Sensitive LCR Meter Measurements

[Martin72]

The ST28006A test fixture arrived today and I'll be taking it home with me at the weekend.
Here are 2 pictures, from the outside the almost 270€ expensive ST version and mine from Aliexpress hardly differ.
But after opening it....

[KungFuJosh]

How do the plastic holders compare in shape?

[Martin72]

I'll take direct detail/comparison pictures at the weekend, but what I noticed was that the spring is smoother on the ST fixture.

[mawyatt]

You can adjust the spring tension by removing the plunger case and/or the end cap. We prefer more pressure on the DUT for better DUT terminal contact.

Make a shield like the ST version, you can use a Cu plate/foil or just Aluminum L Bracket. Also make sure the 4 BNCs are electrically well connected to the case, and the cover has good electrical connection to the base case thru the 2 screws holding it on.

A little rework and the cheap SMD fixture will be almost as good as the expensive one

Best,

[Martin72]

I think I will install some kind of second housing made of sheet brass.
But first I will repeat the measurements on the ST2830/ET3503, i.e. 10p, 100p and 1n, once with my fixture, then with the ST fixture.
Let's see if we can see anything.

[mawyatt]

A little EM analysis will show that all the case/ground/shield currents return thru the 2 end screws, make sure this connection is really good, one reason we prefer the 4 screw version!!

Best,

[Martin72]

That's a pretty exciting thing, I like that.

My 2nd edition of the power inductor checker has been set up and will soon be put into operation.
If it works as intended this time, I will publish it in the forum.

[mawyatt]

The ST28006A test fixture arrived today and I'll be taking it home with me at the weekend.
Here are 2 pictures, from the outside the almost 270€ expensive ST version and mine from Aliexpress hardly differ.
But after opening it....

Martin,

Something doesn't look right here!!

The ST28006A image shows two end brackets to hold the cover. Each support bracket is bolted to the cover with 2 bolts, yet the image of the ST28006A only shows 1 set of dual bolts at one end of the cover, where's the other bracket attachment bolts at the other end???

Best,

[mawyatt]

Here's some plots made with the IM3536 and the Custom Split-Kelvin SMD Fixture. These are 1% 2512 0.001 ohm and 0.01 ohm Resistors.

Edit: Added 50 and 100 Milliohms, later the High "Q" 100nF C0G Capacitor. Note the artifacts in the upper right of the -2 plot around 7.4MHz, likely due to a fixture/setup resonance.

Best,

[Martin72]

I have now also taken the ST fixture home with me, before that I measured 3 capacitors at 100khz at work, once with the ST, then with my nameless one.
The results are congruent, which suggests that the “denser” structure of the ST does not yet play a role in the frequency.
Tomorrow I will do the same with the ET3503, but then of course also at 300kHz.
When I got home, I photographed both fixtures again, the pictures will be split accordingly due to the maximum number.

[Martin72]

The two fixtures photographed from the outside...

[Martin72]

Finally, the clearly different inner life of the two.

[mawyatt]

How is the bracket attached to the cover in the Left side of image 4788 and Right side of image 4789?

There's no thru bolts or screws shown from the top side of the cover image for this bracket attachment!!

Are these something like a welded stud to the cover bottom, then smashed/expanded to hold the bracket??

Best,

[Martin72]

How is the bracket attached to the cover in the Left side of image 4788 and Right side of image 4789?

Normally with screws through the housing, as these also fix the spring mechanism on top of the cover.
This is not necessary on the other side, where threaded bolts are welded on and the bracket is fixed with nuts.
Mechanically completely plausible.

[mawyatt]

I have now also taken the ST fixture home with me, before that I measured 3 capacitors at 100khz at work, once with the ST, then with my nameless one.
The results are congruent, which suggests that the “denser” structure of the ST does not yet play a role in the frequency.
Tomorrow I will do the same with the ET3503, but then of course also at 300kHz.
When I got home, I photographed both fixtures again, the pictures will be split accordingly due to the maximum number.

With the 10pF Caps this is showing ~20ff difference, or 0.2%.

Since you used the same 10pF Cap (C0G), same quality LCR Meter, same basic setup, this leaves this Fixture as the main source of the difference. The difference effects can now be broken down into variations in Cal and main differences in physical fixture (lack of shield).

A simple experiment can verify IF the shield is worthwhile.

Setup the measurement after lengthly warmup with a very large average (~100), do a spot Open Cal at highest frequency the LCR supports (300KHz) with a small piece of plastic about same size as 10pF Cap DUT for proper plunger spacing with the ST Fixture. Now record the result, then replace the Fixture with the No Name and repeat, record the result. Note the small plastic spacer will allow the 2 fixtures to have ~ the same spacing and remove this uncertainty, however they will increase the fringe capacitive effect, so don't rely on this for accurate measurements, only relative differences. The ~100 averages should give good mean values for comparisons.

Now carefully remove the No Name cover and install a temporary shield. Use some art & crafts thin Cu plate and cut a shield and tack solder in place (BNC lug) at the ST Fixture shield location, careful not to disturb the wires. Make sure it's mechanically stable, you can use some hot glue if needed.

Put the cover back on and reinstall the fixture, then repeat the measurement procedure. Suspect you'll see a value closer to the ST fixture measurement, and this should verify the benefit of the shield.

After all, don't think Tonghui (and others) would use a shield if it provided no measurable benefit

Best,

[Martin72]

After all, don't think Tonghui (and others) would use a shield if it provided no measurable benefit

I didn't say that either, see the corresponding post.
Nor did I think so.

[mawyatt]

After all, don't think Tonghui (and others) would use a shield if it provided no measurable benefit

I didn't say that either, see the corresponding post.
Nor did I think so.

Of course you didn't!!

Just say'n it's there for a reason for others that might not have followed the entire SMD fixture threads.

BTW is the shield and case aluminum or steel?

The Tonghui Kelvin Clip base fixture that connects to the LCR Meter has a Steel case, the cover is Aluminum. Suspect this base is for rigidity holding the 4 BNC connectors, rather than any H field shielding. This doesn't have a shield partition, but the 4 wires are coaxial shielded up to the BNC connection with shields soldered directly to BNC lugs.

Best,

[Martin72]

I can test this later with a magnet.
The separation/shielding in the ST fixture should be made of tinplate, aluminum is not so easy to solder.
It “feels” like the NoName is even heavier, although there is less “inside”.
Whether the relatively low frequency of 100kHz already makes a difference would have to be compared with the nomogram, where the accuracy of the ST2830 actually lies when 10p is measured at 100kHz.
Or you can approach it in comparisons, from say 1khz upwards - people with an LCR sweep would be fine there....

Martin

[Martin72]

It “feels” like the NoName is even heavier, although there is less “inside”

It is heavier..
And so are the contact resistances, here as an example from ground to ground.

[mawyatt]

3 ohms is considerable, try measuring between the static plunger and the BNC shell for each. The lever arm doesn't really matter as it's not conducting nor shielding any signal. The BNCs provide the connection back to the LCR Meter are important wrt any shield currents getting back to the instrument.

BTW our "Good" Standard SMD fixture weights ~291g and is all Aluminum, our "Other" Standard SMD fixture has been repurposed to include external DC Bias Capability within, it's also all Aluminum and weighs ~318g but has a lot of extra "stuff" inside for the added DC Bias Capability!

Best,

[Martin72]

3 ohms is considerable, try measuring between the static plunger and the BNC shell for each. The lever arm doesn't really matter as it's not conducting nor shielding any signal. The BNCs provide the connection back to the LCR Meter are important wrt any shield currents getting back to the instrument.

I had measured everything, that was just the most “drastic” difference, which I then photographed.
I am no stranger to good shielding in principle.

[mawyatt]

Along the lines of sensitive measurements with typical Kelvin Clips utilized often with LCR Meters and DMMs, here's some information that users might find interesting wrt to very low Z measurements.

