Different type LCR SMD Fixture

[EC8010]

Interesting stuff. At the moment I don't have a need to measure SMD parts accurately, but the principles are nice to see - especially the roughened surface trick.

By the way, it appears to be an informal standard that the BNCs on LCR analysers are all on the same spacing. Very convenient. My (cheapy) Hameg 8118 talks about guarding in its manual, but with only four BNCs and all of them bonded to chassis inside the instrument (let alone my jigs), I don't see there are enough electrical connections to use a guard. Or maybe I'm missing something?

[mawyatt]

The surface "trick" really helps as we were having much more difficulty in getting contact with the DUT SMDs. This works well and the "trick" is getting the mixture of fine Cu fillings mixed with flux and solder right, we've trashed a few PCBs and still don't have an exact formula but found ChipQuick Sn63/Pb37 T3 with Mechanic Uv559 Flux works well whereas the Mechanic NS83 Solder Paste with said flux didn't.

The Hioki IM3536 has all the 4 BNCs "floating" wrt to chassis ground, and a Guard Terminal (Banana 4mm) next to the Lcur BNC. The Tonghui TH2830 has the Guard Terminal on the instrument rear panel and is also isolated wrt chassis ground but apparently not as well "isolated" as the IM3536.

We reported measurement improvement when using Kelvin Clips using the Guard connected to a large (300 by 200mm) conductive plate (PC Bite). After some thought this made sense as all the Kelvin Clips we own have the clip tip as a long blade thur the handle and this entire length is unshielded which is one reason for measurement variations when using the Kelvin clips handheld!! With the fields so exposed with the Kelvin Clips, having a large area at the Guard potential to terminate the fields helps reduce their effect and this directly shows in the measurements.

Our quality Kelvin Clips from Tonghui (TH260011CS) are only rated to 100KHz, the common cheap types have no rating. We don't put much faith even in the quality clips when using higher frequencies since even a slight cable movement changes the readings, but do get better results using the Guard plate approach mentioned.

Best,

[Nominal Animal]

Actually the Green tip is 3D Printed (PLA tho), along with the Grey Standoff box (also PLA) for the Plunger assembly. With the setup as shown there is just enough pliability between the Tip, plunger arm, box and spring tensioned screws underneath the PCB. You can adjust tension and range with lever arm nut arrangement.

Silicone rubber would be better for the Plunger tip, but the 3D printed PLA Tip seems to work well for now

Yep!

I'm only worried about fracturing thin film resistors and multi-layer ceramic capacitors when the lever goes past the switchover point: if the tip is too hard, it might exert a very short spike of pressure, akin to tapping with a small hammer.  I worry about this kind of stuff because I'm an uncle bumbleduck (in AvE's vernacular) who somehow manages to break things quite often, so playing it safe lets me use more of my small budget for new stuff.  (I like to do things like use rubber paint (the kind used to coat the bed on a truck) on the insides of my Hammond cast-aluminium enclosures, to avoid accidental shorts due to things shifting about... So do consider my suggestions as pretty firmly on the paranoid side.  In my hands, stuff just seems to, uh, fail often. Me too.)

[mawyatt]

There is a flexible 3D filament called TPU, we haven't used this so might be worthwhile to take a look.

Understand the issue with concern about fracturing delicate SMD chips, having seen some of the fractured SMD chip caps like the one Dave had light up when testing a cheap PS.

What's interesting about this approach with the split Kelvin Force and Sense approach where the "connection" between them is on the DUT end terminations rather than the at the fixture to DUT interface, is this requires minimal DUT contact pressure.

We can't get any reasonable repeatability with sub 10 milliohm SMD with the Standard SMD fixture, but have little problem with the split Kelvin approach.

Best,

[gitm]

The Hioki IM3536 has all the 4 BNCs "floating" wrt to chassis ground, and a Guard Terminal (Banana 4mm) next to the Lcur BNC. The Tonghui TH2830 has the Guard Terminal on the instrument rear panel and is also isolated wrt chassis ground but apparently not as well "isolated" as the IM3536.

