Thursday, October 13, 2011

W1GHZ rover transverter for the 3400 MHz - EU band coverage

And finally, the 9cm band cheap rover W1GHZ transverter for the EU band coverage is also finished. As already mentioned in the 13cm transverter post, the band coverage with the original L.O. does not permit operation on the EU portion of the band, in this case 3400-3410 MHz. Some modifications were required and here we go, another easy and cheap way to start with the new band not spending the big bucks. Not yet in the proper housing but even without it no unwanted oscillations or products observed.

Looking from the left side there is a small power amplifier with the AH-102A and a peace of heatsink from the old PC switching power supply followed by the well known W1GHZ transverter board for the 9cm band with the 1/2" pipecap filters. As the transverter will be used at the Europe with the different 9cm band allocation the original 720 MHz local oscillator does not help in this case. Separate local oscillator, S53MV style, already used in the 13cm transverter will be used but with the different output frequency. At the end on the right side there is a so called "sequencer" taking care of switching power and everything that have to be switched in the transverter.

I will start from the oscillator this time. After the multiplying (5x) the original 720 MHz oscillator should give us the 3600 MHz signal on the transverter board after the pipecaps. Mixing with the 144 MHz (LSB) we should reach 3456 MHz. If we want to use this idea in the EU we should have a 200 MHz I.F. radio which is not convenient at all. Going back to the well proved practice, the S53MV oscillator used in the ZIF microwave radios was a cheap and flexible solution offering many multiplying schemes to reach required frequency. For the I.F. I choose the 70cm band where we have 10 MHz (from 430-440 MHz) comparing to the 2m band where only 2 MHz (144-146 MHz) is available for conversion. Looking through my crystal stock (this is where all your "odd" crystals comes handy) I found the one marked 20.585 MHz. After the chain of multipliers on the oscillator board the output frequency was 741 MHz. The exact frequency was tuned with the coil and the trimmer capacitor near the crystal. (The multiplier scheme was 20.58333 x 3 x 3 x 2 x 2) More that 10dBm of the stabile signal was enough for the next multiplier MMIC stage on the transverter board.

At this stage the transverter PCB was not populated completely, just the multiplier parts required and all pipe caps. To avoid an extra multiplier, as on the 13cm modified rover transverter, I decide to use the original pipe caps tuned to the 4th instead of original 5th harmonic frequency. Idea was to multiply oscillator 741 Mhz four times to 2964 Mhz. Mixing with the I.F. of 436 MHz we are reaching the 3400 MHz band. No need for the fancy measuring equipment, all can be done just with the "LNB diode power meter" if you previously measured correctly 741 MHz from your L.O. So how does it work? First, be sure to have all pipe caps with the screws completely inside touching the PCB inside the filter.10dBm at the input is enough to drive the ERA-2 (A4 marked first MMIC) in the saturation producing reach harmonics. Backing out the screw from the first filter and at the same time measuring the power just after the second ERA-2 (marked A5) you will notice just after the few turns (2 or 3) a peak in the power meter. This is obviously the 3rd harmonic at 2223 MHz as this pipe caps are too small to tune 2nd harmonic (1482 MHz). Backing out the screw will bring us to the next peak which should be 4th harmonic at 2964 MHz, the one we are looking for. So you can stop here, or if you want to see what this simple multiplier is capable of you can continue with the screw reaching the 5th harmonic at the 3705 MHz. I even reach the 6th harmonic at 4446 MHz if I remember well.

The next thing is to replace the measuring points and to connect the power meter at the L.O. test point after the second pipe cap. Tune the second pipe cap filter backing the screw to get maximum power. You can fine tune now the first filter to the maximum, but practically the MMIC isolation should be high enough preventing the influence of the filters to each other. At this point 7-10 dBm of clean L.O. signal is available, depending how patient with the screws you are. After tighten the screws I like to secure them with some extra nail varnish.
As you notice, I did make some bridge where the mixer should be installed allowing this way to tune also the filter after the mixer to the maximum at the L.O. frequency of 2964 MHz, of course measuring the power just after the filter. The I.L of the filter should not be higher than 2-3 dB so you should easy measure some 7 dbm just after the filter. Do not forget to remove the bridge :-)
Now it is time to populate the rest of the PCB and to complete the transverter. This time I did follow the project and I didn't swap the MMICs. The only thing that I change are the bias resistors. I don't like to have different RX and TX power supply so I convert everything to the 9V power supply, even the multiplier MMICs are 9V powered. Anyhow I think that the MGA-86576 has the odd value at the original design, not according the data specs.

