From: Mike Monett
X-Mailer: Mozilla 2.02 (Win16; I)
Subject: Re: Comb/Harmonic Generators
Date: Mon, 09 Dec 2002 17:38:11 -0800
NNTP-Posting-Date: Mon, 09 Dec 2002 17:38:14 EST
Organization: Bell Sympatico
> > Vance, speaking of rf applications, can you give us an idea of the noise
> > figure of your 100GHz sampler?
> Hi Mike,
> These answers are application dependent ... I'll take a shot.
> The integrated noise at the output of the IF is about 3 mV rms;
> the gain to the RF input is 1. The thermal noise for 50 ohms in 100
> GHz bandwidth ... 452uV rms. Noise figure of 16 dB.
> Dynamic range has been measured at >2v p-p.
> From one of the developers ...
> In the mixer or downconversion mode of operation, where the sampler is
> slowly "walked" through the RF waveform, the noise is much lower owing
> the inherent averaging occurring in the IF. Noise in this case is
> determined by the IF bandwidth, which can be orders of magnitude less
> the RF bandwidth.
> Does this answer your question?
Thanks Vance, but something doesn't make sense. When I calculate thermal
noise, I get
rms = sqrt(4 * 1.38e-23 * 290 * R * Bw) ; rms thermal noise
= sqrt(4 * 1.38e-23 * 290 * 50 * 100e9)
= 283 uV
So the noise figure of 16 dB depends on how the measurement is made. It seems
exceptionally good for such a wide bandwidth. I would have expected closer to
28 dB or more when used as a mixer.
For dynamic range, I believe the >2V p-p number refers to the maximum input
signal. While impressive, there are probably few 100 GHz signals this strong.
I'm interested in the minimum detectable signal, where
10 * log((S + N) / N) = 3 dB
or whatever number you prefer.
Downconverting to an IF is interesting. What is the IF frequency and
bandwidth? And why is this better than straight sampling to DC? Is it due to
the difficulty of obtaining stable signals at 100 GHz?
The 100 GHz sampling bandwidth is interesting, since it would improve
acquisition of 10 GHz logic signals from Motorola GigaCom ECL, which is my
main interest. Also, I would expect your modules to be considerably less
expensive than the HP or TEK samplers.
Since your sampler can run at 10 Gs/sec, is downconverting to an IF still