Sampling Rate for FieldPoint

[Ching_F.]

Hi Rolf,

Can you please elaborate on what you mean by real sample frequency?  Are you talking about the effective sampling rate for each channel on the module?  Also, will you be using the serial communication with only the FieldPoint driver, or are you planning on interfacing with other communication protocols such as ModBus and/or CAN? 

[hemmerling]

Thanks alot,

1)
real  = effective sampling rate ( sorry for my bad english )

2)
In another thread I learned that the Effective Sampling Rate for all channels
is 0.97 Hz for the  cFP-RTD-124,

and for the cFP-AI-112, there is a formular

Update Rate

The update rate of the [c]FP-AI-112 is determined by the filter

settings and the number of input ranges in use. Higher filter

settings and fewer input ranges yield faster update rates. Use the

following formula to determine the approximate update period:

(number of channels with 50 Hz filter) × 0.061 s +

(number of input ranges with 50 Hz filter) × 0.12 s +

(number of channels with 60 Hz filter) × 0.051 s +

(number of input ranges with 60 Hz filter) × 0.10 s +

(number of channels with 500 Hz filter) × 0.007 s +

(number of input ranges with 500 Hz filter) × 0.012 s + 0.16 s

= Update Period

which results in sample frequencies
8 (channels with 50 Hz filter)  * 0.061s  +
1 (input range, as all channels have similar input sources) * 0.12s
+ 0.16
= 0.728 s.
==> 1.37 Hz sample frequency
Right ?

8 (channels with 500 Hz filter)  * 0.007s  +
1 (input range, as all channels have similar input sources) * 0.012s
+ 0.16
= 0.228 s.
==> 4.38 Hz  sample frequeny
Right ?

So with acitvated 50 Hz filters we do not get "almost 50 Hz or 25 Hz effective scan frequency" :-(.

So the ONLY chance to get the 10 KS/s is NOT to activate ANY internal filter ( 10 Hz, 50 Hz, 60 Hz, 500 Hz ), AFAIK.

3)
Now HOW MANY samples can I transfer by the serial RS-232 device,

which is mandatory for communication just-because the vendor of a certain compiled Labview application offers it as one-and-only communication for temperature instrumentation, so no Modbus, Can, Ethernet.

AND its my job to say "take it or leave it" ( the compiled software, and the Fieldpoint hardware with cFP-1804, so WITHOUT cpu on board .
So I need ARGUMENTS against this Fieldpoint solution ( in favour of a more obvious PCI solution, with the need to write own Labview software instead of buying the readymade compiled software ), and thats
a) the Effective Scan Rate - which can be calculated, ok
b) the samples/s transfer rate of the RS-232 line - which is unknown so far

So from what I learned the Labview application on the Windows PC must setup a "timed loop" and ask the Fieldpoint cFP-1804 frequently about the values of the analog modules.
So now with 10KS/s of one module, even of another 10KS/s of another module,
how much data can I transfer by the Labview timed loop, by the serial RS-232 communication ?

This is of course the difficult question:
How much time needs the RS-232 protocol to transfer a (16-bit ?) analog value to a labview application ?
So how much overhead is there ?

For the temperature measurement, we have the unusual demand for 10 Hz sffective sample rate  of the data, so ON THE LABVIEW PC ( it does not matter what the Fieldpoint module does).
But if we decide to take the Fieldpoint system,
and believe the bashing of the vendor of the compiled Labview application for WindowsPCs who just supports Fieldpoint-without-cpu-and-with-RS232 AGAINST PCI solutions ( they had bad experience with PCI boards and THEREFORE do not offer PCI solutions but JUST Fieldpoint with RS-232),
we must argue the next step against Fieldpoint AGAIN with the next project, where faster sampling is needed.
So to argue against Fieldpoint without CPU module and with serial RS-232 communications,
this project ist the first line of defence...

Message Edited by hemmerling on 01-22-2007 10:55 PM

[hemmerling]

 I wrote
How much time needs the RS-232 protocol to transfer a (16-bit ?) analog value to a labview application ?

For example, is the data transfered in binary form, or like with GPIB /IEEE-488, as "text" ?
If transfer as text, this would decrease the samples/s which can be transfered by the serial RS-232 line, very much.

