Looking for USB 4 channels ADC digitizer 10-12 bits 10+ MSamples per channel

[Vasilich2004]

Hi,

I am looking for USB 4 channels ADC digitizer 10-12 bits 10+ MSamples per channel.

NI USB 5132 is more or less Ok but I want a little better.

Any suggestion?

Thanks, Andrey.

[Anthony_F]

Hello Vasilich2004,

We currently only have two USB digitizers both of which only have 8 bit resolution (5132 and 5133).  The sample rate for the 5133 is twice as fast as the 5132.  Please keep in mind that the USB bus is a limiting factor for data transfer and thus affects resolution, max sampling rate and number of channels. 

I am not sure if your setup allows you to have a desktop computer but if you consider PCI devices you can also look at the PCI-5105 that will meet your resolution, sampling rate and channel count requirements.

[Vasilich2004]

Anthony F

Thank you for answer. I know about 5133 and USB limitation.

I searched in internet but I didn't find suitable device. I had hope that someone uses similar device. 

Probably, it is not very correct to ask about any (not only NI) devices in NI forum. But, unfortunately, I don't know another suitable forum.

[johnsold]

Andrey,

Look at your numbers.

10-12 bits/sample => 2 bytes/sample (assuming binary data transfer).

(2 bytes/sample)*(10 Msamples/s) = 20 MB/s per channel.

4 channels * 20 MB/s/channel = 80 MB/s

8 bits per byte * 80 MB/s = 640 megabits per second.  USB 2 has a nominal data transfer rate of 480 Mb/s.  With the overhead the net data rate is less.

Even if you found a device which used packed binary (12 bits/sample) with no overhead you would still be at the maximum rate of 480 Mb/s.

No USB device can do what you want. Maybe when USB 3 is more widely used, but not now.

Lynn

[Vasilich2004]

to johnsold

thank you for your calculation. I would agree with you but this "10-12 bits/sample => 2 bytes/sample" can be improved!

In any case I want "USB 4 channels ADC digitizer 10-12 bits 10+ MSamples" or 40 MSamples and 2 Bytes. I didn't calculated myself because NI USB 5133 is 100 MSamples and 8 bits which is 25% higher than my wish.

I hopes that NI tested NI USB 5133  So ... what will NI developers say?

[Anthony_F]

Vasilich2004,

As johnsold pointed out the main limitation is the bus type you have specified.  You can sample at the highest speed (100MS/s) but don't expect to do continuous sampling at this rate.  For continuous acquisition you will need to sample at a slower rate.  For higher sampling rates you will need a deep onboard memory and a short acquisition period.

[johnsold]

Anthony F,

Thank you for the link.  I expected that something like that was the case, but the 5233 documentation does not mention that limitation.

Lynn

[Vasilich2004]

Anthony F,

Thanks. Unfortunately, my hopes almost dropped 

What about table from here? I supposed to see transfer rate: 1 channel - 16 MSampels / sec / channel, 2 channels - 8 MSamples / sec / channel, and so on ...

What is real maximum transfer rate could be reached? I suppose it should not be depends to much from on-board buffer if buffer is more than few kBytes. Let say CPU is busy less than 50%. USB controller has only one USB device.

[Anthony_F]

@johnsold:  You are correct, the 5132/5133 documentation does not address this issue and the knowledge base article was created in response to this and other customers concerns.

@Vasilich2004: The actual sample rate will vary depending on a lot of factors in the setup.  The 4 test setups for continuous acquisition varied from 7.2MS/s to 8.1 MS/s for single channel acquisition and 6.1MS/s to 8.0MS/S for both channels.  Your results will most likely vary also.

[Vasilich2004]

and that is strange as that table.

based on your "the 4 test setups for continuous acquisition varied from 7.2MS/s to 8.1 MS/s for single channel acquisition and 6.1MS/s to 8.0MS/S for both channels."

single channel has USB rate 7-8MS/s

both channels have USB rate 12-16 MS/s or twice more.

I didn't look USB interface deeply but it looks for me this 5132/5133 issue.