Wellington VHF Group, Inc: Online.

Article submitted for publication by Simon, ZL1SWW. Our thanks for the contribution! Others with articles to share are welcome to contact us...

Well with the winter months slowly coming to an end, it’s the time to think about preparation for things we like to do in summer time. One thing is hill topping, going out to a high spot and playing around with the SHF bands.
I’ve been doing this for a little while now and have been building up a repertoire of microwave transverters with the aim of covering every band from 6m to 10 Gigs with the exception of 615MHz.
I’m nearly there and have now got the 9cm stuff up and running albeit a decent dish antenna.
Some people may be put off the microwave bands for the degree of difficulty involved in getting things going.
We in ZL have a spot of luck though. There are some bits and pieces around that take a lot of the hard work out of some of these bands. The items I am talking in particular are about the Stratex DXR series of converter bricks and to lesser extent PA modules.
Currently I have two of these units running, the 768 and the 710 variant. The 768 we have managed to bring down to 5760MHz with a bit of work, and the 710 for 10GHz, which basically needed nothing RF to do on it. What we needed to do was to make up a controller to drive the PLLs and control, relay sequencing and switching etc.
Additional to this is a 730 that will run in the 9cm band.
Tim ZL1TN went to a VHF convention (now a couple of years back) and managed to pick up one of the 768 units along with a PA. He made mention of it but it all went quiet for a bit. Some months later I then asked if I could get some photos of its innards to see what made it tick. Seeing inside got me all excited and told my other two mates Harry ZL1BK and Keith ZL1BQE about it. The digital pics were circulated by email to the other two, data sheets were pulled off the net for the things we didn’t already know. We all got on the simplex and basically had the whole thing worked out from the pictures. None of us, Keith Harry or myself, had any hardware yet.
We paid our money and got three units up from Wellington VHF group so we could all have a play.
This article outlines what a group of three of us have been doing here in Auckland to get these going. It is targeted toward the 768 model with converting it to 5760MHz, the other 2 units, 710 and 730 are very easy in comparison to get going as they can already run in our band area with little change. The 768 had been designed for 6.6 – 7.2GHz link use,

The team involves Harry ZL1BK, Keith ZL1BQE and myself Simon ZL1SWW. Each of us has areas of expertise in the field so we were tasked with self appointed jobs.. Some of these units have made it to the UK and Australia as well. Chris Bartram GW4DGU has made some good progress on these units as well.

Keith had already made up some code to drive the PLL (Phase Locked Loop). Some issues with getting the SPI data bits in order were had but soon solved with Keith’s SPI (Serial Peripheral Interface) to binary decoder, effectively a number of shift registers with LEDs on to see that the bits are in the right order. It’s easy to load the data in the wrong way around. We load MSB (Most Significant Bit) first.
SPI is used commonly these days as it saves on lots of parallel pins that used to be used on older PLL chips. Think of it as parallel data that has been converted into a serial stream of bits that go in one pin of the PLL chip. Each bit gets sent per clock toggle and is stored in a shift register within the PLL chip which effectively recreates the parallel data within the chip. The enable pin will have been held in a state during this load and the transition then loads the parallel bit pattern into the dividers within the chip plus several other setup modes needed for the PLL chip to work. It means that you can effectively send a lot of parallel data down one pin with two other control lines in the case of this PLL chip.
With the PLL chip supposedly doing the right thing, we went about making the VCO (Voltage Controlled Oscillator) run at 5616MHz, 5760 – 144 = 5616 for low side LO (Local Oscillator) injection.
Several nights were spent getting the blessed thing to lock, just couldn’t get the frequency low enough. Finally we got it to go. The oscillator is a negative resistance interdigital design. We had to find a way to effectively lengthen the tracks. The most stable way is shown in the picture below…

Click the image for a high-resolution version.

To keeps things short, we had to overcome several issues:-
• Phase noise was unacceptable for SSB
• LO bleed from up to down converter and vice versa was causing beating of the 2 PLLs
• Keeping LO spurs to a minimum as they are only 144MHz away from the main output we want.

The Local Oscillator

Not just one, but two are employed in this unit, one for the transmit mixer and the other for the receive mixer. They are in the form of a PLL that is tripled up to get the required frequency. The reference starts at 10 MHz.
Keith adapted a piece of code we were using on another 2Gig PLL project I was playing with, and using conditional assembly, we make a few tweaks in the assembly code and we can use the same PLL code for both types of PLL 2Gig or 5.7Gig. It just needs reassembly and reprogramming the chip.
The heart of the LO is based on a National Semiconductor LMX2326 PLL chip. It has an internal prescaler that is a divide by 32/33 and is dual modulus. The prescaler can handle up to 2.8GHz.
As mentioned above, the VCO runs at about 1872MHz when modded for 5616MHz LO. The VCO is multiplied up by three to get to the final frequency we need. This is done by a crossed diode limiter as it gives rich odd order harmonics. From there it is filtered and buffered along the way to the Hittite mixer that gets about +13dBm on its LO port.
In our design criteria, we decided that the micro should have a pin for selecting two frequencies. This pin is sampled on a processor power cycle.
The initial thought here was to have it so the micro will send 6.5Gig frequency command to the PLL. This is to allow a constructor to get the PLL up and running to check it works with just the chip connected to the unit in the way I will describe later. It should show the PLL lock LED blink briefly and then go out, you can do a finger check on the VCO to knock it out of lock to see the LED light.
Here is a basic layout of the PLL / Tripler chain as we see it. We have no documentation on these units so we have to make our own.

The TX side.
The units are very similar in the PLL, tripler, filtering and LO amp chain. The differences are in the TX and RX sections.
The TX side uses single mixer with a variable gain amp after it to alter power level. A bit of post mixer filtering is applied before hitting the amp chain consisting of 3 stages, the final being a small power FET. The radial stub on the first device that feeds the negative gate bias needs a touch of silver paint on it to stop an oscillation we have seen on the units we are working on. No doubt caused by the lower frequency of operation.
The PLL supply and the TX supply run off 10v. For -14dBm drive at 144MHz in the IF port gives a relatively clean +16dBm out at 5760MHz. The IF needs very little drive as there is a MMIC (Monolithic Microwave IC) at the IF input that has oodles of gain. I have actually thought of removing this device.

The RX side.
The receive down converter has a two stage FET low-noise amplifier, an image-reject
mixer using a pair of Hittite mixers, and an output amplifier.
Image reject switching is done by a control voltage in the SA630 (high frequency switch in effect) that selects the output from which mixer. This is pertinent when using high or low side injection. All the time we are using low side injection so we put a positive voltage on the control pin to get an overall increase in RX sensitivity.
The front end has a circulator with low insertion loss that works ok at this frequency, it offers a degree of protection to the front end FET as reflected signals get dumped in the dummy load on the third port of the circulator. Also when tested with a network analyser, it looks like 50 ohms when looking back into it. Some FCC regulations need to have the RX front end show as a 50 ohm input impedance. There is some IF amplification by another MMIC on the IF port.
Additionally the RX LO output maybe suppressed if the PLL goes out of lock as the last FET gets put in cut-off and will not pass enough signal for the mixer hence disabling the RX.

Our Home Brewed PLL Controller
The controller board is the same as in the 10 Gig unit, we were tied to this design due to having two 10Gig YIG oscillators that needed to be controlled. Having done the PCB artwork, all we had to do was the code change.
Keith ZL1BQE and Myself ZL1SWW developed a board to control both PLLs from one micro, a Motorola 68HC908qy4. The chip is an SMD unit that offers all the port pins we need.
The control board comprises of the following functions:
• PLL control / Sequencing Micro.
• VTCXO frequency fine adjust POT.
• Two LM317 voltage regulators for 10v supplies.
• Optional TX RF output control pot.

The PLLs are controlled by SPI data lines Clock, Data and Enable.
All the lines are paralleled up with the exception of the Enable lines of which there are two, each one feeding the TX and RX PLL chip respectively. Data is sent twice during a power up or a TX / RX transition..
On Power up the TX data is sent to the TXPLL and then the RX data is sent to the RX PLL. If you connect Lock Detect LEDs to each chip, you will see the delay as the PLLs load.
The SPI data is clocked through synchronously 24bits at a time but only 21 bits are used by the PLL for the load. The first part sets up the Charge pump and LD (Lock Detect) LED function, the other load sets up the reference division ratio. Once all data is loaded to the PLL chip, enable is toggled to transfer what is in the PLL shift register, into the relevant areas in the PLL chip. All this happens for each PLL chip in turn.
Within reason, the code allows the user to input any reasonable frequency span and it will faithfully push it to the PLL chip. Weather it will lock is a different story. Eg, you could put in 4Gig and the commands would be sent to the chip.
Serial comms happens on the same pin and hence we use a small adapter to change from RS-232 / V.24 levels to a 5v signalling. Pin 7 is the data in/out pin for serial comms.
Operation is simple, only 2 commands
‘R’ will return the programmed channels
‘W’ followed by 8 digits will program channel 1 with 1st 4 digits in MHz and
W60005616 ch1=6000MHz ch2=5616MHz in 6G version
The PLL runs at a third of the final frequency of the LO. The final stepping rate is at 2MHz so that means we set the reference division ratio to 666.66667 KHz so from a 10MHz reference we have a division ratio of 15.

Click the image for a high-resolution version.

Making a Common TCXO Reference on TX + RX Units.
This mod is needed for SSB as the TX / RX frequency should be the same for easy transceiving.
The PLL units on each converter have a “Fine Tune” adjust where the VTCXO (Voltage Controlled Temperature Controlled Crystal Oscillator) unit has a voltage trim point to bring it on frequency. (Remember, we are tripling so any discrepancy in the reference will be magnified 3 times at the final output frequency).
On one of the pins, there is a ref frequency out that can be effectively paralleled to the RX unit providing the following is done.
Remove the power to the VTCXO by pulling a Zero Ohm resistor that will remove the power.
Remove the 5 legged device (it’s a single inverter unit) near the VTCXO, this is a buffer from the VTCXO to the Reference pin.
Bridge the existing in and out pads with a resistor to back feed the clock signal.

Phase noise issues.
It has been found that using too higher division ratios causes the phase noise to be bad. Based on past generations / iterations, we have settled on 2MHz steps that results in a division ratio that gives a 666.666kHz reference at the PLL, multiplied by three effectively, gives us a 2 MHz step at 5.616GHz.

Beating PLLs
To fix this, we need to change the frequency so the two PLLs are not running on exactly the same frequency.

To do this, we use a 2MHz offset.
While the unit is in RX, the TX PLL will be the wanted LO frequency + 2MHz.
When in TX, the RX PLL will move up 2MHZ, so wanted LO + 2MHz.

Tuning Microwave Circuits.
Tuning the radial stubs that are prevalent in this system is done by the “find the hot spot” method where you run the unit up on the desired frequency. We then use a short length of tinned copper wire (TCW) of about 5 to 8mm long and jammed in the end of a cotton wool stick (without the cotton wool bud on of course), to find peaks in the output signal. At the hot point, we apply a bit of silver conductive paint, the sort you use for joining rear window demister track if they have been damaged. This stuff dries and then becomes conductive. It won’t be conductive when wet so let it dry. You can often tell if you have painted on enough by then using the cotton bud stick again, to see if the output still goes up or drops. If it drops, you may have too much and need to scrape a bit off by using a tooth pick to scrape off any excess to peak it up. Don’t scrape too hard as the board is Duroid and is quite soft and can be damaged. A bit of practice will get you mastering this technique.

The Final Outcome.
After several iterations of the whole kit and caboodle, I think we have come up with the best solution workable for us.
• We have a controller board that will drive all three units, 3.4, 5.7 and 10Gig. All that changes is the code in the micro and some extra pin jumpering for the 3.4GHz.
• It has inbuilt sequencer for the relays.
• Is accurate and on frequency every time.
• It is also frequency agile by presetting the IF frequency you want by using a simple terminal program such as HyperTerminal.
• All LO’s are low side injection so no worries about USB / LSB inversion.

The PA Unit.
If you are lucky enough, you may have a PA unit that mates up with the converter brick very well. These amps will put a reasonable output on 5Gig with anything from 3 up to 8 watts. The output strip line can be “flaked” with silver paint or a small tab of copper to better match the device at 5760.

The module runs off 10.5v and has several pins, 10.5v in, TX enable Forward power measurement and reverse power measurement.

First thing I do is to put a 10k pull-up resistor from the 10.5v supply to the TX enable pin as pulling high will keep the amp turned off so will not draw a lot of current in RX. To TX, we pull low with one of the outputs of the sequencer, in most cases, output 3.

With a simple pot divider arrangement on the FPM and RPM pins, we can easily get forward power and reverse power measurements that give a relative indication of antenna / feed line performance / matching.

For a clean output, it is highly recommended to get a K&L filter from the VHF group as well so that we are not wasting precious 5760 output in LO + output and other mixing products. Besides, it’s not polite to splatter the spectrum!!
These filters can be easily tuned down to 5760 by adding blobs of solder to the tops of the filter pins internal to the cavities and then reassembling and tuning. They are quite touchy and best done with a specan or wavemeter. The PLL can be set to 5760 and be used as a sig genny as there is enough LO leakage through the mixer and down the TX chain to tweak the filter. If you don’t have access to a high frequency spectrum analyser, there are people around who can help with filter tuning and general peaking of the unit.

Summary.
The description of the above would detail the hardest of the three units to do. The 3.4 and the 10GHz units are needing no RF tuning to make them go.
These units are a very cost effective and easy way to get on the microwave bands.
The Wellington VHF Group are also planning to release 10GHz and later, a number of the 3.4GHz units as availability and time permits. These units come with the original controller cards and would be in a going state.
Our development path we have taken was needed as we did not have controller boards available to us. There are currently 5GHz bricks available and many have already got them running and having fun times with them. Steve ZL1TPH opted to go with a VK5EME crystal controlled oscillator board and is using a Waikato VHF Group multiplier board to multiply up to 1123 and then the tripler is changed to run as a times 5 multiplier to get 5616MHz LO. This is yet another way of doing it with great results also, but not so frequency agile.
Several of these units are in use in the VHF contests with great results and being able to work well over a 320+ km path.

Click the image for a high-resolution version.

For those who are interested in more specific details, please visit http://www.qsl.net/zl1sww for precise details on PCBs, schematics, filters and so on. The site shows the evolutionary process of how it got running to the stage we are at now and explains the reasons for the design approach.
Think about giving it a go and enter the world of microwaves with relative ease. This especially applies to the 10GHz units as they are basically ready to go. Lets make the most of the bands we have!
Most of the hard work has been done and all that is needed is a bit of patience and time. There are a few of us around who are able to assist with any tweaking issues. Micros can be made available on request for a nominal fee.

73’s Simon ZL1SWW