Hi Sean, see below....

"Sean H. Breheny" schrieb:
>
> Hi Jochen,
>
> Thanks for the kind words :-)
>
> The main problem is not the amount of noise, but it's effect on the ad606.
> I had expected Gaussian noise on the input to produce something roughly
> like DC on the output, since in the datasheet, the only real difference
> that they mention in the effect that Gaussian noise has vs. the effect of a
> sinusoid is that it has a different log intercept. they don't mention that
> the output would be a lot rougher. Is it the case in general that the
> output of the amp with Gaussian noise on the input is very noisy?

The distribution of Gaussian noise is a graph determining how much
amplitude probability there is for each possible input voltage. The log
amp will change that into a Rayleigh distributed output, but basically
this is still a probability function for a spot value of the output
voltage, just a different graph. So your output will still "look noisy".

>
> I'm thinking that the problem may be that I'm dealing with noise that is
> narrowband and is not white. For example, if you have white noise with an
> rms amplitude of 2mV, you would expect to see a DC level out corresponding
> to about -20dBm. Because the noise has some frequency components below the
> low-pass video filter cutoff (internal in the ad606), you would see some
> noise on the output, but nothing like a significant fraction of the average
> level. However, if you feed noise which is only 100kHz wide into the ad606,
> then when it gets detected, the nonlinear operation of detection will shift
> some of the noise down into the 100kHz region, where it will pass right
> through the video LPF and become noise on the output. Now, this happens in
> the white noise example, too, but there, the portion of noise that gets
> converted into a 100kHz BW by the detection system is only a fraction of
> the total input noise power, but in my case, it is most of the input noise!

One more word on noise and the associated bandwidths: Perhaps I was
unclear on the subject.....
We have to consider two different signal worlds here - one world is your
input signal of so-and-so dBm, and noise-laden in itself, with a 100 KHz
bandwidth,
and the second world is the log amp input, for which the noise bandwidth
is 70 MHz, regardless of the filtered input signal. To get around this
source, you have to increase the signal input. The voltages you
mentioned yesterday indicated that you are operating very much on the
bottom of the range, where the device has most internal noise.
A logamp is basically a series of cascaded amplifiers, and the noise on
the first one normally is sufficient to get the last amp mildly into
saturation. As the input signal increases, the point of saturation moves
towards the input, and this way the log amp works, the stage where
saturation happens (and how much) will determine your output voltage...

Hope this helps more than it "obfusticates" the issue....

Greets
Jochen Feldhaar DH6FAZ


>
> This is my theory of what is going on (have not had a chance to test it
> yet) and I'm wondering if anyone else has seen this before?
>
> Sean
>
> At 12:57 PM 6/20/2002 +0100, you wrote:
> >Hi Sean,
> >
> >I am always amazed at what you do (and the quantity), and I am glad that
> >maybe I can be of help in this case.
> >
> >I have used the AD606 in a spectrum analyzer, which is about the
> >application you use it for, because at the output of the resolution
> >filter you get a series of peaks out of the input signal.
> >When I had to get more than 80 dB of range out of the 606, my biggest
> >problem was noise on the input pins. The bandwidth of the AD606 is
> >around 70 MHz, so you can calculate the theoretical minimum by
> >estimating 4kTBR (resistor noise).
> >At that bandwidth this gives a BIG figure, and then you have to add the
> >semiconductor intrinsic noise.
> >Conclusion: Maybe the coupling between the filter and the IC is not
> >optimal, maybe it is not shielded and other signals (up to 70 MHz) are
> >coupled, maybe the power supply is noisy, and a lot of other
> >maybes....nut as a HAM and student you will have the knowledge to test
> >for them one by one and eliminate the possible sources.
> >Now for the 10 dB of noise: It is always there, either the resistor
> >noise plus semiconductor, or your input signal. If you look at a
> >spectrum analyzer, you see the baseline very noisy, and the noise
> >quantity will go down as your signal gets stronger. So there are 2 ways
> >for you to score:
> >1) Use a preamplifier between bandpass and log amp
> >2) Use a low pass filter on the output of the log amp, on the spectrum
> >analyzer this is called video filter - but it limits your response
> >speed!
> >
> >Good luck!
> >
> >Jochen Feldhaar DH6FAZ
>
> --
> http://www.piclist.com hint: To leave the PICList
> mailto:piclist-unsubscribe-request@mitvma.mit.edu

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads