Time of flight adc

[Brettn56]

I'm trying to design a system to calculate time of flight for a CW signal with a 200khz carrier. As soon as the signal has finished transmission (TX pulse varies from 0.1ms to 5ms) I need to 'capture' 1000 samples, each sample being precisely 13.3333uS apart (ie 75khz). I then need the I and Q component of the received waveform to be sent via ethernet.

Am I barking up the wrong tree trying to use an Fpga to perform this? Would I be better off using something like a cyclone v soc(arm architecture)?

Any advice would be much appreciated.

Looking to use an Ad8339 (IQ demod) > ad8334 (vga) > AD8021 (current to voltage) > AD7665 (adc). The ad8339 I'd only use 2 channels (one for I, one for Q). Or am I smarter to use and Adc directly to a micro and do all the conversion/calculation in dsp slices?

Thanks,
Brett

 

[KlausST]

Hi,

so you want to do undersampling.

Are you sure you want 200kHz (5us) and 75kHz?
because it´s just 3/8. This may be good or not... depending on what you are looking for.

For sure you need an ADC with high bandwidth and low capture jitter. They are available.
But I think one single channel should be enough and I / Q can be calculated by the FPGA or microcontroller rather simply.

Another .. not that digital processing solution is to two synchronous rectifers 90° phase shifted and a low pass filter each.
Then you get two rather simple to convert DC signals for I/Q.

One big question is: How do you generate the 200kHz sinewave. It needs to be done with the same clock source than the sampling / rectifying.

Klaus

 

[Brettn56]

Hello Klaus,

Many thanks for the reply.
The CW frequency varies from 100us up to 5ms.
Normally I would envelope detect the signal, meaning signals from 200hz to 10khz, nyquist sampling would mean 20khz maximum.

The carrier frequency varies from 20khz to 200kz.

The transmitted signal burst is a short non modulated (CW) signal that lasts anywhere from 100us to 5ms.

Thank you for any suggestions, they are much appreciated.

Brett

 

[FvM]

You are talking about TOF measurement without mentioning time range and intended resolution. Also what's the transmission medium? Cable?

 

[Brettn56]

Hi, sorry for the lack of details. It's my first post on here!

OK, transmission medium is water (so assume 1500ms speed) transmission and reception transducer is piezo-electric. TX pulse is generated from Centre tapped transformer, push pull style. Square waves drive the fets, which when coupled to the highly capacitive transducer result in a modest envelope signal.

Distance measurement is variable from 1mtr, to approx 1000mtr. Hence the intended 75khz sampling interval, which should allow a sample of the signal every 1cm, so intended resolution of 5-10cm.

Adc resolution at 16bits would be more than enough I would think?

Thanks,
Brett

 

[FvM]

Besides ADC time resolution, the limiting factor is transducer bandwidth. It's not clear how things fit in your application. I also guess you are heading for lower resolution for the long time scale.

 

[Brettn56]

Besides ADC time resolution, the limiting factor is transducer bandwidth. It's not clear how things fit in your application. I also guess you are heading for lower resolution for the long time scale.

Hello FvM,

Thank you for your comments. Transducer bandwidth is quite narrow, being nominally 2khz.

The transmitter stage is capable of high bandwidth (20khz to 400khz), however the transducer in use for this application has a very narrow frequency band, being 199 - 201khz.

And yes, with such a long time scale (at 1000mtr) the resolution is less important. At shorter distances (500mtr and less) the resolution is more important. At distances of 100mtr and less the resolution should be the greatest.
Brett