Multifunction DAQ

cancel
Showing results for 
Search instead for 
Did you mean: 

modeling 4461 card

Bonsoir EBL 🙂

>> the model of the fixture impairment (including the DACs, ADCs, cables, capacitors, wires, fingers, noses, and all) is one s-parameter matrix.

I'm sorry, but, I miss something. I understand what you explained, but I miss this sentence. What do you put in the term "fixture impairment", is it all the power loss due to these components ? In this case, this means that all the cables are inside a two port device. I suppose that on one port you put the DAC and  on the other port you connect the dut. But if you model like this, where are connected the two ADC ? If you model all the circuit using a s-matrix, you have to choose a four port terminal (4x4 matrix). In the other case, you must assume that the two adc inputs are perfect, I think you do not make this assumption.

In the second model, I speak today, we can choose to have one  two port device wich is only a 1k resistor and whose characteristics are known. And then, we consider that the fixture impairment is modeled by some measurements errors around this 1k resistor. 

To my mind, but I'm quite novice, the doc of hp uses the second model. To my mind, but I'm absolutely not sure, you seem to work with the first case and consider that an 2x2 s matrix is sufficient to correct the errors. It is unimportant I'm right or wrong. But, I would really want to understand how you work. I will go on with one year of work based on sharp measurements.  So, I need to know what I do. I think, I have to develop the two port calibration after I understand your app. So the discussion is not about maths, RF or labview, but how you model the fixture using s parameters. If you could do a figure... My misunderstanding cames from a little thing I miss.

In any case, merci beaucoup et a bientot 🙂

ps: today, I will speed up to retrieve my pc and have a look at your code



Message Edité par MisterDAQ le 02-12-2008 06:17 PM
0 Kudos
Message 41 of 52
(1,913 Views)
Hi, MisterDAQ.

Take a look at the following article, which does a reasonable job of explaining the reflection measurement:

http://radio.tentec.com/cms-files/mcdermott.pdf

In particular, look at the discussion of reflection calibration (s11), which starts on page 9. The s-parameter error matrix shown in figure 10 represents all of the errors introduced by the non-idealities of the system. Those non-idealities include the instrument itself along with all of the cables and connectors, etc.

If you dig through my code, you'll find a set of "k" constants that correspond to the impairment s-parameters. These k constants are not quite the same as the s-parameters, but my notes show that they correspond as follows:

k0 = Ed

k1 = Er - Ed*Es

k2 = -Es

where Ed is directivity error, Er is reflection tracking (called Et in the article), and Es is source match.

The reason I did this way (a long time ago, I might add) was that it simplified the math. Little did I know that I would have to explain it one day - had I known that, I might have taken the more conventional route.

Cheers,
Ed (directivity error...?)


0 Kudos
Message 42 of 52
(1,893 Views)
Hi EBL,

thx you again for all your information and for being patient.  I just find an interesting  and  self contained tutorial on microwaves. So I have two hundred pages to read. After this, I think I will be more confortable with this work.


"" Little did I know that I would have to explain it one day - had I known that, I might have taken the more conventional route."" !!! We never know, if I
had known that I would study s-parameters one day...

Thx for the pdf, I will study it. I have to confess that I must improve my skills in labview because I only have two weeks of flight on this software. So, see you in one month, or maybe in three days !!!!

Merci beaucoup 🙂




Message Edité par MisterDAQ le 02-14-2008 06:28 PM
0 Kudos
Message 43 of 52
(1,886 Views)
Hi all,
 
Very interesting discussion indeed. I am also working on an impedance measurement project. Your VIs work magic EBL, thanks a lot for sharing.
 
I tried to use FGEN and SCOPE module to perform generation and acquisition so that I could measure impedance at higher frequency (say 100kHz to 10MHz). I just wrote new I/O VIs then integrate into your VIs.
It worked pretty well. However, I encounter 2 problems.
 
One, the resistance readings are quite far off the calculated values. Sometimes the error is greater than 20%, 50%... This is not the case when DSA module is used. What are the possible reasons for this? It is the high frequency or the charateristic of FGEN and SCOPE module being different from DSA?
 
Two, the magnitude graph shows unexpected measurement when the frequency is above 2MHz.
 
 
 
 
Can you give me some advices on this? I will post the VIs if you need to take a look.
 
Regards,
Ho Nam
0 Kudos
Message 44 of 52
(1,772 Views)
Hi, Ho Nam.

