modeling 4461 card

[MisterDAQ]

So, some advice. It's easier to change directly the name of the card in MAX, simply choose your device, press F2 and replace by Dev2.
Then, it's better to extract the llb archive. To do that, run labview, tools->LLB manager, go in the place where you save the llb package, select all the VI, copy and paste to another location. This prevents the CAL file saving from crash. This spares one or two hours of modifications.

Now !! Multitone seems to work. I have something in the ouput. Meter Cal produces nice (and strange !!!) figures, so I suppose it works... But when I use the meter... nothing happens. Z is always zero...

Just a question, when I have to do the "LOAD" test, I must put a 1k resistor in the DUT ? There is always the 1k resistor in the fixture ? In other way, there are two 1k resistors in the circuit during the LOAD test ???

thx a lot 🙂

measurement is really a hard job 🙂

[MisterDAQ]

That WORKS !!!!!!!!!!!!!!!!!!!! yes !!!!!!!!!!!!!!! GREAT !!!!!!!!!!!!!!!!

(after a reboot :D)

ok the app is very nice and work well. So, if I understand well, the meter uses the signal generated by the multitone (the 125 added sines) ? So the graph is built using this 125 points ?

Great job.

Now, after checking the R L C values of a component depending on the frequencies, I want to study the intermodulation noise of the components. I mean, mainly I have three sines (900Hz 1100Hz , 1300Hz). The amplitude of the 900Hz is the response of a physical phenemenon produced by the 1100 and 1300Hz. The problem is that the amplitude of the 900Hz is 60dB under the 1100 Hz and 1300Hz. The main danger for me is that semiconductor components naturally produce 900Hz when they are traversed by the 1100 and 1300. So I have to check this by myself.

So, I suppose I have to understand signal analysis reflectance in order to use the same fixture and adapt it to my problem or just use the existing app and modify it (maybe)

Ouch

I obtain a strange graph for a 1uF cap, from where can come the problem ? Inductors in the cable that produces a resonating signal ?

Message Edité par MisterDAQ le 08-24-2007 08:03 AM

[MisterDAQ]

I obtain a strange graph for a 1uF cap, from where can come the problem ? Inductors in the cable that produces a resonating signal ?

[MisterDAQ]

Problem found, sometimes the acquisitions failed... I have to open Multitone vi and to run it. If I close the block diagram before opening the meter, the acq fails... so its ok.

Little modification :  Mag Z Displays impedance magnitude vs. frequency and LC value: Displays the L, C, or R component of the impedance vs. frequency, as selected by the L, C, and R buttons. But, for L => we display Zl(w)/w and for C Zc(w)*w 🙂

Message Edité par MisterDAQ le 08-24-2007 11:04 AM

[MisterDAQ]

I prepare my fixture. Nevertheless doing many calibrations, I notice that the cal change as soon as the cable moves from some centimeters... so It's quite complicate. At the beginning I unterstood that if I change something in the fixture I have to calibrate again but, now, I think that the location of the cables greatly influences the calibration... difficult

Message Edité par MisterDAQ le 08-24-2007 12:10 PM

[EBL]

Wow. So many questions. Here goes...

>>Just a question, when I have to do the "LOAD" test, I must put a 1k resistor in the DUT ? There is always the 1k resistor in the fixture ? In other way, there are two 1k resistors in the circuit during the LOAD test ???<<

Yes, the load calibration uses your nicest precision 1 k resistor in place of the DUT. However, the other 1 k resistor on the fixture itself need not be accurate at all, as the calibration will take care of that. It should be stable, though.

>>ok the app is very nice and work well. So, if I understand well, the meter uses the signal generated by the multitone (the 125 added sines) ? So the graph is built using this 125 points ?<<

Yes, the graph consists of just those 125 points, which are the frequencies of the multi-tone.

>>Problem found, sometimes the acquisitions failed... I have to open Multitone vi and to run it. If I close the block diagram before opening the meter, the acq fails... so its ok.<<

Yes - you probably have to leave the multitone VI open. Even though it's finished running, you probably have to leave the front panel in memory. That's a driver issue - not my area of expertise.

>>Little modification :  Mag Z Displays impedance magnitude vs. frequency and LC value: Displays the L, C, or R component of the impedance vs. frequency, as selected by the L, C, and R buttons. But, for L => we display Zl(w)/w and for C Zc(w)*w <<

I'm not sure I understand this one. The VI should display L and C correctly already. What's the problem here?

