modeling 4461 card

[EBL]

Hi, MisterDAQ.

The 200 pF impedance is not the limit of the measurement - the calibration accounts for that. That said, the 200 pF does challenge the robustness of the measurement. For example, if your fixture is not stable, the calibration will not cancel the parasitic effects (such as the 200 pF) perfectly, and indeed you will see a limiting capacitance (though it will be MUCH less than 200 pF - maybe 0.2 pF). In any case, it is not necessary to know the exact input or output impedances of the DSA board - the calibration routine will measure those to the extent that they affect the final impedance measurements.

Four things can compromise the quality of the calibrated impedance measurements:

1) Stability of the fixture. If the fixture changes in any way between calibration and measurement, the results will be shifted. This includes, for example, if the leads have more capacitance to a nearby ground when they are moved.

2) Quality of the calibration standards. Chances are your short and your open are very good, but you also have to make sure your 1 k standard is very accurate and has no parasitic inductance (or capacitance).

3) Time of measurement. Because this is a vector measurement, the quality of the measurement improves with the number of vector averages. For measurements near the extremes of impedance (i.e. milliohms or Gigohms), more averages will be necessary. But perhaps worth the wait.

4) Stability of the gain ratio between the two DSA input channels. This is the one thing that's out of the user's control. In my experience with the 4461, this has limited the dynamic range to the region between about 1 milliohm and about 1 Gohm. Anything that will improve the thermal stability of the DSA card will help this, such as keeping it away from other cards that generate heat.

Enjoy your holiday. I should have some code for you by the time you return.

Ed L.

[MisterDAQ]

<<but you also have to make sure your 1 k standard is very accurate and has no parasitic inductance (or capacitance).>>

How can I know that ? Special resistor ?

[EBL]

Well, admittedly it's a case of "Who shall watch the watcher?". How -do- you know that your impedance standard is good?

In my case, I have a couple of resistors for which I've established that the parasitics are low enough that they don't affect my measurement requirements, and I've cross-checked the results with other impedance measurements. As a practical matter, I think any non-inductive resistor that's thin film, metal film, or bulk foil should be sufficient. By non-inductive, I mean that the inductance should be well less than 1 uH in order to prevent impedance errors at frequencies up to 100 kHz; at 100 kHz, even 1 uH is 0.628 ohms, so you really want something that's on the order of 10 nH or less. My resistors happen to be axial, leaded precision metal film. They're probably not as low as 10 nH, but they're probably less than 50 nH.

Ed

[MisterDAQ]

Hi EBL,

so im back and ready to test 🙂

[EBL]

Hi, MisterDAQ.

OK, the VIs are in the attached LLB. No doubt some explanation is required, so also attached is a file describing all the VIs. Let me know if you have any questions or problems. Have you built your fixture yet?

Ed

[MisterDAQ]

Thx a lot for the job

This week, I change my desktop and then I have to upgrade to labview 8.2 because your file were done with it. I was working with the 8.0 version but it seems to be unable to open them. So tonight it works 🙂 good. I will see tomorrow the inner machinery but your code seems to be clean.

At the moment, only one thing bothers me. You seem to work at 52kHz (sampling rate). Why ? I mean why do you choose to work 4 times slower than the maximum speed allowed by the card ? To limit the size of the samples ?

cheers

lilian

[EBL]

>>At the moment, only one thing bothers me. You seem to work at 52kHz (sampling rate). Why ? I mean why do you choose to work 4 times slower than the maximum speed allowed by the card ? To limit the size of the samples ?<<

Fear not. The sample rate will be 204.8 kHz. The only place you might notice indicated sample rate in the whole system is on the multitone generator window. The Parameter cluster on the right is an output, which will indicate the frequencies and amplitudes of the tones in the set, as well as the sample rate for all measurements (which happens to default in the cluster to 51200). But notice you actually set the sample rate with the "Sample Rate (Hz)" control on the left, which is defaulted to 204800 Hz. Once you run the generator, you'll see 204800 pop up in the Parameter window.

Happy measuring,
Ed

[MisterDAQ]

sorry, I miss this

I look to the DKCalcSol vi. Apparently, using it, you create your own calibration constants. But, I do not see how you model the fixture. I mean, we consider that there exist complex impedances around the DUT, but how do you model that ? Apparently three constants are sufficient to solve the entire problem. But, what do they correspond to ?

sys

Message Edité par MisterDAQ le 08-23-2007 08:28 PM

[EBL]

That's correct - three constants (per frequency) are sufficient to model all the linear imperfections in the fixture, and in the cabling, and even in the 4461. Three calibration measurements are taken to solve for three unknowns.

The three constants don't have an easily-interpreted physical meaning; they're mainly structured the way they are for mathematical convenience. Now, there is a one-to-one mathematical mapping between these three constants and the three constants which model a theoretical impairment of a reflection measurement (from signal flow-graph theory). Those three constants, roughly speaking, model the fixture's leakage from the source to second input channel, the fixture's attenuation of the signal, and how close to 1 kohm the fixture appears to the DUT.

So there's really no a priori model of the system that's been created, other than to model all possible impairments generically in the form of the three constants, which are calculated and stored during the calibration measurements.

Hope this helps,
Ed

[MisterDAQ]

<<theoretical impairment of a reflection measurement (from signal flow-graph theory)>>

Ok. The only thing I know in this field consists in modeling a circuit using Kirchoff current and voltage law... So very poor relative to this. I look on the net and I find nearly nothing. I would want to understand the concept, have you any doc for that ?

thx