Why does my Analog Discovery sample at such a slow rate?

[krb686]

I just bought this product recently.  It looks pretty nice, and I'm pretty sure I'm not using it to its full potential.  The specs say it samples at 100MS/s, yet that certainly is not happening with mine.

From the looks of it, I see 2048 samples at 4.005 kHz.

And by counting the marks between the time divisions (1ms apart), it does look like approx 4kHz.  Why is this not running faster? This is a far cry short of 100MS/s.  I also see no setting anywhere to change the sampling rate/speed.
 

[attila]

The sample rate is adjusted by time base setting. Rate = Samples / TimeRange, where the number of samples is in default configuration 4k.

The configuration can be changed in WaveForms main window Device/Manager after checking "Show advanced features".

For sample rates under 100Msps half of the buffer is used for noise detection, which means more than one transition between two sampling points. This can be disabled in Analyzer/Options/General.

 

The currently released WaveForms version has a bug in Logic Analyzer, where the time base is extended for post trigger acquisitions. For instance having trigger position 100ms time base 1ms/div, the actual time base will be 20ms/div (200ms span).

This problem is solved in the version sent you in a private message.

[krb686]

Thanks, that clears up a lot.  If you don't mind, I have a few other questions if you can answer them.  What is the difference between the two versions you have linked in your PM? Is the version 3 just a beta? Patch notes or change log anywhere?

Also, another quick question regarding a specification of the Analog Discovery.  The specs state it is capable of 16K transitions per pin.  So this is an internal buffer limitation on the device? Yet I am not limited to just a single buffer, and can specify many buffers to be filled in WaveForms.  Is there a limit to how many buffers I can specify? And why are these logically separated once they have been transferred to the application on my PC? I only ask because it seems rather tedious to even think about the concept of filling over 6,000 buffers per second at 100 MS/s, and having to dig through all of these.  I get the corresponding limitation on the device, but it seems like something that could/should be abstracted away from the end user.  AKA - one single timeline from start to finish that contains all my samples.  Just my thoughts.  And thanks again.

[attila]

You have the WF2 vs 3 feature list / comparison in PM.

 

Yes, the 16k sample buffer is a device limitation. 

The device performs acquisition up to 100Msps of up to 16k samples, then it takes at least 6ms to read this data to PC and to start a new acquisition. These sequential acquisitions, with time gap between them, are stored on PC and called buffers. Because of the time gap these can't be represented in one continuous capture.

The number of PC buffers can be specified in WF2 Analyzer/Options/General and in WF3 in the gear after the buffer field.

 

In WF3 the Run mode beside repeated, scan screen and shift, allows streaming mode. Using this it can capture continuously up to 1-2Msps of 1M samples.

[krb686]

You have the WF2 vs 3 feature list / comparison in PM.

 

Yes, the 16k sample buffer is a device limitation. 

The device performs acquisition up to 100Msps of up to 16k samples, then it takes at least 6ms to read this data to PC and to start a new acquisition. These sequential acquisitions, with time gap between them, are stored on PC and called buffers. Because of the time gap these can't be represented in one continuous capture.

The number of PC buffers can be specified in WF2 Analyzer/Options/General and in WF3 in the gear after the buffer field.

 

In WF3 the Run mode beside repeated, scan screen and shift, allows streaming mode. Using this it can capture continuously up to 1-2Msps of 1M samples.

Very cool! Thanks for the answers. WF3 definitely looks like a step forward from WF2, that's for sure.  Though how does the device manage to stream to 1 million samples when it can only handle 16K internally? Does it attempt to transfer to the PC while more samples are being taken?

If you folks at digilent are ever thinking about redesigning the Analog Discovery for a 2.0 edition, I would certainly be willing to pay more money for  either 1) a larger internal memory to store more than 16K, or 2) the ability to truly stream the samples to PC (USB 3.0?), and thus only be limited by the amount of PC ram I possess.  I am not sure what the realistic capabilities and limits are of USB 3.0 but surely there would be an equilibrium sampling speed that could be attained indefinitely while transferring over the the PC.

Thanks again attila!

[attila]

In streaming or recording mode, to capture more samples than the device buffer, the data is streamed through USB. The bandwidth of USB limits the achievable sample rate.