10-01-2010 04:20 AM
What Kind of Analog-to-Digital Converters Are Used in X-series USB-6356?
SAR or Sigma-Delta?
Will there be obvious difference between 32M and 64M version of USB-6356 when acquiring 8channel signals with 1.25MHz sampling speed?
10-05-2010 04:22 AM
Can anyone can solve my puzzle? Thx!
10-05-2010 11:11 AM
Hi twdonkey,
The 6356 uses SAR ADCs.
The difference in performance between the 32M and 64M devices is a much less straight forward question.
As soon as data is acquired, it is placed in the onboard FIFO of the device. The device then continually trys to flush it from the onboard FIFOs to the host machine. However, it is possible that other traffic on the USB bus may temporarily starve the USB-6356 of bandwidth. When this happens, data will back-up into the onboard FIFO until it becomes full. If the onboard FIFO overflows, the device will error and the AI task will be aborted. Having a larger onboard FIFO gives the device a better chance to deal with spikes in bus traffic.
How much bandwidth is available is highly system dependent. All devices attached to the same USB host controller share bandwidth, so other devices in your system can heavily impact the USB bandwidth available to the 6356. Also, some USB host controllers have much more refined implementations then others, so the chipset used by a system can also dramatically effect bandwidth. If that's not enough, the operating system and other host software intervene in USB transactions, so the speed of your computer and how busy the CPU and memory are at any given time can effect USB bandwidth.
Due to the large number of variables involved, it is very hard to say how large of an onboard FIFO your system will need to sustain a continuous transfer of all 8 channels. During development of USB X-Series devices, we tested a long-running continuous acquisition using all channels of a USB-6366 (which has 2MHz ADCs) on a fairly modern desktop machine. The USB-6366 was the only device on the USB host controller used, and the system was not doing anything other than running LabVIEW to acquire data with a single USB-6366. In this setup, we were able to sustain the transfer for several days on both the 32M and 64M versions. However, I want to emphasize again that it is very difficult to know how much onboard FIFO space your system will need to sustain a continuous transaction.
Another use case to consider is finite acquisitions. If you attempt a finite acquisition larger than the onboard memory size, data will need to be transferred from the device to the host before the acquisition is complete, so the same bandwidth concerns come into play. This means the only way to be 100% sure a finite acquisition on a USB device will be able to acquire all requested samples is for all of the samples to fit in the onboard FIFO. So, the 32M FIFO can hold about 3.2 seconds worth of data for the 6356 using all 8 channels running at the max rate, and the 64M FIFO can hold about 6.4 seconds worth of data.
William Earle
National Instruments R & D | Staff Software Engineer | DAQ Software