[time-nuts] A "solderless" frequency reference from 198 kHz radio 4 using a lock-in amp

Dr. David Kirkby - Kirkby Microwave Ltd drkirkby at kirkbymicrowave.co.uk
Mon Aug 17 09:52:55 EDT 2015

I know this is not rocket science, but I mentioned some time ago about the 
possibility of using the BBC radio 4 signal on 198 kHz to make a 10 MHz 
reference. My initial plan was to shift the frequency of a 10 MHz oscillator in 
a Stanford Research DS345 function generator to exactly 10 MHz using the 
calbytes. But I had some concerns this might damage the function generator's 
EEPROM, so did not do this.

What I did however was try a proof of concept. This was the setup.

1) Loop antenna (1.0 x 1.5 m) connected to the input of a EG&G 7260 dual channel 
lock-in amplifier. I'm sure this antenna is far bigger than needed, but I 
happened to have it around. If I turned the gain of the lock-in up too much, the 
198 kHz from the loop overloaded the lock-in amp, so that's a sure sign the 
antenna is bigger than necessary.

The loop is not tuned to 198 kHz. At 200 kHz, the impedance is 115 + j 3084 Ohm. 
The resonate frequency is somewhere between 300 and 320 kHz - I don't have 
anything to tie that down any more.

2) 10 MHz GPS reference feeding into the reference input of a function 
generator. Of course this defeats the object of making a reference based on 198 
kHz, but it was useful for test purposes.

3) Set the function generator to generate 198 kHz and feed that to the reference 
input of the lock-in amplifier. This rather unfriendly lock-in had no facility 
to use an external 10 MHz source. Although it obviously has a reference input, 
that is limited to 256 kHz.

4) Look at amplitude and phase of the signal on the lockin amp, as the function 
generator is varied in frequency.

Here's a short video,


where the function generator is set to first 200 kHz then 198 kHz. Contrary to 
the first sentence or two, the antenna is connected to the lock-in, not the 
function generator!

I suspect if instead of feeding 10 MHz from GPS into the function generator, I 
used another signal generator set to approximatley 10 MHz, I could adjust the 
frequency of the signal generator to keep the phase of the lock-in stable. At 
that point I know the signal generator is producing 10 MHz, since it is driving 
the function generator to produce 198 kHz. Any error in the 10 MHz output from 
the signal generator would mean the function generator does not produce 198 kHz, 
so the phase on the lockin should change with time.

One issue is having sufficient frequency resolution on the signal generators. 
Mine range from 1 uHz to 1 kHz.

HP 8665A 10 kHz to 4.2 GHz, resolution 10 mHz
HP 83623A 10 MHz to 20 GHz. Resolution = 1 kHz (upgradable to 1 Hz if I can get 
a software option)
Stanford Research DS345 1 uHz to 30 MHz. Resolution 1 uHz.

The only sensible choice is to generate the 10 MHz with the Stanford Research 
DS345, and feed that to the 8665A set at 198 kHz. The 1 uHz resolution on a 10 
MHz source give a fractional change of 10^-13.

Anyway, something I can play with.

At the moment, I don't have any of these instruments under GPIB control, but 
that is something I can do of course. Once they are controlled by GPIB, it 
should be possible to generate a decent 10 MHz signal with the Stanford Research 
DS345, but controlling its frequency over GPIB. Unlike changing the calbytes, 
which are possibly in EEPROM. regular changes to the frequency over GPIB are not 
going to damage the Stanford Research DS345

Anyway, it was something to play with, even though it uses a fair amount of 
expensive test equipment to make a 10 MHz reference.

Dr. David Kirkby Ph.D CEng MIET
Kirkby Microwave Ltd
Registered office: Stokes Hall Lodge, Burnham Rd, Althorne, Essex, CM3 6DT, UK.
Registered in England and Wales, company number 08914892.
Tel: 07910 441670 / +44 7910 441670 (0900 to 2100 GMT only please)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: drkirkby.vcf
Type: text/x-vcard
Size: 4 bytes
Desc: not available
URL: <http://www.febo.com/pipermail/time-nuts/attachments/20150817/3b978616/attachment.vcf>

More information about the time-nuts mailing list