After we have all parts on the board we can finally tune all together. With L.O. connected and rx side powered up it remain to bring the pipe cap filter after the mixer from previously tuned 2964 MHz to required 3400 MHz by backing the screw (several turns). Approaching the 3400 MHz the noise will also come out louder on the IF radio. The easy way to do that is just to listen any signal on the band :-). Of course, this is not 20mtrs band and it is not crowded, so it is handy to make some oscillator making the "noise" on the 9cm band. The easiest way is to use canned 40 MHz oscillator. The 85th harmonic should be heard with no problem!! Adding a LNB diode on the oscillator output pin, can even improve generating of the harmonics. I make the comb oscillator with the 16 DIL socket, so different canned oscillator can be used. 100 MHz oscillator can be handy for 9cm band as well. It is smart to add the 5V regulator on the sam board, just to stay on the safe side with all wires on the bench during the test and tuning faze.

Tuning the signal to the maximum, listening the radio is not the perfect way but can give us good idea how many turns we have to back the screw out from the pipe cap. Now we should connect the power meter to the TX SMA connector and apply some signal on the I.F. port. Not more then 1 mW on the 436 MHz is required to have 3400 MHz out. Do not forget to power the TX side of the transverter :-). Backing the screw from the pipe cap just before the ERA-5 should give us some output, bringing the screw at the same point as previous pipe cap. Fine tune the both pipe caps for the maximum output power. At the same time this will result the best receiving signal. I reach the 25 mW of the power on the 3400 MHz. Most probably, some more mW can be obtained with the careful tuning but I was lazy to play with it because I had in mind the next amplifying stage.

The RX side performed quite well, the sensitivity is just enough for the rover type transverter. TX part is stabile, no oscillations noticed. You can notice on the picture above that the screws head are just above the nuts, that's because when I tune the filter to the required frequency I like to cut excess length of the screw. Like that the complete transverter occupy less space and lower profile housing can be used. One important thing: before soldering the pipe caps, make the plan how you can do it easily. It is not the same which pipe cap will be soldered first. Try to visualize the job and you will see that there is a place just between the pipe caps that is very narrow and you can not access it easily. So try to make the best sequence and order in soldering the pipe caps to avoid this problem.
Using only 25 mW of the power can result in some qsos from the hill portable location, or the qsos with the big guns but at least 200 mW should be nice to have. Anyhow, half of that power we are losing trough the relays, connectors, coax, swr, coax to feed adapter etc.

Looking through my microwave surplus to find some quick and dirty amplifier bring me to the already proved design using the AH-102A MMIC. I knew that the data sheet said that this component can work up to 3 GHz, but it did not cost me nothing to try this MMIC on 3.4 GHz. The board with the MMIC was cut out from the old 2.1 GHz equipment together with bias network and input/output capacitors. First test did not bring my expectations to reach at least 200mW. On the 1.3 and 2.3 GHz this MMIC easily produce some 450 to 500 mW of power. Trying to match the input with the trimmer capacitor did not bring any improvement and the maximum of 150 mW out with 25mW drive was the result. Yeah, I did the tests with the 5V power supply, and going back to the data sheet I found that the power supply should be 9V!! This is why I change all the power supply in my transverter to 9V at the first, having in mind this amplifier. Applying 9V to the amplifier bring the smile on my face, 250 mW output or 10dB of the gain on 3.4 GHz. This was much closer to my expectations and no further tuning with the in/out SMD capacitors was done. Finally, a piece of  heat sink from the old PC-switching power supply was prepared with the MMIC board attached allowing long TX sessions.

This is the so called sequencer that I am using in most of my transverters. Just a few words about it, because this one deserve a separate posting, maybe in the future. On the PCB there is IF TX-RX switching with the attenuator to reduce the IF radio power to 1dBm. On this place I am using the Omron type relay (black). The blue one is used for the power (DC) switching. Obviously there is a difference in the inner design where the blue type has a very poor isolation between the contacts. The Omron type is probably designed more likely as coaxial relay where much better isolation was achieved. On the same board there is a RF sensing switch, PTT possibility and NE555 hold/timer. The same board is driving also the output coaxial relay and at the same time switching the DC power supply with many options.

As complete project was cheap and easy I decide to use the same approach for the antenna. Cheap WA5VJB 2-11 GHz PCB Log.per antenna was placed in the focus of the prime feed dish 90 cm diameter. I did not care about the ilumination efficiency for the moment, but I found later on the web that the hole arangement is not so bad. Anyway this is rover, cheap and easy. For the first test I just hookup the transverter board with the L.O. on the FT-817nd tuned to the 70cm band. At this step I just want to check the receiving part, so no coaxial relay was used. A few SMA to N jumpers and a peace (2m) of not so lossy LMR-400 was connected to the LPA. After the tuning and all tests the comercial coaxial relay with SMA connector was attached to the unit. Here is the result:

Despite the fact that nothing was screened, the frequency was quite stabile without noticeable drift. Screening just the L.O. will improve the stability much more. Just one active amplifier stage on the receiving side gave us also surprisingly loud signal, much louder than expected. To give you some idea what I was listening on the video, I can tell you  that the beacon was running not more than 500mW (probably less) into the omnidirectional waveguide slot antenna distanced 66 km away. Estimated 4 double slot antenna gain was 5dBi with (+/- 3dB) omnidirectional pattern H plane. With no clear line of sight and the path loss of the 200 dB the terrain slope was not good (what can be seen on the picture bellow) for the microwave experimenting but this is the only source of 9 cm signal in the area.

