ADP2230 data streaming

[M. Hamza Öncüer]

Hi, I noticed that the ADP2230 supports USB 3.0. In the datasheet, it mentions that it can handle rapid data streaming. Could you provide more information about this, please?

[JColvin]

Hi @M. Hamza Öncüer,

These are my personal testing results from a single ADP2230 on my work laptop with an M.2 SSD -- aka a single data point (albeit repeatedly tested from a pedantic engineer). 

Naturally, DDR is enabled on the ADP2230 during all the tests (the default option, though its worth noting that these rates are simply not possible when using the lower power BRAM only mode as the buffer size of 32 kS simply results in far too much USB overhead for all of those tiny transfers).

All recording to file results saved as a 16-bit binary file, interleaving the data sources when more than one is recorded (analog1 analog2 analog1 analog2...).

• For a single analog input (the resolution is 14-bit at high sample rates, but each sample is recorded as a 16-bit value; you can get higher resolution if you sample at 1/2 or 1/4 of the set system frequency), I can repeatedly successfully record 5 G samples at 125 MHz (the maximum sampling rate).   
• For two analog inputs (16 bits each, or effectively 4 bytes per 'sample'), I can repeatedly successfully record 5 G samples for each channel at 70 MHz.
  • I'm not certain how much faster this could be pushed and remain consistent with USB bus activity. I know for example 100 MHz is not realistic as that would require transferring data at 400 MBytes/sec which isn't viable thanks to the USB protocol overhead. 
• For all 16 digital inputs in the Logic Analyzer, I can repeatedly, successfully record 5 G samples at 125 MHz, with or without data compression enabled.
  • Data compression meaning only data changes are occurred, rather than a sample at every 8 ns. Two 16-bit samples are recorded with data compression enabled, first the data at the change, then the timestamp. 
• Using the Digital view within Scope instrument, and recording both analog channels and the 16 digital channels simultaneously, I can repeatedly successfully record 5 G samples for all three at 41.667 MHz (a third of 125 MHz). 
  • I would like to believe that 50 MHz is possible with further refinement, but expecting no issues for the mixed signal recording at the equivalent 300 MB/s is probably unrealistic. Currently I get a "samples could be lost" message, though it does complete the 27.9 GiB acquisition (3 data sources * 16 bits each * 5 billion samples = 30 billion bytes; 1024 bytes per kilobyte, 1024 per mega, 1024 per giga -> ~27.94 gigabytes). 

For clarity, I picked 5 G samples as a representative arbitrarily large number of samples that far exceeds the on-board memory. In principle, you could record far more samples, but I haven't personally tested it.

Here's an example screenshot for recording all the two analog input channels at 62.5 MHz that I took today:

Let me know if you have any questions.

Thanks,
JColvin

[M. Hamza Öncüer]

Thank you very much! The results seem promising 😃.

Edited March 12 by M. Hamza Öncüer

[Luke Aldrich]

On 3/6/2024 at 2:31 PM, JColvin said:

Hi @M. Hamza Öncüer,

These are my personal testing results from a single ADP2230 on my work laptop with an M.2 SSD -- aka a single data point (albeit repeatedly tested from a pedantic engineer). 

Naturally, DDR is enabled on the ADP2230 during all the tests (the default option, though its worth noting that these rates are simply not possible when using the lower power BRAM only mode as the buffer size of 32 kS simply results in far too much USB overhead for all of those tiny transfers).

All recording to file results saved as a 16-bit binary file, interleaving the data sources when more than one is recorded (analog1 analog2 analog1 analog2...).

• For a single analog input (the resolution is 14-bit at high sample rates, but each sample is recorded as a 16-bit value; you can get higher resolution if you sample at 1/2 or 1/4 of the set system frequency), I can repeatedly successfully record 5 G samples at 125 MHz (the maximum sampling rate).   
• For two analog inputs (16 bits each, or effectively 4 bytes per 'sample'), I can repeatedly successfully record 5 G samples for each channel at 70 MHz.
  • I'm not certain how much faster this could be pushed and remain consistent with USB bus activity. I know for example 100 MHz is not realistic as that would require transferring data at 400 MBytes/sec which isn't viable thanks to the USB protocol overhead. 
• For all 16 digital inputs in the Logic Analyzer, I can repeatedly, successfully record 5 G samples at 125 MHz, with or without data compression enabled.
  • Data compression meaning only data changes are occurred, rather than a sample at every 8 ns. Two 16-bit samples are recorded with data compression enabled, first the data at the change, then the timestamp. 
• Using the Digital view within Scope instrument, and recording both analog channels and the 16 digital channels simultaneously, I can repeatedly successfully record 5 G samples for all three at 41.667 MHz (a third of 125 MHz). 
  • I would like to believe that 50 MHz is possible with further refinement, but expecting no issues for the mixed signal recording at the equivalent 300 MB/s is probably unrealistic. Currently I get a "samples could be lost" message, though it does complete the 27.9 GiB acquisition (3 data sources * 16 bits each * 5 billion samples = 30 billion bytes; 1024 bytes per kilobyte, 1024 per mega, 1024 per giga -> ~27.94 gigabytes). 

For clarity, I picked 5 G samples as a representative arbitrarily large number of samples that far exceeds the on-board memory. In principle, you could record far more samples, but I haven't personally tested it.

Here's an example screenshot for recording all the two analog input channels at 62.5 MHz that I took today:

Let me know if you have any questions.

Thanks,
JColvin

I’m looking at buying the ADP2230 and based on your test captures you’ve provided it will easily be able to capture 10Mbit UART for a couple of seconds (a common method I use troubleshooting). The concern I have is that my computer is older, but does have an SSD with 64GB of RAM. I haven’t been able to get a demo capture with the ADP2230 to work without saying it’s losing samples in record mode using configuration 1 or 3. I use this computer with a Saleae Logic Pro 16 to do 10’s of second captures with the 10Mbit UART, but would prefer the higher analog bandwidth of the ADP2230 when I need it  

Should I expect this demo mode capture (with Waveforms v3.22.2) to be an accurate estimate of how my computer would handle the ADP2230? Or could I expect results more similar to your post?

[attila]

Hi @Luke Aldrich

The demo mode is intended to explore the application and device options; triggering, record... are not its strengths. 
The ADP2230, AD3X50 have deep onboard memory so even if the computer is slow on data transfer or storage it should be able to capture up to device buffer size lossless.