>>I prepare my fixture. Nevertheless doing many calibrations, I notice that the cal change as soon as the cable moves from some centimeters... so It's quite complicate. At the beginning I unterstood that if I change something in the fixture I have to calibrate again but, now, I think that the location of the cables greatly influences the calibration... difficult<<

I did warn that stability in the fixture is important. But if you make your fixture well, you shouldn't have a problem. My fixture is built on a circuit board with a ground plane, with three BNC connectors mounted directly on the circuit board. Moving the coaxial cables that connect the circuit board to the 4461 shouldn't degrade the measurement. The only thing that seems to make a difference in my situation is moving the actual test leads and DUT with respect to ground, especially the positive lead, since it's exposed. But even that only makes a difference when measuring very high impedances. And the closer you join the AI1 / fixture connection to the DUT, the less difference you'll see at very low impedances.

No matter what method you use to measure impedance, the quality of your bridge fixture will determine the resolution and stability of your measurements. It's worth investing the time to make it nice.

What does your test fixture look like? Can you post a photo?

Ed

[MisterDAQ]

>>Little modification :  Mag Z Displays impedance magnitude vs. frequency and LC value: Displays the L, C, or R component of the impedance vs. frequency, as selected by the L, C, and R buttons. But, for L => we display Zl(w)/w and for C Zc(w)*w <<
>> I'm not sure I understand this one. The VI should display L and C correctly already. What's the problem here?

yes, you re right, this is L(w) and C(w). But when I select magnitude, this is Zc(w)=1/jCw or Zl(w)=jLw, so this is not the same graph.

I will post a picture soon. My cables are long, nearly 1 meter, so I suppose this is not a good thing for the measurement.

--theoretical impairment of a reflection measurement (from signal flow-graph theory)-- Hum, I'm always interested in this field, maybe you can advise me some references ? I'm sure there is nothing on wikiped 🙂

Have a nice we

Message Edité par MisterDAQ le 08-24-2007 05:59 PM

[EBL]

Attached are two pictures of my fixture for measuring impedance with the 4461, along with the schematic diagram.

The first picture shows the bridge fixture. It's on a pre-perforated circuit board with one side of ground plane. If I had enough time and energy, I would build a shield around the exposed components. Nonetheless, it's neat and tidy, and it gives me very few problems being exposed as it is. I try not to move it around much when I measure, but it really doesn't make a whole lot of difference.

The second picture shows the connection to the DUT (in this case the DUT is my reference 1 k resistor). Note the BNC T-junction in the background, where one coaxial cable goes to the fixture and the other cable goes to AI1. The junction is fairly close to the DUT, but this configuration is actually more for convenience than for precision. When I need to measure very low impedances with precision, I actually make the junction right at the DUT, so there's four clip-leads on the DUT. That pretty much makes a Kelvin connection, which you would normally use for low impedances. Sense (AI1) goes closer to the DUT and force (from the bridge) goes just outside the sense connections.

Everything else not in the pictures is just BNC coaxial cables to and from the 4461. They're all either 3 or 4 feet (0.9 m or 1.2 m) - the length is not critical, though I suppose shorter is better. In any case, 1 m cables are no problem.

I wish I had a good reference on flow-graph theory. I tried searching for flow-graph theory and flow-diagram theory, but really didn't come up with anything. If I find something really useful, I'll post a reference.

Ed

[MisterDAQ]

Hi EBL,

I'm now restarting my work on the DUT we spoke about 6 month ago. From one week, I try to understand the notion of "flow graph theory" you use in your program. This is quite new for me, I confess. It seems that this chapter is linked to the "mason's rule". But, from what I see, this problem seems to be generally studied for signals of more than 1Ghz in order to model reflection... So I'm unable to see the link with your (working, im sure) audio application. I find this doc on the net : http://canopus.icu.ac.kr/~CoSMIC/lecture/mmic/iii-5ho.pdf . Can you confirm that the notions presented in this pdf are the key points to understand the theory used in your app ?

Thx for your help.

Im interested in this theory, because I will have to use fixtures with very long cables and I'm sure that I will have a lot of measurement error due to their locations between each others. As you app is able to correct this problem, I think I can reuse this bright idea for my problem.
 

Message Edité par MisterDAQ le 01-17-2008 08:36 PM

[EBL]

Yes, that's the theory exactly. Reflection and transmission measurements are commonly made at radio frequencies, but there's absolutely no reason the technique can't be used at audio frequencies as well. And so I do.

The basic method of correction is more or less described in this app note, especially on pages 4 - 7:

http://www.smi.oeaw.ac.at/widmann/hfs-lit/hp-vna-error.pdf

The case we're concerned with in our application is a one-port correction, but the two-port diagram on page 6 gives the general idea. You can likely find some more in-depth references on the web as well.

The key is to consider your fixture to be an impairment that's been inserted between an ideal measuring instrument and the Device Under Test. Calibrating using the three standards measures the parameters of the impairment and allows us to correct for it, as though it were never there.

Hope this helps.

Cheers,
Ed