G1HBE

Better than Nothing.

 

This is The Home Service....

 

 

A Droitwich 10 MHz reference

I’ve long had an interest in frequency measurement and been slightly obsessive about the accuracy of my radio equipment, so I suppose it was inevitable that I’d eventually get around to building a reliable 10MHz standard. To this end I investigated the various sources, finally settling on the BBC Radio 4 transmissions on 198 KHz, which are traceable to NPL. After looking at a few designs I decided to do my own thing.
A few years ago when R4 was on 200KHz and did not carry the phase modulation used by the electricity suppliers, this was quite easy – receive the 200KHz and remove the AM in a limiter/clipper, divide down the 10MHz crystal oscillator by 50 and apply them both to a phase detector, the DC output of which held the crystal oscillator in check.
With Radio 4 now on a rather awkward 198KHz, life gets more difficult. Using conventional logic, 10MHz cannot be divided down to 198, so a rather more complicated path has to be taken; I don’t have a full cct diagram of the Droitwich, but the block diagram below and a few notes should enable the dedicated madman to produce a working copy. This is now in Mk2 form and includes a few modifications. Here goes:

The Mark 2 'Droitwich' Frequency Standard



 

 

 BBC radio 4 is received on a ferrite rod and the signal is applied to a single-stage dual-gate mosfet amplifier.  There's a wee bit of  accidental positive feedback around this circuit, which narrows the bandwidth a bit.

TheMk2 version is exactly the same with the addition of a 15pf trimmer which is twizzable externally. This is to allow correction of a slight temperature drift which seems to be caused not by the caps but by the ferrite. A length of screened cable  takes this signal to the selective amplifier in the main unit.

 

 

 

 

 

The MK2 198KHz amplifier (below) is nice and stable, unlike the Mk1. It takes the signal from the head amplifier and brings it up to about 300mV p-p.  This goes into the ‘tin can’ containing the dividers and phase detector.

 

 

 

 

 

 

 

Here, the 198 KHz signal is amplified, clipped and amplified again, by which time it is a modulation-free square wave. This clean 198KHz signal is now divided by 99 by a pair of 4029’s and a 4-input NOR gate, resulting in 2KHz. This signal goes into one half of the 4046 phase detector.

 

 

 

 

 

 

The crystal osc is in its own can and is conventional with the addition of a varicap diode in the earthy end of the crystal. There’s a buffer amp in the can. The ouput at 10MHz is taken into the divider box (above) where it is faced with a string of three 4017 CMOS divide by 10 chips and a further 4017 set to divide by 5, giving a total of 5000 and resulting in a 2KHz square wave which is applied to the other half of the PD. Note the first 4017 needs to be an HC part, as it has to cope wih a 10Mhz input.
The loop filter time constant is fairly long in order to get rid of any trace of the narrow phase modulation which R4 uses to send secret messages to Russian spies.

 

A completely new addition is the AGC board, which has three functions. First is to hold the signal level steady, giving the amplifiers and clippers a less demanding job. Second is to drive the signal meter. Third  - this was suggested by a friend - by using a very fast AGC response it is possible to smooth out some of the modulation, again taking some of the load off the clippers and reducing the likelyhood of a 'de-lock' under heavy modulation.

 

 Page no longer updating.