EME223 630m Band Transverter

Please Note: that all information including pictures on this web page are copyright to Mini-Kits and must not be used on any other web site, or reproduced in any publication.

EME223-630M KIT Description:

The complete high performance 472 to 479kHz ( 630m Band ) Transverter Kit makes it very easy to get onto the 630m band especially with weak signal digital modes, but being linear it will work with all modes including SSB if it is allowed in your country. Construction takes around 6 hours to fully built and test. The Transverter is designed as a transmit down, and a receive up converter, with a fixed frequency range of 472 to 479kHz, and produces up to 5 Watts of linear output when used with a suitable low power < 5 Watt Yaesu FT817 or similar 3.6MHz 80M Band Transceiver.  The design incorporates a 500kHz low pass filter that is used on the receiver for broadcast band rejection, and on transmit as a low pass filter to effectively reject the second and 3rd harmonics. There is further filtering in both the RX and TX paths using a tuned 472 to 479kHz band pass filter to effectively remove remaining out of band signals.   The design incorporates a RF sensing circuit for automatic switching from RX to TX mode. For a higher performance station the board has the option to disconnect the low pass filter from the PA so that the Transverter can drive a higher power RF amplifier, and then the receive converter with low pass filter can be used on a separate receive antenna. The basic 5W output with a resonant Antenna should provide contacts using digital modes of up to 1000km.

The Transverter stability is typically around +/-1 Hz with a 2 minute transmit and 6 minutes receive timing using WSPR mode, but may not stable enough for some of the very narrow digital modes that require stability of less than 0.05Hz over a very long period of time. To use these modes an external 3.2MHz GPS locked reference could be used to provide the stability required.

Kit Constructors Alert:

EME223-630M

1/ With the first initial batch of Kits using the Rev1.1 board, there appears to be an issue with Transistor T1 failing intermittently. The problem never showed up when using digital modes only SSB and was traced to the powersupply switching circuit erratically switching between RX and TX modes with voice when link LK7 is not fitted. This causes high RF power spikes out of band from the PA to bypass the diode switch and appear on the base of T1. The fix is to replace C40 with a 1uF capacitor so that there is a very small delay that seems to completely fix the problem. If the Kit was supplied with a NE46134 for T1 then it will probably fail regardless and should be replaced with either a BFQ19, DXT2222A, or PXT2222A. Please contact Mini-Kits for a Kit of parts for your Rev1.1 Kit if the NE46134 has failed.

2/ On Rev1.1 Kits, the 3.2MHz crystal has been replaced as the original HY-Q crystal appears to have high drift with temperature. After testing for drift on WSPR, the crystal heater is now probably not required with the new replacement crystals.

3/ The Rev1.2 board was sold in a small number supplied with the FQP13N10 MosFET in place of the IRF510 that appears to produce more RF output power up to 10 Watts, but the gain is way too high and can actually cause the PA and driver to go into oscillation which can cause failure of the GALI-84 driver. Some further investigation is required here but there was an issue found with instability due to L5 and C22 being close to each other on the board. Current Kits replace C21 and C22 with 1uF 1206 size SMD capacitors that fixes the stability issue, but you can also remove L5 and fit it on the under side of the board so that it is away from C22. It is also suggested that the FQP13N10 be changed back to the IRF510 so that the GALI-84 is not damaged. Please contact Mini-Kits for a Kit of parts if your Rev1.2 Kit with the FQP13N10 has failed.

