of the V+ lines for the synthesizer chip. Connected to the opposite corner (pin 2, lower left) another short piece of 30 AWG wire is connected to the other side of capacitor C3 to provide the ground.
A small, 1000pF disc ceramic capacitor was inserted into the bottom side of the board to connect to the crystal terminal closest C3 and the "CLK 0" terminal with the other lead carefully formed and bent to be soldered to the upper-left pin, #3 - the output terminal of the TCXO. Once the capacitor is soldered into place it is a good idea to re-heat the capacitor's other lead (the one soldered into the board) to relieve any mechanical stress that might have occurred from bending the lead to fit to the connection.
Before soldering to the TCXO - but after it has been glued to the board - it is recommended that a small amount of liquid flux be applied to the connections and that they be tinned using a hot iron with a very fine tip: The ceramic package tends to draw away heat quickly, making it a bit difficult to solder and tinning it before-hand assures that a solid connection has been made. Don't tin the unused pin as it's an easy way to identify the pins of the device while it is inverted.
Assuming that the connections are good and that the pins were properly identified, the synthesizer may be plugged into a ProgRock as normal. If all went well, the output frequency will be pretty close to what it was before - but slightly low in frequency. While the nominal frequency of the original crystal is 27.000 MHz, in this circuit the frequency is usually 2-5 kHz high in this circuit so the "default" clock frequency of the ProgRock is set to 27.003 kHz to compensate. With the TCXO being within 1PPM of its intended frequency, register 02 of the ProgRock will have to be set to the new frequency, hopefully within a few 10s of Hz of exactly 27.0 MHz: If you have a means of precisely measuring the frequency, use that number for register 02, otherwise use 27.000000 MHz. Once this is done the programmed, output frequencies will be quite close.
Once everything was checked out I put a few more dabs of adhesive on the capacitor and flying leads to make sure everything was held into place.
I put together two of these TCXO-based ProgRocks and when compared to a GPS-referenced source, I found one to be 1 Hz high (e.g. 27.000001 MHz) and the other to be about 13 Hz low (26.999987 MHz) - both well within the 1PPM specification. These frequencies were programmed into register 02 and CLK0 was set to precisely 10 MHz and I found the output to be within 1 Hz of the intended frequency. I then heated and cooled the units and observed that the frequency stayed well within the 1PPM spec, indicating that all was as it should be.
Stable receiver local oscillator:
An immediate need for a stable frequency source came about recently while I was putting together a module that is designed to cover the entire 40 meter amateur band in two segments using two "SoftRock Lite II" SDR receiver modules. Normally these ship with crystal oscillators, but the use of a single ProgRock module allowed a pair of these receivers to collectively cover the entire 40 meter band with very good frequency stability - important if digital modes such as WSPR are to be considered.
Figure 3 shows the result. Both receiver modules and the ProgRock were mounted in the lid of a Hammond 1590D die-cast enclosure and a simple 2-way splitter using a BN-43-2404 binocular core was constructed. The end result - when coupled with good-quality 192 kHz sound cards - is a high-performance, stable receive system capable of covering the entire U.S. 40 meter amateur band - with a bit of overlap in the middle and extra coverage on the edges.
Replacement of a crystal oscillator on a phase-modulated transceiver:
These days it is increasingly difficult to source custom quartz crystals for older "rockbound" commercial radio gear. An example of this is the GE MastrII line of VHF (and UHF) transceivers that require a crystal for each transmit or receive frequency. Even though this equipment is now quite old, it is still useful as it is quite rugged and has excellent filtering and when properly prepared, it has been proven to very reliable.
These radios use crystals in the 12-13 MHz area for transmit and 16-17 MHz area for receive so a ProgRock can be easily programmed to be used in lieu of a crystal with a slight modification of a GE "ICOM" channel element. Because the MastrII transmitters use phase modulation, the signal source is never modulated - and this is an advantage if you happen to need to set several overlapping transmitters to the same frequency and you need their modulation to "track" precisely. By using a TCXO (or QRP Labs' OCXO) rather than the 1PPS to maintain frequency stability, the possibility of occasional "clicks" in the audio due to frequency correction steps is eliminated.
With any synthesizer the concern is that it will produce spurious signals and/or additional phase noise that will degrade the transmit/receive performance, but preliminary testing has shown that even when multiplied to 70cm, the resulting spectra is quite clean - probably good enough to be used on a repeater. If one does do this, there are a few things that should probably be kept in mind:
- Even though the ProgRock can output two frequencies at once, there is a small amount of crosstalk between them and when multiplied to the ultimate VHF/UHF frequency, these low-level spurs could end up on the output. For this reason it would probably be a good idea to use two separate ProgRocks, located apart from each other, in a full-duplex radio. For half-duplex, a single ProgRock could be used with the RX and TX frequencies being toggled by selecting a pre-programmed "bank".
- Its worth noting that in many of these radios the LO frequency of the receive frequency is immediately multiplied by the next change. Testing was done on a receiver showing that the ProgRock could be set to the output frequency of this multiplier stage. This cannot be done for the transmitter as the oscillator's output is immediately phase-modulated at its operating frequency.
- It would probably be a good idea to place some high-Q band-pass filtering tuned to the synthesizer's output frequency to minimize any low-level spurs at frequencies removed from the main output frequency that might be present on the synthesizer. Initial testing didn't show any obvious problems, but using such a filter would be a sensible precaution.
- A very small amount of added "hiss" - probably from low-level phase modulation - was observed at UHF. In normal use, this would probably have not been noticeable unless one did an "A/B" test.
This page stolen from ka7oei.blogspot.com