At the moment the transverter is using  90cm prime focus dish, the same one I am using for the 13cm activity with the same WA5VJB LP array. Counting the active stations in the region and the overall activity on the 9 cm band, most probably, the transverter will be boxed together with the 16dBi gain short-backfire antenna in the same w/tight housing creating a compact and portable solution for the rover activity. If you want to activate a new band or just to participate in the contest gaining your overall result this can be cheap and easy way to do it.

So, what's next? Rover 6cm transverter.......soon ready :-)

Sunday, October 2, 2011

RA18H1213G simple 1296 MHz amplifier

How to get simply more power on the 23cm band? After building (read spending time) or buying (read spending money) the 23cm transverter you end-up with the qrp equipment and after making the initial qso-s with the nearby (read 400-500km) stations the appetites are bigger and bigger. Most of the todays transverter are not exceeding the 27dbm of output power and some way of increasing the power is required. Using the old style technique with the chain of low gain transistors is not cheap and simple process. At the end the price of used trimmer capacitors is exceeding the price of semiconductors used in the project! Investing (read spending more money) in the Mitsubishi RA18H1213G power module can be the good value for the money. 100mW input power delivering 20w of output power for only 65$  is what you get (according to some already built projects).

The sample used in my project comes from old ATV transmitter with note written on the plastic bag: 50mW IN - 11W OUT. Module was in continuous 24/7 operation for a couple of years where output power dropped from 20W to 11W at the end. Removing the plastic cover, quick inspection using the magnifying glass showed that all bonding wires are looking OK, anyhow nothing can be done even if they are burned, chip with bonding wires is secured with some kind of hard transparent epoxy. So 11W or nothing .... Actually, the idea was to test this module despite the fact that many articles (clik link) speak about the oscillations that may occur and the way how to cure this problem. The amplifier was designed following the original data sheet using just the basic parts, nothing more - nothing less. 5V voltage regulator with blocking capacitors is also included with the output power indicator circuit at the end.

Old milled aluminum housing from the Comtek 900Mhz amplifier was a simple and easy to reuse for this project. Good RF shielding and grounding, adequate thermal contact with the hea tsink and a good quality SMA connectors was a plus. Already built SMA and power supply connectors dictate the component layout and the size of the PCB. For the PCB i choose the double side FR4 laminate 1mm thick to achieve tight connection with the central SMA input and output pins and 50 ohms PCB tracks. Input and output lines are 50ohm and there are enough grounding screws on the PCB preventing the self oscillation of the amplifier. Pay attention to the screws close to the SMA connectors, module input and output pins and grounding point for the capacitors that are blocking the power supply pins.

First test with 50mW drive gave 11W out. Further increase of drive power did not gave more power out, even 100mW of drive did not change nothing. The amplifier is very stable and no oscillation problem noted. The amplifier was than tested changing the drive power and power supply voltage to invoke some oscillations but the amplifier was always stable. With the 9V of power supply the output power was 7 watts, not bad at all. Probably, more power can be squeezed out with proper input and output circuit match but I was lazy to do that. Last test was done also with the parts in place that are measuring the output power. This power indicator can be very useful for monitoring the output where no extra SWR meter or power meter is required. The amplifier was tested in tandem with the latest AD6IW prototype low noise 23cm transverter with excellent results during the July 2011 uW activity contest from the portable location.

And yes, the latest AD6IW 23cm transverter is the state of the art ......

Thursday, August 18, 2011

FOR SALE!!! 5.7GHz 120W amplifier "For Big Boys" ====== S O L D!!! ======

First of all - WYSIWYG, this unit is for SALE        SOLD!!!

This is the commercial V-sat 120 watts 5700-5900 MHz amplifier, CODAN manufacturer used in the v-sat equipment in the 24/7 regime. This unit comes without the waveguide combiner and no datasheet nor documentation attached.

What is inside the aluminum box housing? There is a small semiconductor S36B driving the TIM5964-4 driving one TIM5964-16. Then, TIM5964-16 is driving four TIM5964-8A where each one is driving the final TIM5964-35SLA. At the end we have four 30-35 watts outputs. Following semiconductors are inside the amplifier:

TIM 5964-4           1pc
TIM 5964-8A        4pcs
TIM 5964-16         1pc
TIM 5964-35SLA  4pcs

Output part

There was no buyer for the previous one, so it was cannibalized and sold in parts. Most of the components are sold separately where good feeback received with the success in constructing the separate stand alone amplifiers. What is left from the previous one, still available:

Contact me on the mail adam9a4qv x for the info.

Sunday, May 22, 2011

24 GHz transverter (conquering the last cm band)

Be prepared :-) 'cause this is the project where you will use more a screwdriver than the soldering iron !!!

Going higher in the microwave bands means spending more money and making less qso's. Reaching the 24 GHz band does not have to be so expensive like it seems. Spending 400$ on the various parts can be enough for a simple but effective transverter bringing us to the last amateur centimeter band. The transverter presented here is the result of smart planning and cheap building. Most of the critical parts are available on the e-bay and the rest of the electronics should be constructed by the builder. I have to thanks to Goran, AD6IW who generously send me a measured and controlled main RF parts with the idea how to use it most efficiently.

Block diagram is showing the "guide line" and the idea how to assemble a simple transverter. The transverter was built around the down/up converters as a separate blocks. The 24 GHz sides of converters are already prepared for the WR-42 waveguide flange connection and the good practice on this bands is to avoid unnecessary waveguide/coax transitions, SMA connectors, semi-rigid cables and all other lousy parts. Instead of lousy coaxial relay for the antenna RX/TX switching I use the 3 port WR-42 waveguide circulator. No need for fancy smart sequencers and 24 volts DC relay supply. The converters are connected directly to the waveguide circulator. The third port is connected through the short peace of flexible waveguide to the antenna.

The YIG oscillator should work on half or even one third of required local oscillator frequency because of the fact that multipliers are already included in the converter units. The signal from the YIG oscillator is supplied to the converters through the 3 dB splitter.

According to the available data, both IF converter ports are covering the range from 400 MHz to 4 GHz, so wide range of IF frequencies are available. I choose 438 MHz and 1298 MHz for the IF due to easy tuning to the "round" frequency. All parts from the picture down below are commercial radio parts and we do not require expensive measuring equipment to tune and check the finished transverter. All job can be done using the "LNB diode" power indicator/meter and a frequency meter or a surplus wave-meter.

This is the "heart" of the transverter. Both, up and down converters are connected to the circulator (Ports 1 & 2) to obtain the best isolation and the short peace of the wave guide is connected to the Port 3 where antenna should be connected. L.O. and I.F. SMA connectors are protected with the plastic caps. Separate I.F. and L.O. ports are available, so there is a possibility to use different up/down bands for the base radios.

Down-converter is really small but with excellent characteristics. The noise figure is around 3dB and the conversion gain is 22dB. As I already mention the IF is covering the wide range from 400 MHz up to 4 GHz. Very convenient is that the down-converter is requiring LO/2 or LO/3 for a proper operation and the level of the signal should be around 10 dbm. As the LO multiplier is inside the converter we can use any 7/8 GHz or 11/12 GHz oscillator depending on the IF we want to use. The converter require positive 6.5 volts and the negative -5 volts supply. The RF antenna port is accessed through the standard wave guide connection WR-42 size. The inside L/4 pin is probably bent and look odd. Do not try to touch or correct the position of the pin because the pin is matched for the best RL already. The IF and the LO ports are using standard SMA connectors.

The up-converter is using the same approach to reach the 24 GHz band. The local oscillator should work on LO/2 or LO/3 frequency giving out 8 db of the signal. This will be enough power for internal multiplier to produce the signal for the mixer where 0 dbm of IF signal is required. The IF is again covering the range from 400 MHz up to 4 GHz. The up-converter sample that I have from the TRW miliwave company has the gain of 28 db with the IP3 of  30 dbm. The output power is from 23.7 -  26.5 dBm. Consumption is 617mA / 12 V dc power supply. The output RF port has the standard WR-42 waveguide flange connection ready for WR-42 circulator. Again, do not try to touch or modify L/4 waveguide inner pin even it is bended. It looks strange, but this is done by the factory matching the unit to give us the best results. The I.F. and L.O. ports are using the standard SMA female connectors. Recently, I saw a couple of similar units on the e-bay that are advertised together with the isolator ready for WR-42, with some lower output power than mine sample, but with the very good price. The seller advertise that the unit can be used as the beacon, if no IF signal is supplied the module is working as a simple multiplier. Depending of the applied L.O. frequency, a simple 24 GHz beacon can be built. Very useful is also the DET pin (detector) to monitor the output power simply by connecting the voltmeter.