Or maybe there is a "fieldbus protocol overhead".

Please tellme.

What is your experience, how many sample items can be transfered by RS-232, per second ?


Sincerely
Rolf

[Ching_F.]

Hi Rolf,

I have looked into this some more, and have found that the AI-112 will be unable to meet your sampling requirements of 10Hz.  The best that it can do appears to be approximately 13Hz. 

As for the AI-118, it is possible to achieve the sampling rates you are looking for.  The all channel update rate according to the manual is 3.45kHz with a 10Hz filter and 10.42kHz without a filter.  To determine the effective sampling rate, a few other things need to be taken into consideration.  The effective sampling rate needs to consider the all channel update rate, the communication time, as well as the software loop rate.  In most cases, the communication rate is negligible compared to the all channel update rate of the hardware, and not factored when calculating the effective sampling rate.  If you are interested in factoring that in, you can assume that all 8 channels need to transfer 16 bits of data.  Use your baud rate, divide by the amount of data being transferred over.  The protocol beneath the FP Read is Optumux, so you can try to find the overhead added.  If you do not want to do that, you can assume that it is in the micro-second range.  To find a precise effective sampling rate, you must add the software loop rate, communication time, and the all channel update rate. 

Please let me know if you have any further questions.

[hemmerling]

Hello,

Ching P. wrote
"I have looked into this some more, and have found that the AI-112 will be unable to meet your sampling requirements of 10Hz.  The best that it can do appears to be approximately 13Hz."

Hmm.. 10 Hz is more than 13 Hz 🙂

1)
The
AI-112 manual
manual contains a special formular of the effective scan rate, resulting usually in a small scan rate,
while the AI-118 manual does not ?
So as I don´t want a 10 Hz filter ( which prevents scans of more than 5 Hz ), I really get 10 Khz with the AI-118 while I don´t get this with the AI-112 ?

2)
Unfortunately you did not answer the question  how many samples per second I may transfer by serial RS-232 line.
I hoped that you can name  a typical "experimental" number,
instead of naming parameters and influcence factors, as I can do in theory too.
So might you ask at NI internally whats a typical value ?

Remember that with any other PCI or USB system, its very easy to name the number of samples / second,
the same with a Fieldpoint system with CPU onboard.
Its JUST that the cFP-1804 has no CPU so the data MUST be transfered by the serial RS-232 line...
and JUST that this is an offered "solution" in a very special application...
thats why I ask.

Sincerely
Rolf

[Ching_F.]

Rolf,

I apologize for the mistake I made.  I meant to say 3Hz, not 13Hz.  The fastest sample rate is 0.27s, or 3.7 Hz.

1.) That is correct, you should be able to achieve 10.42kHz without filtering when using the AI-118.  You will not be able to get this with the AI-112.  They are two different modules that were designed with two different specifications.  We have different modules with different specifications to meet the needs of a wide range of applications.

2.)  I cannot provide you a typical value for the number of samples you can transfer per second via the RS-232 line because again, this is code dependent.  R&D does not provide a typical value for this as code may vary greatly from one application to the next.  The amount of data you can transfer is highly dependent upon the loop rate of your code.  I would recommend asking the providers of this "offered solution" for these benchmarks.  They would be better able to provide you with these numbers as they have the loop rate information.

Please let me know if there is anything else I can do for you.  Best of luck on your application, and have a wonderful weekend!!!

[hemmerling]

Thank you very much for naming to the difference between AI-112 and AI-118.
Thank you very much for describing the difficulty to determine a sample transfer rate for the RS-232 connection.
We understand now that we must ask the solution provider.

Just for all the other readers:
The solution provider told us, that his compiled Labview application running on a Windows PC with Fieldpoint cFP-1804 and temperature module, connected by RS-232, can do a scan of 10 Hz .. 20 Hz in normal mode, and 60 Hz in a special "ramp mode".
It is not obvious if that limit is by the software or by the RS-232 device,
at least it is a limit which cannot be exceeded with the application at all ( even for different purposes than temperature measurement, and no matter what effective scan rate the measurement module offers ).

Sincerely
Rolf