Measuring Inductance with Network Analyser....

[Catalyst]

Hi there folks ,

I am trying to connect up a planar spiral upto a network analyser to measure the inductance at around 10MHz - 20MHz. Is this possible and is it accurate? The predicted value of the spiral is a few hundred nanohenires.

I am also trying to figure out how to connect the spiral to the network analyser, not sure what connector to use and how to calibrate the measurement properly.

Sorry about all the questions

Any advice is much appreciated

Kind Regards

Cat

 

[G4BCH]

Hi cat
It is easy enough to get a reasonable measurement of inductance using a network ananlyser.

You will need an open ended coax lead with a connector to fit your network analyser.
Make sure that the open end is large enough to connect your inductor.

Connect the lead to Port 1 and set measurement to S11, probably the reset condition.
Set the frequency range to 10 - 20MHz, or what ever you feel most suitable.
Select Smith chart and turn the marker on, set to your desired frequency.

Short circuit the open end of the coax in the same way as it will be when connected to your inductor.
Normalise the display, Data to Memory & Data/Memeory on the HP8753.

Add a phase offset of 180 degrees.

The marker should now show an open circuit or close to.

Connect you inductor in place of the short.

Read off the inductance from the marker.

Different analysers may disply the answer differently the 8753 reads inductance and capacitance directly. You may have to work it out from the reactance if that is al that is displayed.

It is not a perfect measurement but will get you close enough for most purposes.
You could just normalise to open circuit if the lead lenghts are very short and you can live with the extra error.

HTH

Peter

 

[Catalyst]

Thanks for the information

Does that method include the calibration stages? is that what the open and short stages are for?

why do I need to perform open and short calibration?

Regards

Cat

 

[G4BCH]

You calibrate on an open or a short. an open if you want a measurement that does not take into account the residual inductance of your connecting leads, or a short if you do. If the leads from the end of the coax are short then an open is good enough.

The calibration is only a normalisation to find the end of the coax. If the network analyser does not know the length of cable it can't work out the phase of the reflected signal and thus can't work out the reactance.

If you calibrate to a short you need to add the 180 degree offset because the network analyser expects you to normailse to an open circuit ( 0 degrees refelection ).

If you want to be more accurate you can do a 1 port calibration for S11, but I've not found that it makes that much difference for measuring reactances. You are only interested in the phase of the reflection If the inductor is any good the plot will be on the edge of the chart.

Peter

 

[Catalyst]

Thanks for the replies

Back to the issue of calibration. It looks like I'm going to have to solder two wires to the two flat terminal pads of the planar inductor. How can i calibrate out the wires from the measurement?

Can i measure the inductance of each wire sepereately? Then measure with both wires connected to the inductor and simply take the difference between them to get the true inductance of the planar inductor?

Regards

Cat

 

[G4BCH]

Maybe I should have been a little clearer witt my original explaination.
The open ended coax should include the leads you are going to use to connect to the inductor. Keep them as short as you can to minimise any errors.
When you short them together and normalise the display you are then taking the conecting leads into account.
The network analyser expects you to have normalised to an open circuit, that is a reflection phase of 0 degrees. A short circuit has a refelection phase of 180 degrees. If the display is showing a smith chart you will see a spot at the open circuit point. This is why you have to manually add 180 degrees phase offset to bring the point back to the 0+ j0. The display now shows what you actually have, a short circuit, including your connecting leads.
Connect your inductor and read off the value.

Peter

 

[Catalyst]

Hi again,

I've tried what you have said. But I am getting very unstable results. Even when i put a known inductor of 22uH, the value varies right in front of me from 21 to 23 uH. The range of inductance i wish to accurately measure is in the range from 200 nanoH to 1000 nanoH .

It just doesn't seem to be feasible to measure such small inductances with this setup.

1. Are there any guidelines for improving stability in the measurements?

2. Or is there another method for accurately measuring such small inductance values? At the test frequency between 1Mhz to 20 MHz.


Regards

Cat

 

[G4BCH]

22uH is probablly too high a value, things start to get inaccurate as the reactance gets higher, the nornmalisation was done at a short circuit.
I have just measured a 47nH 0805 chip inductor ( the first one I found ) and got 45nH using a piece of RG174 coax 140mm long, SMA one end and 10mm flying leads the other. The ground lead even had a solder tag slodered to it ( again the first lead I found ). Network analyser was HP8753B.
I put the inner of the coax into the screw hole in the solder tag and normalised that for the short circuit. Held one end of the inductor against the solder tag and pressed the inner to the other. Not what you might consider a very accurate proceedure but It came up within 2nH of nominal, which is about what I would have expected. for this mesurement.
Just to check your upper range a 1.8uH inductor measures 1.81uH, too close;-)

The open circuit value reads about 12uH, and doesn't quite make to to the edge of the chart, again about what is to be expected for a simple noramlisation. Calibarating for a S11 measurement would have improved this by takin the open circuit case into consideration.

You should be able to measure the range you want, it works here.
Peter

 

[sheela]

can please let me know how to proove the kuroda's identities or if anybody has the solution please send 2 my email address prakash@venture.com.sg.

 

[Catalyst]

Peter,

I have a HP 8753E VNA. For calibration it requires the open, short and load to be completed before any measurmeent can take place. In the case for your chip inductor measurement, what load did you use?

Thanks

Cat

 

[Gakusei]

The most accurate way to do this would be as follows:

1) Include a 50Ω stripline into your board design to connect the inductor to an sma connector fitted to the board. At 10-20MHz yo don't need to be too accurate with the transitions.

2) Using an SMA lead and cal kit, perform an open, short, load calibration on the analyser.

3) Connect the SMA lead to your board.

4) Use the analysers delay/reffering function to remove twice the stripline and connector length from the measurement. Make sure you take the velocity factor of the stripline into account.

If you look at the plot in polar or snith mode, you should now see a reasonably tight ball or cusp directly below the centre line of the plot. If you don't, you either reffered to the wrong place, or your stripline is way off 50Ω.

If you now take the centre of the ball or cusp you can calculate the inductance from this.

 

[G4BCH]

I did not perform a full calibration, just a normalisation.
Assuming the 8753E is much the same as the 8753B the key presses are

Display
Data to memory
Data / Memory

This normalises the display to 0 degrees and whatever reflection amplitude you have.

Scale Ref
Phase Offset
180 x

puts in the 180 deg offset.

Gakusei's method above is probably as close to ideal as you will be able to get, It allows you to use the network anylaser cal kit. You can get similar results by performaing a 0ne port cal for S11, just leaving the flying leads open for the open circuit, shorted for the short and using a pair of parallel 100 ohm chip resistors for the load. As Gakusei says you do not need to be over picky at 20MHz, just keep the lead lengths short.

Peter

 

[Catalyst]

Which type of chip resistors are recommended? any particular model / manufacturer?

Cheers

Cat