What I can see from some already built transverters and written articles, one on the problem was the antenna switching. Some coaxial relays are rated up to 24 GHz but the loss are high. Some authors are using manual waveguide switches while the others are using mechanical parts with the electrical motors for the same purpose. There are also low loss waveguide switches, mostly home-brew like the one from the I3OPW but they are bit expensive. If we plan to use a low output power a simple solution can be 3 port circulator. The one I am using have the isolation of at least 25 dB and loss of 0.3 dB. This can be enough to protect the RX downconverter from burning the front-end. As the circulator is  WR-42 flange ready it is easy to connect the converters straight without any adapter. Using the circulator will save us from using the sequencer and 24V DC required for commercial grade coaxial relays.

The up and down converters are requiring the the same level of the signal from the local oscillator. The easiest way to do that is by using a simple 2 port power divider. As the L.O. is working from 11-12 GHz  and the  appropriate divider can be purchased from the e-bay or can be found on the flea market. 3 dB loss will require the 11 dBm of the L.O. output power to reach at least 8 dB of the signal on each output port.

As already mentioned, the converters are requiring the LO/2 or LO/3 to work properly. There are several ways of doing it, and after comparing some market prices, I choose to use the YIG PLL oscillator working on the LO/2 frequency. This may look quite expensive, but it is NOT!! It is very strait and simple and no tuning is required comparing to the endless multiplier chains used in some oscillators. Making the L.O. in old fashioned way will require the good quality X-tal (30 USD), some filters (more USD), MMICS (even more USD) and some good quality laminate. Of course, the housing, heater for the oscillator and some instruments and time to tune properly all the multiplier chains. Of course, at the end the frequency stability will always be questionable comparing to the commercially available products. Looking through the e-bay where many different types fo YIG are available, I purchase the Verticom MTS1500e-151-01 YIG PLL oscillator (from the picture) covering the range from 11.2-12 GHz. E-bay price 75 USD.

This type of YIG PLL oscillator require the SPI control. The job has been already done by the group ( Dave Robinson, WW2R, G4FRE ) in their project where simple PIC micro controller was developed for the SPI control. All you need is PIC 12F675 loaded with the proper code, a few voltage regulators and you are ready to have frequency stable 10dBm signal out from the YIG oscillator. MTS1500e-151-01 is not covering wide frequency range but enough to use IF of 432 MHz or 1296 MHz. The advantage is the lower phase noise comparing to the wide frequency coverage YIG oscillators. This YIG has the internal oscillator and the reference signal of 26.25 MHz  with the step of 0.416667 MHz. To make the calculation easy we have already prepared excel sheet with all formulas where code-words were created entering only the YIG model and the required frequency. Nice feature is that the microcontroller is capable of creating two different frequencies so I use this feature to generate the L.O. signals for 432 MHz or 1296 MHz IF radio. As the PLL step is odd number not every "round" IF frequency is available but my radio is covering both, 438 MHz for the L.O. output 11805 MHz and the 1298 MHz for the L.O output 11375 MHz. Quickly, using just a simple switch any frequency can be selected for the operation.

The controller was assembled on the small pcb together with the 5v regulator for the 12F675, while the power required for the YIG oscillator is obtained through the voltage regulators 7808 and 7812 mounted on the small heatsink. At the end, complete unit is requiring only 13-15V power supply. The small LED on the PCB is going on only when there is no PLL lock, just briefly at the power-up. Simple way of checking the YIG is using the old wave meter and a simple power meter/indicator. A minute after powering the YIG , stable frequency dip on the wave meter can be found indicating the L.O. output frequency. Switching of and resetting the power back, the time required to reach the same frequency can be measured giving us the idea how much the YIG oscillator need to stabilise on the correct frequency. In my case, after 30 seconds the YIG was ready for narrow band operation. This time is required for heating the reference oscillator inside the YIG oscillator. Connect the SMA output using the low loss semi-rigid coaxial cable with the power divider and further to the converters and the first test can be done even without antenna. Actually, the open waveguide will give us some 6dB of the gain.

Of course, some kind of antenna should be used instead of just a waveguide. Building the microwave antenna that will work on this band is not an easy task. It may look simple and straithforward from one point of view but huge precision is required to meet the low SWR. Simple horn antenna can be a good start, or penny feed with the parabolic reflector is also good choice for the homebrewer's but at the end you will require some kind of measurement to prove that you antenna is actually working.
I prefer to use commercially fabricated antenna for the 24GHz band where no tuning is required. This option can cost you a lot of money for a small antenna if you are buying new one. Good idea is to check the flea market where old cell link equipment can be found. The best and cheaper source is to "Know a guy who knows a guy" :-) Not so many people use this bands today and sometimes this "pots" are available for the beer or two. Just connect the wave guide to the original flange and you are ready to conquer the last microwave centimeter band.

Do not even think to "tune" the feed for the best signal. If antenna is not damaged, it will work more than good on our ham frequencies.

That's all, hope to CU on the 24GHz.........

Monday, April 4, 2011

Going up to 13cm W1GHZ rover modified transverter

The first thing that discourages builders around the Europe from building the W1GHZ transverters is the different band plan on the 13cm and 9cm. Moreover some countries can not even operate on the 9cm band. Some counties, like Croatia can use the both portion of the band (2304 & 2320 up to 2450 MHz) but all activity is concentrated around the 2320 MHz. Using the originally designed oscillator is not of much help. The problem can be solved using the lately wide spread oscillators based on the Si4133 synthesizer with direct  L.O. frequency for the 23cm and 13cm bands. The oscillator can generate the frequencies between 400 and 1900 MHz what he is OK for the 432 MHz I.F. on the 13cm band. Of course, this is no more simple and cheap project, but just an option for the builders. The one I have is coming from Australia, offering 16 different pre programmed frequencies, with possibility to reprogram any required frequency in the mentioned range with the step of 1 MHz. DEMI is offering their version (ApolLo) of the oscillator based on the same synthesizer.

As I plan to use this synthesizer for another 6cm transverter project, the same was replaced with the old fashioned L.O. based on the crystal oscillator with the chain of multipliers. Already proved S53MV oscillator design, the same one in use in the 23cm simple beacon, was assembled also for this project. After choosing the I.F. of 432 MHz, the required L.O. frequency was 1888 MHz. Approach was simple, first generating the signal of 629.333 MHz and than multiplying the same to 1888 MHz. The oscillator multiplying factor is 36 (3 x 3 x 2 x 2) with the crystal oscillating on the 17.481 MHz. This is where all your odd frequency crystals  ripped out from the old service mobile radios are coming handy and finally useful. If you have an inventory list of all your crystals you will have in no time the right one ready. If you do not have the list, you will have in no time your fingers dirty digging out for the right one :-) .

The final VXO oscillator frequency can be tuned quite wide, so the crystal frequency does not have to be exactly as the required one, making easier to look for the right one. After a minute or two the frequency is very stable despite the the big multiplier factor. The same principle and oscillator was used in many projects. The first 720 MHz oscillator for the 23cm rover transverter was the same one. Multiplying the 20 MHz crystal from the old hard disc was easy. Multiplying the 18 MHz computer grade crystal was used in the 23cm simple beacon project and  the next 13cm rover transverter will have the same oscillator.

Driving directly the transverter board with the 629 MHz seems to be a nice idea, but I was not sure that 1" pipe-cap filter can tune efficinetly down to 1888 MHz. Some experimenting and tuning was required on the pipe-cap filter and for that reason I decide to use another (no tune) multiplier. The S53MV Beacon 99 script is full of designs that can be used for such multiplier. I decide to use the oscillator multiplier from the ZIF SSB transceiver for the 9cm band (3400 MHz). The same article is also showing the oscillator that I am using for most of my simple projects.I knew that the filter in the multiplier chain is designed for the 1700 MHz, but I expect not so sharp filter response due to the design. The original design suffer a minor modification on the first stage where ATF35376 was replaced with the simple SNA-386 mmic from Sirenza. I just don't like this HEMT devices.Of course, the bias network was changed as well and the gate resistor was removed also. Idea was to saturate the MMIC to get enough harmonics, filter the 3rd one and amplify the 1888 MHz signal with the BFP420 transistor. I drive the SNA-386 with the 6dBm from the 629 MHz oscillator. At the end I was happy with the 14dBm of the signal on the 1888 MHz.

The multiplier was assembled on the 0.8mm FR4 laminate where no tuning was required. After the initial tests the SMA attenuator was replaced with the SMD resistor network just before the SNA-386 on the PCB. Vias required for the MMIC and BFG420 grounding are replaced with the grounding done by drilling the 2.5 mm holes filled with the melted solder wire. The ground side of the PCB is covered with a peace of thin copper foil soldered for the PCB creating a good connection with the pins.

Once I had the oscillator and multiplier ready and running properly the rest was easy. As I ordered a few rover PCBs it was just about following the Paul's procedure. There was no need to solder the original multiplier part with two pipe-caps just the part from the mixer and the front end. First, two 1" pipe-caps were soldered using the flat iron technique and they are separately tuned to 2320 MHz using the procedure explained in the 13cm Pipe-cap filter post. Soldering the pipe-cap may be tricky if you are using just a low power soldering iron. On the other side if you are using high power iron, the job may look ugly and the PCB can be damaged as well. I prefer to preheat the pipe caps on the flat iron and to position one by one using the tweezers on the marked place on the PCB. Than you can use your standard soldering iron to properly solder the pipe-cap for the ground side of the board. Soldering the RX and the TX front was quick and the only thing you have to care is the position of the MMIC blocking capacitors. I was short of MGA-86567 and instead I install the MGA-86563, the first I found in my drawer with the similar characteristics. The PCB is just enough small to fit SOT-363 (SC-70) surface mount package.

Connecting the L.O. and a FT-817d with the transverter board with a peace of semi-rigid coax was enough for the first RX test. Everything was looking OK and the MMIC current was by the specs. I checked the rx part tuning the filter pipe-cap to the maximum signal transmitted using the alias frequency from my  AD9951 DDS oscillator. To check the TX side i just insert the 25dB attenuator before the mixer I.F. port and the TX was pushing out 40 mW. It seams that Sirenza SNA-586 was running by the specs. delivering the max. output P1dB power. Initial tuning with the antenna was with the transverter assembled in the old PC switching power supply.

Just a small power amplifier using the AH-102A MMIC delivering almost 27 dBm of power is hiding bellow the orange heat-sink This will help to check the TX part running some more power to the antenna. This setup can not be used outdoor, so another w/tight housing was prepared with more room to accommodate a bigger power amplifier. The same one can be mounted on the mast, just behind or close to the antenna to reduce the loss of the signal in the coaxial cable. Looking from the top, there is a 629 MHz oscillator followed by the x3 multiplier stage. Below the multiplier there is well known transverter board. From the TX SMA connector on the left side the semi-rigid cable is connected to the power amplifier while from the RX SMA connector the semi-rigid is connecting the commercial 13cm 25dB preamplifier. At the bottom the coaxial relay is connected to the amplifiers and the antenna connector. Dummy proof sequencer is positioned on the left side of the housing assuring the carrier or PTT operating mode.

The transverter was tested from the home and the portable location. With only 450mW and 90cm dish antenna with the WA5VJB log. per. feed the contacts up to 250 km were established during the winter time. The summer will bring more activity, better propagation and some ducting as well. The video is showing the 200 km qso on 2320.100 MHz with the Italian station. 13cm band was used as a talk-back frequency trying to establish the contact on the 10 GHz.

Job remained to be done before the summer microwave season is to replace the noisy commercial preamplifier with the latest AD6IW "state of the art" uW LNA followed by the 13cm bandpass filter. The left side of the w/tight housing is leaving enough space for the 6 watts power amplifier. After that the transverter will be ready some DX-ing on the 13cm band. Latest AD6IW uW LNA ready to be inserted:

 The same principle will be used for the 9cm W1GHZ rover transverter. Oscillator is almost ready and the original multiplier on the transverter board will be used, avoiding the extra multiplier.

Saturday, March 12, 2011

Yet another W1GHZ rover transverter modification

After several successful modification it was necessary to unite all in one project. The transverter was tailored according to the parts that are found in the drawer, and some improvements are possible indeed. The PCB is a copy of the original W1GHZ project with a few minor modifications. As seen in the diagram, another type of mixer (RMS-30) was used. The MMIC on the RX side was replaced with the INA-02186 and lately with the MAR-6. The output MMIC ERA-5 was replaced with SNA-586 and the resistor combiner/splitter was replaced with the PIN diodes.

The first thing you notice is that the PCB is almost half size because there is no L.O. multiplier included. Instead of that there is just a SMA L.O. port connector so IF of 144 or 432 MHz can be used with the appropriate external local oscillator ( 1152 MHz or 864 MHz). The RMS-30 mixer has a slightly different pin-out so small modification was required also here. The top PCB TX side remained untouched, together with the part where PIN diodes are inserted. The PCB RX side is slightly changed, now allowing 2 MMIC on the RX side if required.

The PCB bottom side is mainly the ground layer. The PIN diode modification was done mainly on the bottom side. The splitter resistors are replaced with the diodes and a holes for the wires are drilled through the board connecting both PCB sides. After that at the bottom side, the pads are cut with the sharp X-Acto knife creating the soldering pads for the PIN diodes network (resistors, coils and the capacitors). The outcome  is higher isolation between the ports and lower loss comparing to the resistor splitter. The PCB tracks are then extended to the RX/TX power supply pads allowing the switching using the same voltage.

The transverter was ready for the test. Connecting the external L.O. ( 864 MHz ) and the IF radio ( 432 ) Mhz the RX part was tested first. Immediately, there was a high noise on the RX side indicating that the RX part is oscillating. Adding a few extra via to the MMIC ground pads helped a little bit but under the some conditions the INA-02186 was oscillating again. The known problem was not easy to solve, the 1.6 mm FR4 was too thick for INA-02186. At the end the RX MMIC was replaced with the MAR-6 with surprisingly good result.

It's true that MAR-6 can give us less gain and higher NF comparing with the INA-02186 but the transverter is very stable, not showing any sign of self oscillations under various conditions. Housed or not, with or without the extra LNA on the input, with the high input signal the MAR-6 was a simple solution for the home-brew PCB with via done by simple wire. The TX side of the transverter performed very well after the first test. All the MMIC voltages were according to specs. and no sign of oscillations were noticed. The output power is 40 mW.

The transverter was tested with the IF drive of 1 mW on the 432 MHz from the FT-817nd through the 50 ohms attenuator and with the 6 mW of L.O. power on the 864 MHz. The signal was clean and stable. Both the transverter and the L.O. were housed in the box made from the scraped nescaffee can metal sheet. The final version can be built without the numerous SMA connectors using just the semi-rigid teflon coaxial cable for connecting all parts of the transverter. Using so many SMA connectors can be costly and some builders don't have the access to e-bay or any other possibility to buy the connectors. At the end, the idea was to build simple and cheap rover transverter. 
On the other side, the modular approach will allow us to build and test all units separately avoiding the headache when trying to resolve the problem on the finished not working transverter what can be very important for inexperienced builders. More over, this approach give us the possibility to experiment with the various L.O., LNA, final power amplifiers, changing the set up very quickly and choosing the right one.

Various test were performed with the transverter to assure that all is working properly. On this video you can see the RX part receiving the beacons ( lower signal are beacons 100 and 250 km away). The strong signal is the 1 watt local beacon running on the 50 ohm dummy load located in the next room not more than 10 meters away. The noise appearing every 10 seconds is the AN/FPS-117 radar running some 30 km from the location with the clear LOS. Inserting the sharp response 23 cm filter before the front-end did not improve the situation because this is the in-band disturbance. The RX MMIC was just a MAR-6 so don't worry to much especially if you plan to use extra LNA on the RX side. The antenna was just a simple 25 el. loop antenna.

The second video is showing the same setup but with the YU1AW LNA using the BFG-540x as active component. Even not screened the LNA and the transverter performed excellent showing no sign of instability. Adding the LNA I improve the overall NF and a total conversion gain. The same beacons signals were checked. It is not easy to notice the difference on the video due to the local QSB but improvement is obvious.

Presented modular approach give us the possibility to use different L.O. with the different I.F. radios to reach the 23cm band. 144 or 432 MHz or some other I.F. depending on the radio you have requiring the proper L.O frequency. For the L.O. we can use the PLL, DDS, or some other way of generating the signal, today quite popular Si4133 with the possibility to program any frequency from 400-1900 Mhz in 1 MHz steps. Bearing in mind the simplicity of the rover transverter project I choose the old fashioned crystal multiplying chain starting with the 96 MHZ crystal. After multiplying, the final frequency was 864 MHz. This frequency constrain us to use 432 MHz for the I.F. which is simple but not the best solution considering the well known mixing products problem when using 432 MHz for the 1296 MHz I.F. Anyhow this is more acceptable than running the 1440 high side injection with the reverse 144 MHz LSB tuning. Most of the radios are blocked and can not even tune below the 144 MHz and other suffer with the lover sensitivity out of the band. The good thing is that the filter response is also better if we use the 432 MHZ I.F.
The simple and good oscillator chain is basically S53MV 23/13 cm transverter oscillator with the simple modification. Instead of the original 576 MHz oscillator following the 96-288-576 MHz chain I tune just the last multiplier to achieve the 96-288-864 MHz chain. The signal was clean and stable, power 6-7 mW. If you prefere to use 144 MHz (right side up) for the I.F. than the same 576 MHz oscillator can be used adding another multiplier stage to reach 1152 MHz.

At the end just a brief report regarding the cheap relay that Paul is using with the rover transverters for the antenna switching. During the last year Friedrichshafen HAM flea market I get the surplus aluminum milled box with 5 SMA connectors, several MMIC, smd directional couplers and other handy parts including 2 peace of the OMRON G6Y-1 relays, all that for just a few Euro. The unit was part of 900 MHz GSM equipment. It is true that this relay was the 5 volts type but at the end this was the advantage because we are using 8 or 9 volts for our transverter RX/TX stages.

Relay, 78L05 voltage regulator, 2 smd blocking capacitors, 1N4006 protecting diode, a peace of 0.8mm FR4 and 3 SMA connectors will give us the coaxial relay able to handle 10 watts of the RF power (sequenced) on the frequency of 1 GHz with the I.L. of 0.5 dB. Not bad at all! The double side 0.8 mm FR4 PCB was tailored where 50 ohms tracks are cut with the X-Acto knife. SMD blocking capacitors are soldered on the bottom side just close to 78L05 pins, together with the relay protecting 1N4006 diode. Even lower I.L. can be achieved soldering the coax directly to the PCB without using the SMA connectors.

So, that's it ....