How to design and characterize the Test Fixture?

[ninhawk]

Test Fixture: how to?

Hi guys:

...anybody has information about how to design and characterize an Test Fixture?

...mainly from microwave frequencies, but it would be interesting in mmw also.

Thanks a lot. Best Regards

ninhawk

 

[nandopg]

Dear friend,

Take a look in the this App Note from Agilent where an approach to built a test fixture for R&D is described. You can use this approach to characterize active and passive devices.
I hope this help you.

NandoPG

 

[harkonnen]

Re: Test Fixture: how to?

Hi!

What do you want to do? If you want to design a Test-Fixture for prototype devices then TRL Calibration Boards would be the best.
If you want to deembed a production-test fixture you will have to do more work or find a way to perfom a LRM calibration.

So it really depends on what your goal his.

Greetings

 

[yingyang]

test fixture

You may like to apply the black box de-embedding technique instead of using TRL.....

 

[harkonnen]

Re: test fixture

yingyang,

What do you mean by black-box de-embedding?
Do you mean measuring the S-parameters of the input and output part of the fixture respectively and then inversing the blocks in a linear simulator?

That only works if you have connector access to each half of the fixture.
What do you do if you have a MMIC footprint?
Measuring the complete fixture by replacing the DUT with a Through requires symmetry of the fixture and knowledge of the THRU in order to split the S-Parameters.

Is there any easier or better method?

Regards

 

[nandopg]

Dear Friends,
Please, take a look in the article uploaded at the second post of this thread. Using the technique described there and with an easy and trivial full two port calibration you will be able to characterize devices, no matter the symmetry of the footprint, with very good consistency at frequencies as high as 6GHz.

NandoPG

 

[marcomdd]

AN 1287-9.

Bye

 

[hr]

i made some test fixture to mearsure microwave filter and amplifier above 10ghz freq.

 

[yingyang]

test fixture

By black-box modeling, one simply measure the fixture without any circuits or mmic modules on it. The module should contain two open-ended microstrip line at the two connector. This S-parameters are in turn saved for future de-embedding.

The S-parameters of the similar open-ended microstrrip line simulated.
By taking the fixture Y-parameter in parallel with the two simulated Y-parameters of the 50 ohm lines, the unknown test fixture can be derived.


The mmic module with the test fixture is next measured. An appropriate substraction is next performed and the S-parameters can be found.

I have used this method up to about 50GHz with perfect agreement with TRL calibration.

 

[loucy]

What YingYang described, it seems to me, is replacing some calibration standards with the simulated results. It doesn't look like a full calibration of the fixture.

 

[yingyang]

fixture

It is not really with simulated results....
Only the lines are simulated.....

This step can be replaced by measurement but to save time, the preferred method would be to simulated the open-ended lines.


The derived fixture S-parameters are measured quantities and are reliable......


No need to derive any equivalent circuit for the fixture....

 

[loucy]

YingYang,

This "black-box" technique is new to me. Can you give some reference?

I've spent a lot of time on the TRL method. It involves some very subtle assumptions and mathematics.

From what you describe, I really don't see the equivalence between TRL and your black-box method. I think you get good results for your projects because long transmission line (the part you simulated) and also the matching is actually quite good with the DUT in place.

This is an interesting topic.

Loucy

 

[harkonnen]

yingyang,

you got me very interested in this Block-Box method.
Could you go more in detail of the procedure?

So, you simulate the (one-port) microstrip portion of the fixture and then derive the S-Parameters from the total fixture using a one-port measurement? These are all one-port measurements, or?
It sounds as if it needs significant effort to derive the loss and phase of the fixture-half by only measuring an open (no short).

Why do you put the simulation (one microstrip line) and the measurement in parallel with Y-parameters, why not in series with Z-Parameters? What does that help?

How good was your MMIC-device when you got "perfect" agreement with TRL? Were you concerned about the phase? A digital driver (fiber optic modulator driver) might be fine, but how about a linear amp with good return loss? Even a TRL/LRM-fixture seems hard to realize up to 50GHz.

See, I have lots of questions.
Regards