There are at least two reasons why you might have trouble with the higher frequencies:

1) It depends on exactly what your hardware is. But if there are no antialiasing or anti-imaging filters, images from frequencies above one-half the sample rate generated by the DAC could be aliased by the ADC, causing impedances at out-of-band frequencies to interfere with measurements at in-band frequencies. See the discussion at:

http://forums.ni.com/ni/board/message?board.id=170&message.id=304473

2) The requirements for electrical and mechanical precision in your calibration standards are much more stringent for 10 MHz than they are for 100 kHz, where my casual standards are starting to run out of gas. The parasitic inductances will become significant at those higher frequencies and will result in faulty calibration constants. Unfortunately once you move into the realm of many MHz, you have to consider using well-controlled calibration impedance standards, such as those found in a vector network analyzer calibration kit. It may still be possible for you to make your own standards, but you'll have to be exceedingly careful to consider the parasitic inductances and capacitances.

And then of course comes the question: "Who shall watch the watcher?" How will you -know- that your standards are good? At 100 kHz there's already no assurance, and at 10 MHz it would be pretty easy to go off track in a hurry.

Hope this helps,

Ed L.

0 Kudos
Message 45 of 52
(1,758 Views)

 

0 Kudos
Message 46 of 52
(1,548 Views)

Hi EBL, here is Misterdaq !! This was a long time ago. I do not forsake, but time was missing and difficulty was high. I succeed in finding the Pozar Book, then with your explanations and the last document you gave me, I think I now understand the calibration for VNA. I'm not very familiar with Labview and the spaghetti style is difficult to read. So I prefered to work with my pen. I undertand what you do in your VI for VNA calibration.

 

Nevertheless, I have a question about the computation of S11m, maybe my last question this time. I put the vi just above (simplified figure). The two vi on the right bothers me because I do not know precisely what they do. So I see you compute these two values V0/2 and V1-V0/2, the first seems to be the incident wave, and the other "your" evaluation of the reflected wave. After that, you compute something for S11...

 

if 

I = Incident Wave = I e i(wt+phiI)

R = Reflected Wave = R e i(wt+phiR)

 

Do your S11 correspond to : S11 = R/I*e i (phiR-phiI) ?

 

thx 

 

 

 

0 Kudos
Message 47 of 52
(1,545 Views)

 

Here is the summary of your equations. Left here for the next generation 🙂 

0 Kudos
Message 48 of 52
(1,544 Views)


Bonjour MisterDAQ,

Your understanding of the code seems to be correct. Indeed, s11m is basically the ratio of the reflected wave over the incident wave. There is only one slight complication: in order to reduce noise, I take a vector average of this ratio. You'll notice that in Refl Acq Multi-Tone.vi, there is a cross-correlation, which is a vector power measurement. By averaging these vectors, the noise in the reflected measurement that is not present in the incident measurement is reduced, depending on how many averages are taken.

 

Hope this helps,

Ed

Message Edited by EBL on 05-20-2009 03:56 PM
0 Kudos
Message 49 of 52
(1,527 Views)

Hello Ed,

 

I am very appreciative of your code and impedance measurement system.  I have previously implemented it with a USB-4431 card with no hiccups whatsoever.  Now I am trying to use a NI USB-6356, however, the analog input configuration is limited to differential. I am having problems with the system.  The calibraion curves all oscillate and when I take a measurement of the load 1kOhm resistor, it is not nearly a straight line at 1000 as I have seen in the past.  I have concluded, rightly or wrongly, that it is a grounding issue.  In your impedance fixture circuit, AO, AI0 and AI1 all share a common ground.  I am not sure how to wire this up with differential analog inputs.  The pinout of the USB-6356 has AI+,AI-, AI_GND.  As the fixture circuit is wired up, AI0- and AI1- and AO_GND are all connected.  I have consulted the manual for the USB-4431 (the card on which I implemented you impedance measurement system successfully in the past) to determine what the difference betwen the USB-4431 and the USB-6356.  The 4431 are 4 input/1 output BNC connections, with no obvious separate ground connections.

 

I am wondering if you could enlighten me on how I might succesfully implement your impedance analyzer with differential analog inputs.  Differential is the only analog input configurtaion available for the USB-6356.  I would have tried NRE if that was available.

 

Any assistance and direction you can offer would be appreciated.

0 Kudos
Message 50 of 52
(1,026 Views)