[time-nuts] Time interval counters and low S/N signals.
Dr. David Kirkby
drkirkby at medphys.ucl.ac.uk
Fri May 20 08:56:50 EDT 2005
Sorry if this is slightly off-topic, but I guess people here might have a better idea than most about
this than anywhere else I can think of.
Please reply off-list, as one could say it is not directly related to the purposes of the group,
although one might argue it is quite relavant when comparing oscillators to signals such as GPS with
lots of jitter.
I have a professional application (a research project) requiring the measurement of the delay between
two sine waves at 70MHz. One is clean (direct from an RF oscillator). The same oscillator will also be
used to RF modulate a laser. The light passes through human tissue and is detected on a photodector.
Obviously the light detected is amplitude modulated at 70MHz too. The problem is the latter signal is
weak and noisy. Amplifying it is easy - removing the noise less so.
Stanford Research sell a lock-in amplifier (SR844) that will go to 200MHz and should be well suited to
recoving this signal and computing the phase (time) between them. But I don't have access to one.
The idea came to me of using my 5370B time-interval counter for this. Feed one signal into the start,
the other into the stop, and measure the time directly.
The problem is one signal is noisy, so whilst the start can be very regular, the stop will not be. But
hopefully the effects of this would be reduced with the averaging functions built into the TI counter
and/or those performed later after data collection on a computer. I'm just wondering how well TI
counters are likely to work in this manner.
Ideally I'd like to do this at a low as S/N as possible. Perhaps 10dB, or as low as 0dB, or even <
0dB. Lock-in amps would work with the noise > signal, but I'm not sure how well a TI counter would
perform under such poor S/N's, even with averaging. It is clearly not what it was desgned for.
The bandwidth is essentially 0, and is in fact bandpass filtered at 70MHz with a crystal filter.
The obvious this is to try this. Unfortunatly, the TI counter is quite heavy so I don't fancy taking
that into work, although I will do if this looks workable. Borrowing the test kit from work to do it
at home is no less inconvenient.
Anyone ever tried this sort of thing? Comments?
Dr. David Kirkby PhD CEng MIEE,
Senior Research Fellow,
Department of Medical Physics,
Mallet Place Engineering Building,
University College London,
London WC1E 6BT.
More information about the time-nuts