RDS on a Ramsey FM25B

The RDS signal is sent on the FM signal using a 57 kHz subcarrier, by a piece of hardware called the RDS encoder, which is separate from the transmitter. The RDS subcarrier must be in phase with the 19 kHz pilot tone used by the FM stereo system, so I had to modify the transmitter to feed a signal including the pilot tone to the RDS encoder, and to get the RDS signal to the transmitter.

As I was still learning, I started with the first diagram from Pira showing the stereo encoder as a separate device:

The FM25B transmitter had the stereo encoded integrated with the transmitter. I found where the stereo MPX output connected to the transmitter and connected the RDS encoder there:

This didn't work, probably due to a number of causes, but now I believe the biggest problem was the use of the computer's power supply (see this). This introduced an objectional noise in the received audio. It also appeared that the RDS wasn't working well, but that may have been because I was using a terminal program to connect to the RDS encoder rather than the software available from Pira.

The Fix

As I mentioned before, I replaced the computer power supply with a wall wart and the regulator supplied with the RDS encoder. I also changed the connection method using a different diagram from Pira:

In this connection, the MPX signal from the stereo encoder built into the transmitter is used to extract the 19 kHz stereo pilot tone, which is used to keep the 57 kHz RDS signal in phase. I modified my transmitter as follows:

Here's a picture of my "dead bug" assembly inside the FM25B. Note that the resistor labeled "47k" in the schematic above is actually 56k in practice.

I simply routed the cables out the back through a hole in the plastic back panel:

Noise

I did a test to see how much noise this modification adds to the signal. I started with an unmodified FM25B. I did this test away from the computer, with nothing connected to the audio input. I turned the audio gain on the radio to a point louder than would be comfortable if there were a signal with actual audio. Then, I switched to the modified FM25B. Surprisingly, there was an AC hum that was louder on the unmodified transmitter. It was still not very noticable. Then, I turned the RDS encoder on and connected it. It did generate a little noise, but it would probably be hard to hear if there was actual audio turned to a comfortable listening level.

Moving the boxes around changed the noise a little, so I lined the plastic case of the encoder with aluminum foil. This seemed to help a little bit.

I'm going to try it later actually connected to the computer and playing music, but at this point the noise is at such a low level that I'm not going to worry about it.

Software

In addition to the hardware changes, I also downloaded the Magic RDS software from Pira. Using this is a world easier than a terminal program, and it probably fixed some things that were not right in the encoder.

Update

I was still a troubled by the small amount of noise, so I took a closer look of my modifications (above). This involved understanding how the Ramsey transmitter works. The transmit signal is generated by an oscillator in the chip whose frequency is controlled by the parallel LC circuit connected to pin 9. The L is L1, the .18 uH coil. The C is basically composed of D6, D7, and C11 in a series-parallel configuration. (The .10pF capacitor connected to pin 9 couples the LC circuit to the oscillator; the .001µF capacitor is a DC-blocking component. They have little affect on the LC frequency.) The varactor diodes D6 and D7 change their capacitance depending on the DC (or audio frequency) voltage applied. The voltage to D6 passes through a low-pass filter, so this voltage controls the transmit frequency, but does not have audio. The voltage to D7 is the audio frequency.

In my circuit, I was passing the RDS signal (at 57kHz) to D6. This signal was being attenuated by L1 and the .001µF capacitor. However lower frequencies (noise) were not being attenuated.

So I simply moved my new 47kΩ resistor from D6 to D7 as shown here:

Fixed! Much less noise in the output. Later I might want to replace the 47k with a 100k resistor, because I believe the RDS signal is a bit too strong.

Later, I did a similar modification to a FM100B, with much better results.