A note clipped from a Hioki data sheet illustrates the importance of "Proper" Zero/Short Meter/Accessory Compensation. We've added some notes of our own to help illustrate the effects of reversing the Kelvin Clips and why the proper orientation for a quality Zero compensation is important for quality low Z measurements.

With a proper Clip orientation with the Kelvin tips Hcur and Lcur directly on top each other as shown, the LCR Meter Forcing Current flows thru from the High Force side (Hcur) directly to the Low side (Lcur) and little current flows towards the Sense terminals (Hpot & Lpot) because of the high impedance these terminals present. In our notes because of this Ix ~0 since Id and Ic ~0. Under these conditions then Lpot and Hpot sense the same voltage Vx which is the mid terminal connection point between Hcur and Lcur, and the meter computes the effective Z as ~0 since Z = Vdut/Idut and Vdut is VHpot-VLpot or 0 since they each read as Vx as shown in the schematic.

When the Kelvin tips are reversed with Hcur now directly connected on top to Lpot, then Hpot, and Lcur, the full DUT Force current flows thru all 4 Kelvin Tip ends and introduces a voltage drop between Hpot and Lpot which creates a DUT effective impedance as shown as Rx in our notes.

Of course things get a little more involved and as frequency increases things can get way more involved, but this serves to illustrate the concept.

To show these effects we've employed our Tonghui TH2830 LCR Meter with TH26011CS Kelvin Clips. These images show the Kelvin Clip connections on a Guard Plate, proper and reversed orientation and the subsequence meter readings.

First a Zero/Short Compensation was preformed with the Proper Clip orientation, then the Clips disconnected and reconnected to show the Zero Reading, then Clips were reversed (L Kelvin Clip flipped over)  and measurements repeated. 16 averages were used at results shown at 100Hz, repeated measurements with connect and disconnect showed reversed connection variation from 2 to ~4.5 milli-ohms, whereas with proper orientation these variations were under 20 micro- ohms (Note the Th2830 resolution is 10 micro-ohms).

Would be interesting to see what others find with their LCR Meters and Kelvin Clips.

Anyway this is a subtle point wrt to Kelvin Clip usage for low Z measurements, and hope some find this useful.

Best,

[mawyatt]

We've added results for the TH2830 Rdc Measurements and the Keithley DMM6500. The TH2830 larger Rdc 46 micro-ohms with Proper Orientation is due to the Thermal EMF in the Kelvin Clips and Instrument, this effect is negated with AC measurements.

Note negative readings with reversed Kelvin Clip Zero measurement, reason is left as reader exercise

Best,

[EC8010]

I had to stop and think a bit about those Kelvin clip strictures.

Seeing Martin72's internal photographs, I had to open up my SMD fixture. It's identical to his expensive fixture (although the screen is steel) but has Hameg HZ188 written on the outside because it came with a Hameg 8118. I seem to remember seeing a Digimess LCR tester from before 2007 that had a similar jig.

[mawyatt]

Thanks to Martin for showing the ST version

Think the ST and many others from Tier 1 sources (R&S) are directly rebrands from OEM Tonghui, the lower Tier sources are various clones of the Tonghui with varying degrees of quality and various rebrands.

We have a couple clone fixtures, one is very well made, the other not so!! At lease on the inferior SMD fixture, the plungers seem the same and good quality, can't say so for the case tho

We decided to "repurpose" the inferior SMD fixture into a self contained SMD fixture which will allow external DC Bias applied to the DUT. This works really well and is compact and more convenient for SMD devices than using the DC Bias Adapter we developed for LCR Fixtures, which works well with other LCR Meter fixtures such as, Leaded, Kelvin Clip & Tweezers.

Also check out the Split-Kelvin SMD Fixture here:

https://www.eevblog.com/forum/projects/different-type-lcr-smd-fixture/

So we've pretty much got all the bases covered for SMD devices 

BTW if you use Kelvin Clips please note posts 71 and 72, the Zero Ohm Compensation does make a difference with Kelvin Clip Tip Orientation as shown

Best,