Issue with Signal Tap dropping data samples throughout capture

[Altera_Forum]

Greetings, 

 I am using a Cyclone V FPGA in a motor control application. Over the past year, my colleague and I have noticed that signal tap consistently fails to capture clumps of data. We tend to monitor things that look like sine waves or triangle waves. It is particularly easy to notice dropped data in the time history of this type of data. Typically our data looks good for several thousand samples, and then we loose somewhere between one and ten samples across all of the variables we are tracking. This results in a discontinuity in our sin/triangle waves (which is very easy to see). In a 32k depth capture, we typically see roughly three clumps of dropped data. The data is often dropped at around the same point in the capture from one capture to the next.  Here are some details on system:  FPGA: 5CEFA7F23C8 Quartus: 14.0 USB blaster: USB blaster I, USB blaster II (both have the same problem) Computer: Lenovo thinkpad running Ubuntu 14.04 LTS  Is this a known bug? Is there a known solution?  Thank you in advance,  Michael Norcia

[Altera_Forum]

Attached is a screen shot showing the issue described above. The data from this shot was exported from signal tap. We verified that the discontinuity is present in signal tap (thus it is not an issue that occurs during exporting data from signal tap and importing it into matlab). 

 http://i41.photobucket.com/albums/e295/thefonzer/discontinuity_zpselfhjr0l.png (http://s41.photobucket.com/user/thefonzer/media/discontinuity_zpselfhjr0l.png.html)

[Altera_Forum]

Attached is an example of the phenomenon described above.

[Altera_Forum]

I've always gotten occasional hiccups with SignalTap over the years. Probably most often when doing continuous captures. I've always attributed this to the complexities of USB + JTAG and just lived with it. It doesn't happen often enough to really hinder debug for me, but I can see that it might be more of a problem in some situations.

[Altera_Forum]

Windows isn't very good at guaranteed throughput real time operations unless the drivers and apps are written very carefully. One thing that can help is minimizing the amount of background stuff going on. Virus checkers are some of the worst offenders, followed by crapware/adware/spyware.  

 For best results do your data collection with a pristine PC: 1) Start with a fresh windows install, use a windows CD, not a vendor CD preloaded with crapware. 2) Never enable automatic updates or checking for updates.  3) Don't connect the PC to the internet for any reason.  4) Use only passive virus scans. Active scans will cause data losses as checks are done in the background. 5) Transfer files with USB sticks, or SD cards, or a network that has no connection to the internet. Virus check all media before it is inserted into the pristine PC.

[Altera_Forum]

Does the sample point loss happen in a single run or are you using repeated acquisition (autorun)? In the first case, do you see the same faulty data when reloading it by the "Read Data" button? Are you using a storage qualifier and are you sure that it doesn't involve timing violations?