AWR2544LOPEVM: 12-bit Data Streaming via mmWave Studio

Hi, 

I have already configured this setting, when I run with 16bits it's successfully creates a bin file, when I try to run with 12/14 bits no bin file is created.

Data is 12bit but probably streaming out 16bit via ethernet, can I control the number of bits streaming via ethernet?

Can you help with this issue?

Thanks, 

Shlomi

Hi Shlomi,

Can you share the mmWave Studio output log?

Regards,

Samhitha

Hi, 

The log is simillar in both 12 and 16 bit configurations, despite the option of the adc config that you mentioned in the picture above and SW1 on the DCA that switched to 12bit.

Somehow 16bit works, it creates the correct bin file, and 12bit don't even create the file.

Can you confirm that DCA1000 can stream 12bit via ethernet?

Thanks, 

Shlomi

Hi Shlomi,

Shlomi Shvartzman said:

Can you confirm that DCA1000 can stream 12bit via ethernet?

You should be able to capture 12 bit ADC data using DCA1000. 

Shlomi Shvartzman said:

The log is simillar in both 12 and 16 bit configurations, despite the option of the adc config that you mentioned in the picture above and SW1 on the DCA that switched to 12bit.

Did you click set after changing the number of ADC bits in ADC config?

Regards,

Samhitha

Hi Samhitha,

Thanks for sharing the LUA script. I found the issue — the CaptureCardConfig_Mode was set to (1, 2, 1, 2, 3, 30) instead of 1.

Currently, BIN files are being successfully created for both 12-bit and 16-bit settings. However, the file sizes are identical and match exactly the expected size for 16-bit data. Based on this, I suspect that the LVDS is indeed streaming 12-bit data, but the Ethernet interface is padding or aligning it to 16-bit before streaming and storing it on the PC.

Is there any way to configure the Ethernet streaming to output raw 12-bit data without padding?

Thanks,
Shlomi

Hi Samhitha,

Thank you for the information.

Assuming I’m capturing 12-bit data via LVDS, I don’t see a reason to go beyond 12 bits for a 12-bit ADC sampler. So, LVDS is no longer a bottleneck in this case.

However, it seems that the Ethernet interface pads the data with zeros before streaming. Is there any concern about potential data loss if the data is being written faster than it can be streamed over Ethernet?

Thanks,
Shlomi

Hi Shlomi,

Let me check internally to find the potential cause of padding data with zeros. Is there is any specific reason why you want to use 12-bit ADC data? We haven't tested 12-bit ADC data extensively as 16-bit ADC data is the widely used mode.

Regards,

Samhitha

Hi Samhitha,

The ADC resolution is 12 bits, but when I analyze the captured data, I observe values that exceed the expected 12-bit signed range of [-2^11, 2^11-1], that's probably because I configured 16 bits in ADC Config.

Previously, as you mentioned, capturing 12-bit data (configuring 12 bit in ADC Config) results in 16-bit output with zero-padding.

However, now that I’m capturing 16-bit data directly (configuring 16 bit in ADC Config), could you please clarify the exact process by which the 12-bit ADC values are extended to 16 bits?

From the real data I’ve collected, it doesn’t appear to be zero-padded.

Can you confirm that the signed integers are saved in two's complement method?

Thanks,
Shlomi

Hi Shlomi,

Shlomi Shvartzman said:

The ADC resolution is 12 bits, but when I analyze the captured data, I observe values that exceed the expected 12-bit signed range of [-2^11, 2^11-1], that's probably because I configured 16 bits in ADC Config.

Let me check and get back to you by Wednesday.

Regards,

Samhitha

Hi Samhitha,

I'm a bit confused and hoping you can clarify something.

According to the chip's user guide, the ADC has a resolution of only 12 bits. However, in the ADC configuration, we’re able to set it to 16 bits. It seems that the 12-bit ADC data is somehow being represented or converted to 16-bit format.

Could you please explain how this works? Am I misunderstanding something here?

Thanks,
Shlomi

Hi Shlomi,

It seems to be a typo in AWR2544 datasheet. Let me check with the team and get back to you tomorrow.

Regards,

Samhitha

Hi Samhitha,

Thank you for the clarification.

As you mentioned, the 2544 supports only 16-bit configuration.
Now, regarding the 2E44P — I was able to successfully capture 12-bit data, and the datasheet confirms that the ADC resolution is 12 bits. So we can probably agree that the ADC's actual resolution is indeed 12 bits.

However, if I configure the ADC to capture 16 bits, what exactly will I get? How are the 12 bits converted or mapped into 16 bits?

Thanks,
Shlomi