10-31-2011 03:06 PM
I'm assembling an audio testing 'workbench' based on NI hardware and custom C/C++ software. The primary purpose is to test all characteristics (THD, Freq Resp, etc) of input and output audio channels of our products. Our products utilize 16 bit codec devices and provide an electrical interface to a monaural headset. The workbench may also be used for headset testing.
Is there any documentation to provide a comparison of the DSA hardware (e.g.. 4461 board) to non-DSA hardware in this application. The price differences are significant. One obvious technical difference is that the non-DSA boards are mostly 16 bit input, with a couple of 18 bit boards. I'm hoping to better understand what the additional cost of the DSA hardware is providing.
Again, my application will mainly be custom s/w in C/C++, but we may also be using LabVIEW/TestStand for other applications.
Thanks.
11-01-2011 03:17 PM
The major difference with our DSA cards is that they use delta sigma modulation. It is this converter that is responsible for the features that make DSA devices well-suited for frequency domain measurements in audio testing. These features include excellent linearity, dynamic range, and transparent digital anti-aliasing filters.
While we don't have any documentation that compares DSA to non-DSA, I would recommend reading the relevant portions of the NI Dynamic Signal Acquisition User Manual to learn more about these features and their relevance to your application.
http://digital.ni.com/manuals.nsf/websearch/C7F1AE47873FDE428625773700589591
Please let me know if you need any more clarification on about the advantages of DSA.
11-01-2011 03:25 PM
Thanks. I'll do more reading and ask additional questions as necessary.
11-10-2011 04:02 PM
As I'm diving deeper into my audio testing system design I have a few additional concerns and questions.
Our device testing can be divided into at least two categories. The low or electrical level testing undoubtedly can take advantage of the accuracy and low noise of the DSA modules (e.g. NI 4461). However our higher level testing does NOT need the 24 bit resolution or low noise floor. In addition, the high level tests require digital I/O, timers, and triggers and seem to more in-line with the multi-functional DAQ modules. Perhaps even a USB module for portability.
All of that said, my question/concern is about analog output bandwidth on various module (DSA and DAQ families). One use-case for an analog output to be sent a 'wav' file of audio samples. Do the modules specify some max bandwidth? Not the DAC sample rate, but rather a max bandwidth of the interface (perhaps USB) and the drivers. I might want to have two streams of the audio samples from two analog output channels at something like a 40ks/sec rate. I'm using the modules as a high quality and accurate 'sound card'.... Do these specs exist?
Thanks.
11-11-2011 12:23 PM
Hi again,
I think this is what you are looking for.
http://zone.ni.com/devzone/cda/tut/p/id/2876
http://digital.ni.com/public.nsf/allkb/F7DEF910D0728ACC86256BE2006FA78C?OpenDocument
The first link describes the different bandwidths between PXI and PXIe. It also has a good graphic that compares all bus types.
The second link is pretty much the same thing but provides a little different information.
11-11-2011 01:19 PM
OK, so the point is that the constraint is purely bus bandwidth (and ability for host to support it).
So I should no problems with a NI 4461 PCI device running two analog input and two analog output streams at 40 kS/S. Correct? Or a USB-4431 USB module running two input and one output stream (again at 40kS/S)? Or maybe the USB-6212 running 2-in and 2-out? Or maybe toi USB-6343? Do the new X series offer increased performance?
Thanks. Perhaps I need to acquire some modules and build my tests to validate the performance. I don't see another good way to do so from specs alone. There are probably too many variables such as host performance.
11-14-2011 03:55 PM
Yes that's correct. The other limitation, that you'd have with any card, is that you can only have one task of one type running at a time. For example you can have an AI and an AO task running, but not two AO tasks. You can write to two AO's if they are in the same tasks however. You should not have a problem with the 4461 if you plan on combing the AO's into one task.
X series has hardware timing, increased sample rates, extra counters, etc. Both the M-seires and X-series should work for your application.
11-16-2011 08:29 AM
OK, I understand the one task/function limitation. Thanks for pointing out this limitation. Since I don't yet have the H/W and only writing test specs and instrumentation specs, I at first assumed that the two AO channels would be separate streams. I'll need to adjust my requirements for test scripting. I always assumed that I'd capture the two AI channels in the same 'thread'. Thanks.
11-16-2011 10:03 AM
Just so you know, you can simulate the the 4461 hardware in Measurement & Automation Explorer. This will allow you to develop and test your application while you wait for the actual hardware.
http://digital.ni.com/public.nsf/allkb/75CF478A58545DFC86256FCC006E25A2?OpenDocument
Good luck!
12-20-2011 03:17 PM
One thing you should note is that most of the Multi-function DAQ devices have nasty glitches on their outputs. Check out the M-Series DAQ series data sheet. The glitches are somewhat hidden in the documentation, but if you put a scope on them it's very obvious. Glitches make the DAQ series outputs unusuable for precision measurements audio measurements, so unless you can tolerate lower precision, or figure out how to filter away the glitches, your stuck with a DSA card.
Settling time, full scale step 15 ppm (1 LSB) |
6 μs |
|
Slew rate |
15 V/μs |
|
Glitch energy |
|
|
Magnitude |
|
100 mV |
Duration |
|
2.6 μs |