EME223-630M Kit Construction Notes:

The pictures below show the basic constructed Transverter and the Kit construction notes and overlay diagrams should be carefully followed when constructing. For larger pictures to check orientation of some components click the images below. Construction is quite straight forward with although the bifilar and tri-filar transformers can be a bit tricky to wind correctly. The alignment of the band pass filter and local oscillator are quite easy to do without test equipment. BUY THIS PRODUCT

EME223-630m Top View
Top view of the EME223-630m. The picture shows the basic EME223-630m Kit with SMA connectors that have been used for easier enclosure mounting.
EME223-630m Bottom View
Bottom view of the EME223-630m. The picture shows the minor SMD components on the bottom side of the board
EME223-TF1-Winding
Winding of the Transformer TF1 consists of 10 Turns of the 0.3mm bi-filar wire wound on the FT37-43 core
EME223-TF1-Mounting
How Transformer TF1 is mounted onto the board, Be sure to get the polarity of the windings correct.
EME223-TF2-Winding
Winding of the Transformer TF2 is 5 Turns of the 0.4mm wire twisted tri-filar on the FT50-N49 Toroid.
EME223-TF2-phasing
How the windings of TF2 are orientated for correct phasing of the windings
EME223-TF2-Mounting
Opposite side view of Transformer TF2, , Be sure to get the polarity of the windings correct.
EME223-T50-3-Winding
How Inductors L9, L10 and L11 are wound, 28 Turns 0.5mm ECW on T50-3 Toroids
EME223-L8-winding
How the winding is wound and spaced for the 120uH inductor L8, 15 Turns 0.5mm ECW on the FT40-N49 Toroid
EME223-PTC-Mounting
How the optional 40deg C PTC Crystal Heater is fitted to the 3.2MHz Crystal if required for better stability.

EME223-630M Performance Tests:

EME223-RX-Amplifier-Gain
EME223-630M Lower gain is a 15R degen resistor and 2K7 feedback. the +26dB gain is the final design and uses a 3.3R degen and 2K7 feedback resistors. Overall RX Transverter gain is 0dB
EME223-630m-FL1-Filter-S21
EME223-630M Plot showing the Band pass filter response of the 472 to 479kHz filter FL1 which has a 6dB loss to get the sharp response.
EME223-630M-TX-Driver-S21
EME223-630m Gali-84 TX driver Gain
EME223-12CM-TXOUT1-S21
EME223-630m IRF510 Gain with the bias set correctly to just being turned on.
EME223-630M-500kHz-LPF
EME223-630m 500kHz Low Pass Filter Response showing the high attenuation at 1MHz
EME223-630M-Harmonics
EME223-630m harmonic output from the 500kHz Low Pass Filter when driving the IRF510 to +37dBm 5 Watts. Both the second and third harmonics are 50dBc or greater.
EME223-630M-4MHz-LPF
EME223-630M 4MHz Cheby Chev Low Pass Filter on the output of the 3.2MHz Oscillator
EME223-630m

Mounting into a Hammond Die cast Enclosure:

It is recommended that the Transverter board is mounted inside an enclosure like the Hammond 1590BBS Die cast enclosure for RF shielding and to provide heatsink cooling for the IRF510 / FQP13N10. The heat-sinking seems ample when the Transverter is used with a 2 minute transmit cycle on WSPR mode with a maximum of 5W output, but it is recommended that a small fan is fitted and run at a lower voltage to keep the board cool to minimize frequency drift due to the Crystal heating. It is suggested that the mounting holes for the board are made before the components are fitted to the board. Some care needs to be made when measuring the holes that need to be drilled, as any inaccuracy could result in the holes on the front of the enclosure being out of alignment. The 6.5mm holes for the SMA connectors may need to be enlarged to 7mm as the board is fitted into the enclosure at an angle as extra clearance may be required. Make sure that you fully tighten the MosFET to the enclosure for maximum thermal contact, and check with a multimeter for shorts from the drain to ground.

EME223-ENC-1
How the 3mm holes need to be drilled to suit the mounting of the PC board into the Hammond 1590BBS enclosure. Mark the position of the upper left hole and place the PC board onto the enclosure to mark the other holes. The hole for the IRF510 can be marked as the centre of the hole in the board.
EME223-ENC-2
How the holes need to be drilled to suit the SMA connectors and DC socket on the front of the enclosure. You need to be careful and double check after marking the positions before drilling the holes.
EME223-ENC-3
Standard spacing perforated steel sheet was used for the lid to allow ventillation of the circuit board. A 40mm +12vdc fan with a 47ohm series resistor is connected to the internal +9v regulator for cooling. The fan is placed over the crystal for constant temperature cooling for stability

Antennas and Operation:

The Transverter should only be used on an Antenna that is close to 50ohms to avoid damage to the IRF510. To get onto the 472kHz band quickly, an existing Antenna like a 40 or 80m band dipole could be used as a vertical against ground. If the dipole is already fed with coaxial cable then the inner and outer can be connected together, and a series inductance added to resonate the Antenna down on 472kHz. If you have an old switch mode power supply lying around, then the transformer could be used as a series inductance by connecting together windings in series and selecting taps to tune the Antenna closer to resonance. A simple transformer using a FT140-43 core may be required to then match the 50ohm impedance of the Transverter to the feed point. In reality I found that my Antenna was close to 50ohms, but this is probably due to the ground not being very good as it should be much lower than 50ohms. When you are ready to get serious, then a variable inductance ( variometer ) can be constructed which is much more efficient and can be tuned to resonate the Antenna. The trick to it all is to get the vertical part of the antenna as high as possible and away from metal objects like poles, towers, and trees, as it's the vertical part that radiates not the horizontal wires that are capacitance to ground.

472kHz Variometer
This picture shows a variometer and loading coils wound on a standard 100mm PVC plumbing pipe. The variometer at the bottom has a 200 to 250uH range and the coils at the top 200uH and including the variometer have a 450uH maximum inductance.
472kHz Variometer Internal View
This picture shows the 80mm diameter x 60mm long PVC pipe with 2 x 8 turn windings that swivel on 20mm electrical conduit. Holes in the centre of the 20mm conduit allow the windings to exit through the conduit to be connected to the larger coils on the 100mm pipe.
472kHz Variometer3
This picture shows how the wires from the internal coil exit the 20mm conduit and connect to the two larger coils. My design was based on the G0MRF design and is 2x 33 turns on the 100mm tube with a 48mm spacing. I did not get the wide 250 to 400uH inductance range that G0MRF got
FT140-43 Transformer
This picture shows the matching transformer 2 x separate 18 turn windings of 0.8mm ECW on a FT140-43 core with taps every 2 turns on the secondary. In reality the impedance was close to 50ohms anyway, but the transformer does isolate the Antenna and coax cable ground to reduce noise pickup.
Wisprnet-Map-01
This picture shows the longest distance so far is to VK4YB at 1603 kms at -28dB Signal to Noise. This also confirms that the 5W EME223 Transverter with a simple Antenna is quite capable of being heard at large distances when conditions are good.
Wisprnet-Map-02
This picture shows Stations that heard VK5LO over 2 weeks. Signal conditions are not ideal and should improve as we move into winter.

Kit Changes and Repairs:

EME223-472kHz

Please Note: That the Kit notes are changed at times for improvement or changes to the circuit and have Rev numbers. The PC board also has a Rev number and there was an initial 7 Rev1.1 Kits sold with Rev1.1 boards. All Kits after Rev1.2 are supplied with Rev1.2 boards.

1/ The receiver does not receive any signals and appears deaf. D7 may be leaky due to excessively high RF Transmit power being applied to the Transverter. Measure or replace the BAR64-05 D7. Also check the 5 Watt 51R resistor R16 for any damage due to excessive RF power being applied.

2/ With the first initial batch of Rev1.1 Kits using the Rev1.1 board, there appears to be an issue with Transistor T1 failing. The problem never showed up when using digital modes only SSB and was traced to the powersupply switching circuit erratically switching between RX and TX modes with voice when link LK7 is not fitted. This causes high RF power spikes out of band from the PA to bypass the diode switch and appear on the base of T1. The problem can normally be seen when testing on AM mode ( carrier ), and going from TX back to RX mode there is a high spike reading on the S-meter. The solution was to change C40 to a 1uF capacitor so that there is a very small delay that seems to fix the problem. T1 was a NE46134 in the Rev1.1 Kits and is sensitive to failure and can be replaced with a DXT2222A, or PXT2222A, as they are more rugged, or you can use any general purpose transistor that has at least a 40V C-E rating.

