IWR1843BOOST: IWR1843BOOST

Hello,

Can you link me the design document SPY007 you are referring to? I can not seem to find the document being referenced.

Best Regards,

Pedrhom

Hi again Pedrhom,

oops, I think my typing to in the way: spyy007 - "Using a Complex Baseband architecture in FMCW radar systems", May 2017.

I assume this is relevant to the 1843 etc devices.  Certainly, mmwStudio allows the complex_X1 & complex_X2 selections, and I need to understand just what each selection means.

A quick update, as we've spoken before.  My "new" radar system is now all working, bursted continuous mode, synchronised to my external system, initial DSP algorithms in there, all fully controllable and data-down-loadable.  Externally similar to the original OOB demo with an expanded command protocol, but doing something quite different. I've started using it for real-world measurements, hence why I now need another level of understanding of what happens in the DFE, and what the different programming features do to the data after the adc. As said, particularly interested in frequency shifts and filter bandwidths.

many thanks,

Alan

Hello Alan,

Please give me some time to look into these details.

Best Regards,

Pedrhom

Hello Alan,

In complex 2x mode, there is no frequency shifting employed.  The real part and imaginary part are sampled and sent out as is.  So, in this case, both inband and image band is visible at the final ADC output. There is no image rejection in this case.

In complex 1x mode, frequency shifting and image reject filtering are employed.  The frequency shift is done such that the IF bandwidth of interest (which is 90% of Nyquist for the chosen sampling rate) is centered at DC.

For example, if the complex 1x output sampling rate is 5Msps, then the actual IF bandwidth of interest is 4.5MHz.  In this case, the frequency shifting is done by 2.25MHz, and a low pass filter is used to do image rejection.

Best Regards,

Pedrhom

Hi Pedrhom,

great - just what I was looking for, thanks.

I'm most interested in X2 mode for now but, for completeness of understanding: does this mean that, in X1 mode, the BSS does a 0.9 calculation to get the mix frequency?  Using say 1/2 the clock rate would seem simpler ... I just need to be fully up to speed on what the device does with the ADC samples.

Similarly (and still on DFE, so hopefully not really a different thread), I note the numbers coming out in CONTINUOUS mode are pretty small.  As the returned signal will be essentially dc (actually sin & cos, but at 0Hz) - then this is an effect of the two HPF sections?  Is it possible to bypass these ... i.e. I'd really like to get at raw adc samples, with nothing done to them?  I know this doesn't match what the device was designed for but, for my research, I need to understand what happens between the output of the RF sections, and the samples I can get at.

many thanks

Alan

Hello Alan,

I have a hunch on the answers to your questions (yes to the 0.9 calculation, no to the bypass HPF sections), but let me confirm for sure. Give me some time to look into this to confirm and be certain.

Best Regards,

Pedrhom

Hi Pedrhom,

you're probably having a couple of days off now - enjoy! 

I think we can forget about the HPFs etc, as I've found a more important reason why the numbers coming out are so small.

It looks, on the surface, that my code is all working exactly as it should.  I'm  start/stopping bursts in continuous mode; chirpAvail interrupts are happening in both MSS & DSS; its taking the correct numbers of samples, and they're going in the right place in AdcBuf.  However, there isn't really any data there.

I don't have any other signal source/receivers to test this against (can't apply for budget to but another 1843 just yet), so I had a look in mmwStudio re loop back functions.  I got far enough to show this would be useful, but my DCA1000 has stopped working for some reason.  It at least let me find the code necessary to add loopback.

So, I've taken studioCLI (i.e. does contMode), and added code to: enable PSloopback; my code to let me read adcBuf & L3 memory via UART; plus my code to add burst-control of continuous mode, triggered from GPIO_1.

This all works, and shows a very good sin/cos in adcBuf, at around +/-5000.  So: should be identical code in studioCLI, and mine - but I don't get anything out (noise, about +/-50) of mine (using the exact same profile set-up file for both).

I think I've missed something when porting contMode code from studioCLI into my code.  I don't want you to spend lots of time delving into CLIStudio, I'm hoping you can help me find where to look for the error from the symptoms:

• It looks to me like the "back-end" is fine i.e. all the sampling looks OK - but there is something in the DFE, ahead of the ADC, that isn't enabled.

Does that bring up any thoughts about what I haven't done in setting up the DFE?

Hoping for a quick solution i.e. easy for you to answer, so we can get this thread closed.

many thanks

Alan