Part 5: MSF detector - turning the radio signal into data

by Dr Jonathan Hare, The Creative Science Center, Sussex University.

click here for the whole MSF series

The MSF transmitter sends time data bit-by-bit over the 60 seconds of each minute by dropping the 60 kHz carrier. It's a basic CW signal [1]. Two bits are transmitted every second by modifying the length and number of carrier drop(s) per second (see the second article in this series). As is often the case in timing circuits a negative logic is used. When the carrier is ON the data being sent is a 'zero' and when the carrier off it is a 'one'. It is the drop in career that is accurately timed i.e. the ON to OFF state rather than the OFF to ON state. I suppose dropping the carrier (ON to OFF) is a clearer transition than the OFF to ON transition, as fading might confuse the exact ON transition but the carrier going fully OFF is quite distinct.

Reading the data sheet from the NPL it appears that they do more than key the carrier ON and OFF to get the precise signalling. This may be because the high Q antenna / output circuit will 'ring' for too long after the transmitter keys off. They imply that the ON / OFF data is a combination of data pulse and actual switching of power to the antenna system [1].

S-meter / signal strength detection
In principle we can take the changing voltage from the receivers automatic gain control (AGC) i.e. the S-meter circuit as the basis of the data detection signal. The S-meter voltage can be feed into a comparator / trigger to convert the change in signal strength into logic level (0 and 5V) data pulses. If the MSF signal is very steady and strong (i.e. if you are relatively near to the transmitter) then this will work well. This is shown at the top of the first picture. A little bit of hysteresis built into the comparator circuit will help to stop 'chattering' at the transition edges at switch ON and at switch OFF.

Problems with signal strength decoding
The main problem is when the signal strength is weak or fading. When this happens the comparator will have problems because the transition point derived from the AGC voltage will be changing and so we won't be able to detect the data reliably. You can see this in the bottom of the first picture. This problem is nicely discussed in the section on 'data communications' in the latest RSGB handbook [2].

BFO - audio tone detection
A much better way of detecting the MSF CW signal is to use a beat frequency oscillator (BFO) on the receiver and detect the audio tone that is created, rather than try and monitor the changing signal strength. From a signal processing point of view you are taking the signal away from the AGC 'DC' region, where there is a lot of noise (drift and low frequency noise) and moving it into the 'audio' spectrum where simple filters can dramatically improve the S/N ratio.

If you don't have a BFO (CW or SSB) on your receiver (e.g. if you are using an old AM radio) you need to make up a mixer circuit that will take the IF o/p of the receiver (455 kHz) and 'mix' it with a local oscillator frequency to produce an audio product.

I used an old 456 kHz crystal to make a stable local oscillator (LO) and fed this into a NE602 mixer to make up a product detector. The o/p is the difference ca. 456.196 - 455.000 kHz = 1.196 kHz, a tone which I then feed into a simple phase locked loop (PLL) (NE567) tone decoder [3,4]. If you can't get hold of a 456 KHz crystal, you can use a 455 IF transformer (de-tuned by ca. 1 kHz) as a simple oscillator coil and use the NE605 internal LO circuitry, see diagram. I also used a two op-amp audio band pass filter which I put between the mixer and the PLL tone decoder (as indicated on the diagram) to get better signal to noise ratios but I don't think it is really needed for the UK MSF signal strength.

To get a decent signal for the BFO mixer you will need to wire into the IF of the radio. You can do this on the very last IF transformer; find the AM germanium detector diode which is usually directly on the output of the last IF transformer (don't remove it as its also used to generate the AGC voltage) using a piece of miniature coax and connect between the earth (e.g. the can case) and the output. Connect a scope or audio amp to pin 5 of the NE602 mixer ic and adjust L2 for a ca. 1 kHz tone and then peak L1 for maximum signal.

BFO superhet radio

CW SSB receiver
If you are using a communications receiver with a built-in BFO then you don't need the NE602 mixer stage. In this case you can simply connect the headphone (or external output) output of the radio directly into the NE567 tone decoder (via the 0.22uF cap). Adjust RV1 so that the PLL oscillator runs at about 1kHz (measure on pin 5).

BFO-mixer circuit

NE567 PLL tone decoder
The NE567 has a 'flying-collector' output so it needs to be 'pulled-up' by a resistor (ca. 4k7). The result is that the PLL drops its output (to zero) when it detects, 'hears' the tone [3,4]. In the MSF PIC decoder, (which I will discuss in the next article) we need an inverter to create the correct polarity data for the PIC input port. I have used a single BC108 to do this job but if you use a quad inverter you can use the other gates to drive LED's to provide a visual indication of the detector o/p states. These LED's are useful to keep a check on the quality of the data coming from the receiver but can of course be switched-off if they are not required.

Results with tone decoding
So far the only problem I have found is when trying to demonstrate the device in a public lecture [5] as the data projector I was using seemed to produce a strong 60 kHz signal! Other than that the BFO PLL detection is very reliable.

next time ...
In the next issue I will describe a PIC MSF decoder that can log the bit A and bit B data, do the time calculations at the end of the minute and show much of the time information on an LCD display it makes a great radio shack clock.

References and Links
[1] For the Time and frequency section of the NPL and also for the MSF data sheet (PDF)
[2] Radio Communications Handbook, 10th Edition, see section 20, digital communications, p. 20.1
[3] NE567 tone decoder data sheet:
[4] Design of phase-locked loop circuits, H M Berlin, 1988, Tabs book. ISBN 81262 21545
[5] WADARC (Worthing and district amateur radio club) talk; 5th May 2010.


Dr Jonathan Hare, The University of Sussex
Brighton, East Sussex. BN1 9QJ.

home | diary | whats on | CSC summary | latest news