Quadrature Sampling

Define "cheap". If you use something like a dsPIC to generate your signal, would that do the job? I don't know the constraints with SONAR, so this may be off base, but couldn't you just store the waveform in ROM and reel it out to a DAC?

Come to think of it, Xilinx Spartan parts aren't too spendy these days. One of them could do your chirp generation (or at least waveform-playing) and the preprocessing on the incoming signal.

For low volume I'd use a good fast ADC and sample it as fast as your DSP chip (or FPGA) could handle. This makes your ADC better, and lets you relax the constraints on your band pass filter.

Keep in mind that "narrow bandwidth" means that your Nyquist rate can be well below your carrier frequency (at least until you implement that chirp, and widen out your bandwidth).

I'm not sure what you mean by "quadrature sampling". If you mean a pair of analog mixers going to a pair of ADCs, it probably only makes sense these days if you're working at RF.

The last time I looked you could get video ADCs that'd sample at 60MHz and give you 14 bits in the clear with much more available after narrowband filtering. If there aren't ones today that'll do the same speed and give you 16 bits I'll eat my hat.

Given one of those guys, you can just sample way fast and do quadrature decimation in your FPGA, if you're so inclined.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

On Mon, 20 Oct 2008 15:56:12 -0700, Tim Wescott wrote (in article ):

Generating the chirp waveform is not hard. Al you methods will work. An efficient transducer for water is the hard part. most SONAR uses piezoceramics with a resonant frequency of 50 kHz or 200 kHz.

Taking samples in pairs multiplied by reference frequencies 90 degrees apart to get amplitude and phase in a single sample treated as a complex number.

In this case, the reference frequency equals the carrier frequency. If you google "complex sampling" you will probably find a bunch of papers on SSPS statistics sampling, but I should be calling it complex sampling. I don't have a DSP book handy. ALL my books and tools and instruments are in a climate controlled shipping container about four miles away. Very frustrating.

-- Charlie Springer

Quadrature sampling makes a _lot_ of sense internally if you're dealing with narrow-band data, because a quadrature sampler followed by a low-pass filter equates to a nice bandpass.

But I'd do it digitally, after sampling. You seemed to be thinking of doing the quadrature sampling in analog hardware, however, which is what I was trying to clarify. This probably isn't what I'd do for a low volume, 200kHz-carrier application.

My knee-jerk suggestion would be to do quadrature sampling digitally, possibly after under-sampling the bandpass-filtered signal if power or cost limitations kept me from sampling at 4x the carrier.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

o, I

iplier

ist

m

lso,

n

Quadrature sampling detectors using bus switch devices are popular with radio amateurs for low-cost software-defined radios.

Leon

On Mon, 20 Oct 2008 18:49:15 -0700, Tim Wescott wrote (in article ):

Given the option, I will sample at many times the carrier and use an optimal filter (correlate with the carrier in this case by multiplying after Fourier Transforms). Why? So that if I manage to find a cheap enough chirp transducer the software doesn't change but I get tremendous pulse compression for much higher resolution.

-- Charlie

Reminds me of a project I did with an RTX2001 running a DDA algorithm. My client only wanted a "sinewave ping" to be transmitted but was very interested in amplitude and phase of the returns (on several channels).

Sometimes you can obtain these quanta without too much complexity. Have you considered getting the amplitude as a DC signal from normal AM receiver techniques. The phase is then just a case of squaring the reference and return in order to compare time difference between the same edge of each.

What characteristics do you need for the chirp? Might be a simple way of doing that.

--
********************************************************************
Paul E. Bennett...............
Forth based HIDECS Consultancy
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-811095
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************

[[ This message was both posted and mailed: see the 'To' and 'Newsgroups' headers for details. ]]

On Tue, 21 Oct 2008 04:03:48 -0700, Paul E. Bennett wrote (in article ):

The chirp, or linear FM chirp, would be something like 75 kHz to 125 kHz in

100 cycles. The biggest problem is an efficient transducer for water. The best I know are expensive magnetostrictive or piezo polymer systems.

-- Charlie

I know absolutely nothing about matching transducers to water, but at least when matching a transducer to air, the exponential horn can produce quite significant sound pressure levels with very low power. Typically, only a few watts is required to access a large crowd of people using an exponential horn loudspeaker in a public address system, compared to hundreds or thousands watts required if closed speaker boxes are used.

Of course the drag caused by the exponential horn in a moving vehicle would make this idea unpractical.

Paul