Confused about number of analog tasks on cDAQ

[jjjcjj]

I'm trying to build a new compactDAQ system to perform simultaneous capture of 64 analog channels.  I'd like to use the eight NI9222 modules.

Is it possible to share a sample clock, and capture samples from, 8 of these units at the same time in the same chassis?

I'm confused because when I look at the knowledge base I see articles saying that the maximum number of analog tasks is only 3.  I'm not sure if this would be more than one task or not.

[crossrulz]

You can put all of the channels into a single task.

The maximum number of analog tasks would matter if you needed different sample rates, each task having its own sample rate.

[jjjcjj]

Excellent, thank you.

Are there any throughput concerns?  For an 8-slot ethernet cDAQ chassis would there be any problem simultaneously capturing 64 channels at 500ksps with eight NI9222 cards?  That's roughly 500 megabits per second total.

[crossrulz]

---

@jjjcjj wrote:

Are there any throughput concerns?  For an 8-slot ethernet cDAQ chassis would there be any problem simultaneously capturing 64 channels at 500ksps with eight NI9222 cards?  That's roughly 500 megabits per second total.

---

Yeah, that would be a concern.  However, it might actually be worse than you think, depending on how the data is transmitted from the cDAQ to the PC.  Assuming the values are sent as doubles, you would get 500000 S/sec/channel*64channel*8bytes/sample*8bits/byte = 2Gbit/s.  Assuming the raw 16 bits are transmitted, then you are looking at the 500Mbit/sec.  Either way, I would be concerned over even a gigabit Ethernet connection.  You might want to consider using a PXIe or PCIe DAQ card to do your acquisition.

Though, 8x NI-9222 will only give you 32 channels.  So what are your actual requirements as far as sample rate and number of channels?

[jjjcjj]

Oh boy, that was a mistake.  The actual requirement is 64 channels.  Ideally 200ksps.  I can likely make do with 100ksps but it would require that I add more filtering.

[crossrulz]

Based on your channel count, I would be tempted to give a USB-6255 a try.  It can have 80 single ended channels with data rates up to 1.25MS/s, still with a 16-bit ADC.  It looks like it needs external power.  I would also go for the screw terminal version instead of the VHDCI connectors.

[jjjcjj]

Differential was also a must that I forgot to list, but it looks like the USB-6349 can handle that.  It's only 32 channels but it appears that the trigger and sample clock can be shared via PFI signals, but I'd have to check on that more.

At 500ksps, 2-bytes, 32 channels, that's 32 megabytes/second.  That's well within the USB 2.0 spec, but I wonder if its actually achievable, especially with two units where its possible they could share the same USB hub internal to the PC.  Seems like something that might have to be tested.

[Jerry_X]

USB 2.0 devices on the same USB host controller share bandwidth. USB 2.0 in reality tops out around 30-32 MB/sec with NI USB devices if you are using a good USB 2.0 host controller. There are PCIe cards with multiple USB host controller chips on them in the market, but they are rare.

[santo_13]

Looking at your high throughput requirements, you have to move towards PCIe or PXI based hardware for reliable high speed capture

[Jerry_X]

Another alternative is USB 3.0 cDAQ chassis like cDAQ-9179 - NI, 14 slots of NI 9222 (4 channels), that is 64 channels and USB 3.0 will be able to meet your bandwidth requirement.