XTAL-CONTROLLED FM TX (V6)
by Harry Lythall - SM0VPO

Introduction

My FM Wireless microphone V5 has been very popular, both as a beginners project and as a kit. It is interesting that it has also been the biggest cause of lots of e-mail, the most common question being How can I make it crystal controlled?. The oscillator can be made crystal controlled; replace the 2p7 with a 5th harmonic crystal. The oscillator will then be very stable, so stable in fact, that it cannot even be modulated! This means that you could only use it to send CW (morse code).

One solution to the problem would be to use a low-frequency crystal oscillator and modulate it. Since a crystal can be modulated by typically only 1KHz or 2KHz, then the multiplication factor would have to be around 72. This would require a 1.3888MHz crystal and several multiplier stages, each individually tuned. The first stages would also have to be double-tuned if the transmitter were to have low spruious signals; at total of eight tuned stages (plus output filter)! For example:

StageOutput Freq.Stage
Oscillator1.388888MHzsingle tuned stage
Multiply x34.166667MHzdouble tuned stage
Multiply x312.50000MHzdouble tuned stage
Multiply x225.00000MHzsingle tuned stage
Multiply x250.00000MHzsingle tuned stage
Multiply x2100.0000MHzsingle tuned stage
Output filter100.0000MHzdouble tuned stage

This is clearly impractical for a wide-band FM transmitter, so my solution to this problem is to use a Varicap diode to modulate an oscillator, then lock the oscillator in a Phase Locked Loop, and compare it to a crystal oscillator. That is exactly what this project is all about. The result is relatively simple to build and with only one adjustment it is ideal for the absolute beginner. But before I continue, let me just make one thing clear. With Wide-Band deviation, the VCO is never locked in phase, only in frequency.

You will be able to find the PCB foil pattern and component overlay drawings for this at my site download section. Please give me a little time to put it there before writing to tell me you cannot find it. If you cannot find it in a week or so then you may complain to prime.minister@10downingstreet.com. Everything is always his fault!

Circuit Basics

The circuit follows that of a simple textbook synthesiser comprising a Voltage Controlled Oscillator (VCO), a damped Loop Filter, a Reference Oscillator (Crystal) and a Frequency Comparator.

The only difference is that I have added a frequency divider chip to divide the VCO frequency by 64. This means that if the VCO operates at 100MHz, the output from the divider will be 1.5625MHz. If the crystal oscillator is also 1.5625MHz then the loop will be "in-lock." The control voltage from the filter to the VCO steers the frequency of the VCO so that the output of the divider is ALWAYS 1.5625MHz. Any deviation from this will result in a change of the loop voltage to move the VCO back to 100MHz.

Audio frequencies are then added to the loop voltage that control the VCO frequency. It is in this way the synthesised transmitter is modulated.

Circuit Specific

I will not delve too deeply into the in's and out's of synthesisers. I have already written several pages of information about them and the stages that are needed to make a working synthesiser. If you need more information to understand this circuit then visit my Synthesiser basics page for a better understanding.

The specific circuit is shown above. The "Prescaler" (divider) chip needs to have a supply voltage of only 5v (+/- 0.25v) so the LM317 has been included. I used the LM317T due to it's larger can size (and I have got a lot of them) so it will tolerate a supply voltage of greater than 13.8vDC without burning. All has been somewhat over-engineered.

IC1 (CD4001) is the crystal oscillator with an extra gate used as nothing more than a buffer stage. This feeds the frequency comparator of a CD4046 (IC2). The comparator output is filtered with a "slack- handfull" of resistors and caps to feed the VCO; a BC547. A second BC547 has been used to isolate the VCO from the antenna. Without this device the loop would have the tendency to jump out of lock if you touched the antenna. The VCO is also coupled to the divider, IC3, which can be any one of a selection of chips. MB501, SA701, SP8704 and CA12022 are all the same device.

The filter time-constant is a couple of second or so. This makes it take about one second for the loop to stabilise. If this were not the case then the modulating frequency would be seen as a frequency error and the loop would correct the error (remove the modulation). The modulation input is via a 100K resistor and 3n3 capacitor which provides the little pre-emphasis needed for an FM broadcast transmitter. If your AF input comes from a stereo encoder then remove the 3n3 since the pre-emphasis must occur BEFORE encoding. I hope to post a stereo encoder soon, but I give no promisses.

If you are one of those who likes to dissmantle circuits, then you may notice that I have used the CD4046 Signal and Reference the wrong way around - I have used the signal input for the Reference frequency and the Reference input for the signal frequency. This is because the Reference input is designed to be fed from an external source and so it will respond to small small signals (ca: 200mV) whereas the Signal input is designed to be fed from the CD4046s own CMOS logic level oscillator. IC3 is an ECL device with only 1v output signal and is therefore NOT CMOS compatible. The result of this is that the output sense of the frequency comparator is reversed! That is why the Varicap Diode (BB105) is reference to +5v and NOt to Gnd.

Frequency Programming

Not shown on the circuit is the PCB arangements around IC3 pin 6. This pin can be connected to either +5v or Gnd to select a divide rate of 64 or 65 respectively. This allows one crystal to give you two different frequencies. The PCB has a solder-blob patch that is pre-shorted to select divide by 64.

With a 1.5625MHz crystal the transmitter will operate at 100MHz using the 64 setting. If you then cut the track between the pads and place a solder-blob between the other pads then the TX frequency will become 101.5625MHz. If you want to return it to 64 again then remove the solder blob and blob the 64 track again.

ON NO ACCOUNT may both tracks be shorted or there will be a dead short across the regulated 5v supply!

The Prototype

My original prototype was developed on PCB with the only coils fabricated on the board itself. I like this method since it removes any instability, microphony and allows the circuit to be reproduceable. My finished project looks like this:

My camera is not so good with the resolution, so perhaps you may like to look a little closer at the finished PCB. The first picture is the crystal oscillator and CMOS logic, the second picture shows the VCO, filter and the top left, the output buffer stage.

Performance And Alignment

The prototypes I built all deliver about +2.5dBm (2mW) with a 9v PP3 battery as the supply, but the circuit may be driven from 13.8v (nominal 12v) to deliver +9dBm (8mW). Since the unit is frequency-locked to a crystal then the frequency is independant of supply voltage. The AF input sensitivity is about 200mV which makes it ideal for driving with the LINE OUT from your computer or stereo system. You may need to add a preamplifier if you wish to use a microphone at the input.

There is only one adjustment. Using a voltmeter with internal resistance of over 100K-Ohms (normal 20Kohms/volt on 5v range), measure the voltage between the battery -ve and the CD4046, pin 13. Adjust the preset tuning capacitor for a reading of 2.5vDC. Make slow adjustments so that the PLL can follow the tuning changes. It takes about one second for the reading to stabilise after an adjustment. If the voltage is anywhere between 2v and 3v then the synthesiser is in-lock, but the modulation level will vary, depending upon this voltage. If the voltage is higher than 3v or lower than 2v then the synthesiser is out-of-lock (or close to it).

If using a digital voltmeter and an eratic reading is obtained, then add a 100K resistor in series with the meter +ve and connect a 1uf capacitor across the meter terminals. This will slow down the reading and prevent it being so eratic. The output at CD4046 pin 13 is a square-wave, often high impedance (Tri-State).

Suggestions? Warnings?

A couple. If you remove the crystal, then you can couple the output of my simple CMOS synthesiser into IC1 pin 2 and then you can cover any frequency you like in the VHF broadcast band. Just decide on a frequency and divide that by 64 (or 65) and program the CMOS synth accordingly. It may also be worth-while replacing the 4096KHz crystal in the CMOS synth for a 3200KHz crystal, then the DIP switches on the CMOS synth will be direct-frequency programming in 100KHz steps, or double frequency in 50KHz steps.

I shall do a little experimenting and see if I have the time to make a decent PA stage to give a little more output power, but this is a low priority for this project. I really would like to work on a few more ham-radio projects. I have the need for an HF bands QRP TX/RX and I may just use this synthesised project as the starting point.

WARNING! I have got to say this. This project may be illegal to operate, or even posess, in many countries. Australia, for example, allow the use of this type of transmitter where the output does NOT EXCEED 10uW. New Zealand allow something like 300mW and the USA allows up to 5 units in any one household and the output signal radiation shall not exceed 500uV per metre at 3 metres distance. It is YOUR responsibility to check this out before attempting to build this project. If the secret police break into your home, guns blazing and you get killed, then don't come crying to me. I know of one chap who was arrested entering Saudi Arabia with a Daytong morse tutor. The fool left the battery in the thing, so when the customs officer switched it on he heard morse code. The guy was arrested for espionage. He got out of jail after three days, but he never got the morse tutor back. The moral here is be carefull, even it it is 100% legal.

I hope that you have a lot of fun with this project. If you have any complaints then please address them to Save$$$$$@savecrazy.spammers.com. Complements, praise and simple hero-worhip may be sent to harryvpo@hotmail.com. Money is always greatfully received and rarely refused (only kidding!!).

Very best regards from Harry Lythall, Lunda, Sweden.

Return to INFO page