3/ Frequency drift problems have been directly traced to heat from the 5W resistor in TX mode migrating across the board and heating the crystal which causes an increase of the 3.2MHz frequency so a decrease at 472kHz. The 3.2MHz crystals only have a stability of 30ppm so they will never be great with temperature changes, but the stability can be improved by keeping the Transverter as cool as possible with increased ventilation over the low pass filter and and a small fan mounted over the crystal. Some small improvement was also seen when the crystal was lifted slightly off the top of the board, as the ground plane heats and then directly transfers to the crystal casing. Teflon Crystal insulators are readily available and are recommended, along with possibly trying a 40 degree clip on Crystal PTC heater on the crystal. 60 degree clip on Crystal PTC heaters were also tried with limited success, as standard non oven crystals are really cut for room temperature and when heated are probably outside the maximum temperature range. Tests on the other components in the oscillator showed no drift due to temperature changes, only the crystal was prone to change with temperature. See 7. below

4/ Some Rev1.2 Kits were sold in a small number supplied with the FQP13N10 MosFET in place of the IRF510 that appears to produce more RF output power up to 10 Watts, but the gain is way too high and can actually cause the PA and driver to go into oscillation which can cause failure of the GALI-84 driver. Some further investigation is required here, but there was an issue found with instability due to L5 and C22 being close to each other on the board. Current Kits replace C21 and C22 with 1uF 1206 size SMD capacitors that fixes the stability issue or alternatively, fit L5 on the bottom side of the board so that it is a distance from C22. Stability issues will normally show up as a change in RF output power when the lid is fitted onto the enclosure.

5/ In Rev1.3 Kits, the receive amplifier has been modified by replacing T1 with a DXT2222A, R3 with an 56R, and R1 with a 390R which decreases the collector to emitter voltage, and increases the current through T1 to around 40mA which could improve the strong signal handling performance. TR1 if it has failed can be substituted with a DXT2222A, PXT2222A or similar transistor with at least a 40v C-E and 100mA minimum rating. If you get desperate then a BD639 with the flat part mounted against the board with a think smear of heatsink compound is also fine to use.

6.With Kits supplied with the Rev1.2 boards a higher DC voltage can be fed to the IRF510 to produce more RF output power by disconnecting Link LK8, and connecting a second DC power supply to the A5 connector. Currently this has not been tested and it is expected that the 9:1 Transformer TF2 may need to be changed to a 4:1 for impedance changes with the on the output of the MosFET. DC to DC step up switching power supply modules that are rated at 150W are readily available, and can easily supply at least 3 Amps @ 24vdc. It is up to the individual at their own risk to possibly damaging the IRF510 if you want to try a higher voltage, but the MosFET will need substantial Heat-sinking. To avoid any damage to the Transverters circuitry, LK8 must be disconnected when using voltages higher than +14vdc on the A5 connector.

7. In Rev1.4 Kits R15 has been changed from a 680R to a 470R that allows 1W at 3.674MHz to be used instead of 5W. This greatly reduces heating of the circuitry and also frequency drift. This change still allows up to 5W to be used and has been fully tested with a FT817 on 3.674MHz which produced +27dBm low1, +30.5dBm low2, +34dBm low3, +37dBm high.

8. The Transverter is not designed to be hard keyed from a Transceiver and doing so could instantly damage T1depending on how you connect the keying to the Transverter. A small delay with C40 1uF is required to slow down the Transverters TX to RX switching else a very high RF spike from the PA will destroy T1. The suggestion is to find an output on the back of the Transceiver that has +12vdc in TX mode, and feed it via a 1k resistor to the top of LK7. That way when the +12v is disconnected when going back to RX mode, there is a slow delay going back to RX mode due to the discharging of C40. If the Transceiver uses an open collector transistor output that grounds the collector in TX mode, then you will have to change C42 to a 1uF and connect the Transceiver via a 1k resistor to C42. None of these suggestions have been tested so it will be up to the individual if they want to try this.

MF 472kHz 630m Band Links: