2.3GHz ATV TX by Ian G6TVJ
Here is a design for a synthesised ATV transmitter covering the 13cm band and beyond up to about 2.6 GHz. This transmitter has been designed for optimum video performance and can achieve a broadcast quality signal when received with a professional TV microwave receiver. The design is based around a commercial voltage controlled oscillator (VCO) which forms the heart of the transmitter. This unit allows for the complete absence of RF alignment requirements and also aids in the construction and reproducibility of the design. The transmitter also makes use of a commercial power amplifier module made by Down East Microwave producing about 1.6W. The transmitter exciter could be operated on it's own if required and produces about 50 mW.
Click on the circuit image to see a larger version.
The microwave signal is initially generated by a VCO made by Z-comm. This company produces a wide range of VCO products and so the one chosen meets our requirements very well. The module tunes about 2.2-2.7 Ghz and provides an output of about +10 dBm. The unit can almost be considered a little TV transmitter in it's own right. The tuning voltage input spans about 1-10V and has very low capacitance and so excepts high frequency video modulation with ease. The VCO is very stable as it was originally designed for narrow band applications, it also simplifies the design of the transmitter considerably. A colleague of mine has built a miniature TV transmitter using one of these units and an amplifier which measures just a couple of inches square. The VCO feeds two destinations via a 3 dB attenuator and a resistive splitter. Z-comm recommend this approach as it improves the isolation of the VCO and prevents it from "pulling" with changing load. The splitter is made up of three 18R resistors and so provides further loss and hence isolation. One output of the splitter feeds a VNA 25 mmic amplifier which brings the level from about 0 dBm up to about +16dBm. The VNA 25 is made by Mini Circuits and is a "Very Nice Amplifier". These surface mount units operate on 5V, require no coupling capacitors and work from 500 Mhz to 2.5 Ghz, they are useful little gain blocks ideal for the 23 and 13cm bands. The VNA 25 is what can be considered the output part of the transmitter exciter. The second output of the splitter feeds a Plessey SP4982 devide by 8192 prescaler I.C. The I.C. provides the synthesiser with a signal devided down from the transmitter output frequency, producing value of about 280 Khz. The divided transmitter output frequency can be monitored at this point. The 4982 is a consumer device rated to about 2.5 Ghz but has worked successfully to about 2.6 Ghz with no problems. The device is driven with about a milliwatt which helps extend it's frequency range.
This transmitter uses the Motorola MC145151 cmos paralell synthisiser I.C. This device is extremely useful as it is programmed with parallel data and so does not require the use of micro-processors to communicate with it. It is my favourite I.C. and undoutably a classic, I have built endless synthesisers with it from HF to this frequency, it is also useful in ATV for use with sound subcarriers. I have fitted one of these chips to the Aztek transmitter in our own repeater GB3ZZ. Most other ATV transmitter designs use a Plessey SP5060 synthisiser, these devices can provide good results but in most cases the loop filter characturistics are inapropiate for FM TV modulation. Everyone follows the Plessey handbook which specifies a network which is intended for use with a frequency doubling mixer not a TV transmitter. The video low frequency response is ruined by the network and it can lead to rolling and tearing pictures, particularly in difficult circumstances like working through repeaters, working long distance or through long chains of video equipment. Only one transmitter breaks with tradition and is the only one which I could recommend. The G4WIM unit sold by Tim Forrester addresses this problem and uses a MC145151 synthesiser and a different filter network. The transmitter appeared in CQTV 165 in February 1994, I have one of these units and it gives excellent results.
The MC145151 can be tailored more exactly to the requirements of video modulation. Incidentally the 151 is about the same price as an SP5060 excluding the extra cost of the prescaler. The MC145151 further devides the transmitter signal down to a disriminator frequency, the amount of division inside the IC is programmed by switching of its pins, anything from a divide by 3 to 16383 is possible, this feature allows us to tune the transmitter in steps. A 5 Mhz crystal reference frequency is also divided down to the discriminator frequency by a fixed amount of 1024. The two divided frequencies are compared at the discriminator and an error signal is developed which is used to steer the VCO frequency right back up at 2.5 Ghz. For an output frequency of 2.330 Ghz (TV Simplex) the signal is divided by 8192 in the prescaler and divided a further 492 in the MC145151. The division ratio can be changed and programmed from dip switches or an eprom connected to the I.C. This transmitter uses two eproms one to drive the synthesiser and the other to drive an L.E.D channel display. If you do all the maths to cover the amateur band a significant bit changes in the middle of the band which cannot be accommodated with a simple BCD switch. BCD does not decode readily to drive two L.E.D displays so a second eprom is used here to code convert and produce a display of 1 to 16.
Due to the various division ratios the transmitter step size is equal to the reference frequency of 5Mhz. The least significant programming bit is not switched via the eprom so the step size increases to 10 Mhz. A -5Mhz switch could be fitted to the LSB.
The error signal from the MC145151 is filtered by an op-amp and a capacitor network. It is this network which effects the video signal in a number of ways. The discriminator frequency if present at the error voltage connection to the VCO will cause modulation to appear on the video signal and interfere with the transmitted picture. The transient response of the network must be slow in order that the system does not "See" the video modulation, interpret it as frequency drift and try to remove it from the transmitter output. This is what happens with many SP5060 designs and manifests itself as LF tilt, ramping or distorted field syncs and you know what that leads to! So how did I arrive at the component values? The Motorola handbook quotes a load of maths which I tried to understand once but with the lack of a decent explanation I found it easier to experiment. I have looked at some commercial designs to get a starting point. The behaviour of the synthisiser can be monitored in a number of ways:-
1. The VCO error voltage can be monitored on a scope, an excessive oscillation as the PLL comes into lock will indicate incorrect values. 2. A CW HF receiver can be used to listen to the "pre-scaled" signal as the PLL locks up to further gain the best characturistic. 3. The reaction of the synthisiser to the video signal content using special test signals will also indicate how things are performing.
Changes in average picture level will effect the error voltage. Also no LF video modulation should be present on the error signal.
Dual capacitance substitution boxes are very useful here, also ganged potentiometers can be used to ascertain fixed component values.
From cold the transmitter takes several seconds to lock up and will unlock briefly if the channel switch is moved quickly around the channels. The MC145151 has a lock indicator output, this drives an L.E.D, it is possible to use this output via some logic to mute the transmitter output if the PLL goes out of lock. The network shown operates great with the values stated, the output from the discriminator is a differential signal and this is combined in the op-amp. The critical components are in the feedback paths of the op-amp. Splitting the op-amp inputs and decoupling them to earth increases the filtering and so does adding an extra RC network at the output of the op-amp. The filtered error signal is combined with a standard pre-emphasised video signal using a video op-amp. The error signal sets the DC conditions of the amplifier and hence the VCO voltage. The video signal is fed to the inverting input via a small network to slightly correct the HF response as viewed from a broadcast receiver. This arrangement does in fact invert the video modulation, but it comes out OK as viewed via an "S"band LNB which uses a high side local oscillator. For positive modulation an additional video inverter should be placed at the input to the transmitter.
The transmitter boasts no AC coupling in the video path so will produce perfect LF results something often lacking in other ATV transmitters. It should be noted that the DC component of the video signal is still lost in transmission due to the average carrier frequency shifting with average picture level. The synthesiser will react slowly to the modulation keeping the average frequency in check. This is a characteristic of all FM TV transmitters and analogue Satellite uplink modulators and is not a fault. No attempt should be made to clamp the outgoing signal in the transmitter, video clamps are horrible things and should not be used in transmission chains. All sorts of strange diode circuits, networks and transistor junctions seem to crop up in other designs, only to succeed in stripping off the syncs and crushing the video. As I write another strange circuit has appeared in CQ-TV, a preculiar clamp circuit is used to interfere with the action of a synthisiser loop in order to establish a carrier frequency. The circuit is quite complicated and completely unessessary.
The power supply consists of regulators to supply 10V to power the VCO, 5V to run the synthesiser, VNA, Prescaler and channel display. A DC-DC converter is used to supply a negative 12V rail to the op-amps.
A diode in the filter op-amp negative supply prevents
the VCO error voltage from going excessively negative in the event of a fault
condition. A negative bias to the VCO could possibly damage it as there would
be a forward biased varactor diode inside it. A clamp diode at the video op-amp
output also protects against this condition. The video circuitry is simply a
CCIR 405 pre-emphasis network and a pot to set the deviation.
A commercial two stage power amplifier supplied by Down East Microwave is used to provide a transmitter output of about 1.6W. I found this unit while searching on the internet and it is available for about $200 and $15 postage. This two stage amplifier comes as a ready built module un-enclosed complete with a negative bias PCB and transmit relay. The output from the exciter is attenuated by 3dB to give the recommended drive level to the power amp and provide some isolation from the exciter. The unit is rated flat out at 2W but I couldn't get that out of it. To be fare it is actually rated from 2.3 to 3.4Ghz (To cover 9cm) so I am using it at the bottom of the band, it may "Hot up" as you go higher. The amplifier is very stable and no instability problems were encountered even though both the P.A. and the exciter are not completely screened from each other.
The transmitter is built in several parts. An SHF module supports the VCO, Prescaler, VNA and associated microwave circuitry. The devices are interconnected using 50R micro-strip lines etched onto ordinary double sided fibreglass board. All the SHF devices are of surface- mount construction. I have designed a double sided PCB layout for the SHF bits, the PCB mounts inside a tinplate box in a micro-wave committee style. All the other low frequency stuff can be mounted as convenient as long as due attention is paid to the usual grounding and decoupling. I build most circuitry using special prototyping sticky back pcb pads called "Wainwright Mini-Mount",these are extremely useful and allow very speedy development of circuitry by sticking them down onto a pcb ground plain. The eproms can be built on to vero board together with the two 4511 display driver I.C.s
Not actual size.
A little imagination is needed with the Power Amp, it can be supplied in a box but that costs another $100. I mounted it SHF side down using brackets screwed to two unused holes on it's SMA terminations. The aluminium body of the power amp ideally needs some extra heatsinking, by screwing a plate to it. ,p>
I have operated the transmitter on 2.330 Ghz and it has provided excellent P5 results kindly sent back to me on 23cm by Phil G1HIA. The transmitter is mounted in my loft space away from the shack to keep the feeder lengths to an absolute minimum. The P.A. is keyed remotely from the shack but to change frequency I will have to climb a ladder! A sound subcarrier was inserted down in the shack and fed up the video cable to the transmitter. With this arrangement a full duplex high quality QSO was possible. The transmitter can be tuned up into a frequency allocation used for broadcast video links. This feature has allowed me to test the transmitter into a broadcast receiver operating at 2.460 Ghz (Known as channel 1) via a path simulator. By passing various test signals through the transmitter the performance can be assessed by looking at the received output on a scope. Passing a 50Hz square wave through the TX revealled no LF distortion at all. A squarewave is not an alien video signal as it closely mimics an outdoor scene with half sky half land. The HF response was measured using a calibrated "Pulse & Bar" signal, the result was close to 100%. I am cheating a little in claiming broadcast performance from a home built transmitter. Getting or approaching a broadcast quality signal from a transmitter is not that difficult, the tricky stuff is in the receiver. In a receiver you have bandwidth defining filters, demodulator performance and group delay characteristics which all have to be compensated and corrected for. Getting a picture from A to B is easy, getting it from one end of the country to the other combined with decodeable teletext and sound-in-syncs is not so easy, hence the complexity of broadcast designs.
The transmitter is not particularly complicated and it demonstrates the possibilities for amateur equipment. If you want rolling tearing pictures then insert a 47uF capacitor in series with the video signal. May be amateur signals aren't amateur if they aint a little wobbly, but a noisy but undistorted video signal will lock a monitor at only P1! This design can be used at 23cm with no mechanical alterations to the PCB layout and would be ideal for where the best possible quality is required i.e. a TV repeater output.
If you have the appropriate licence then the transmitter could be used outside the amateur band.
|Chan Number||Frequency||Address||Data (Hex)||Chan Number||Frequency||Address||Data (Hex)|
|1||2.320 GHz||0||28||9||2.400 GHz||8||30|
|2||2.330 GHz||1||29||10||2.410 GHz||9||31|
|3||2.340 GHz||2||2A||11||2.420 GHz||A||32|
|4||2.350 GHz||3||2B||12||2.430 GHz||B||33|
|5||2.360 GHz||4||2C||13||2.440 GHz||C||34|
|6||2.370 GHz||5||2D||14||2.450 GHz||D||35|
|7||2.380 GHz||6||2E||15||2.460 GHz*||E||36|
|8||2.390 GHz||7||2F||16||2.500 GHz*||F||3A|
* Frequency not intended for Amateur Useage