I’m confused by this statement.  When I view the mfg’s picture of the TH2830 back panel, it shows a connector marked with a ground symbol.  I don’t see a “guard terminal”.  Is the “ground” an actual ground or does it carry a guard signal?

What is the resistance between the rear panel “ground” to mains ground on the power cord?  I’m wondering, if it is really a ground, how much improvement you get connecting it to the guard on your fixture?

BTW, very nice work on the fixture.  I need one of these.

[gitm]

Wondering what happened to this thread?

[mawyatt]

The Hioki IM3536 has all the 4 BNCs "floating" wrt to chassis ground, and a Guard Terminal (Banana 4mm) next to the Lcur BNC. The Tonghui TH2830 has the Guard Terminal on the instrument rear panel and is also isolated wrt chassis ground but apparently not as well "isolated" as the IM3536.

I’m confused by this statement.  When I view the mfg’s picture of the TH2830 back panel, it shows a connector marked with a ground symbol.  I don’t see a “guard terminal”.  Is the “ground” an actual ground or does it carry a guard signal?

What is the resistance between the rear panel “ground” to mains ground on the power cord?  I’m wondering, if it is really a ground, how much improvement you get connecting it to the guard on your fixture?

BTW, very nice work on the fixture.  I need one of these.

On the TH2830 the BNCs and rear terminal are directly connected, from the rear terminal to the earth ground on the AC measured to one of our PS with earth ground terminal we get 3~4 ohms. So not a true Guard, maybe a choke involved but don't know. The Hioki IM3536 Guard is basically galvanically isolated from the earth ground.

We haven't used the Guard on the TH2830 yet, but the the Guard on the Hioki provides an improvement in measurements when using Kelvin Clips & Cable as mentioned with a Guard Plate, however the Split-Kelvin SMD Fixture it's not as pronounced if any. Honestly we need to do more detailed testing to verify the improvement, if any, and how much. Time is limited as today was spent building up another DC Bias Adapter which uses a pair of ultra-high "Q" C0G capacitors to improve the high frequency performance and we need to see how well this works.

BTW our lab is far from "electrically quiet" and we should be receiving some clip-on chokes for the Power Cords, Hioki recommends these for "noisy environments" and we'll see how well these work.

Best,

[tszaboo]

Wondering what happened to this thread?

There are two similar threads with the same pictures on this forum, that were recently active.
I was also wondering, as I seemed to remember more pages.

[mawyatt]

Wondering what happened to this thread?

There are two similar threads with the same pictures on this forum, that were recently active.
I was also wondering, as I seemed to remember more pages.

Maybe the links in post #21?

Or this one: "Cheap SMD LCR Fixture, the Good , Bad and Ugly!" We used an "!" in title which causes a link issue, so do a search.

Best,

[mawyatt]

The Split-Kelvin SMA Fixture also works with a DMM requiring 4 BNC to Banana Adapters allowing Hcur, Hpot, Lpot and Lcur to be routed to the 4 Banana inputs of a DMM as shown.

Here a 40 milli-ohm 1% 2512 SMD Current Sense Resistor is being measured with the 4 Wire Ohms capability of the Keithley DMM6500.

Best,

[mawyatt]

Few more 1, 10, 15, 20, 25, 50 milli-ohms 1/2 and 1% 2512 Resistors.

Best,

[mawyatt]

Using same Split-Kelvin Fixture with DMM6500 as shown above, we used a KS34465A and made some measurements with same DUT Precision 2512 Resistors. Compare with above

Think these agree well with DMM6500 measurements, and show the ability of this unique SMD fixture.

Best,

[mawyatt]

We've had some requests for the Gerber files for this fixture, and included the updated original (V4) and latest which has the Guard capability (V5), both work. Also included are the 3D files for the Plunger Tip, Base and Bottom Supports.

If you intend to copy this PCB and fixture, be sure to study this entire thread to understand how the fixture operates and how it interfaces with a Bench type LCR Meter, also how to "roughen up" the contact surface for better connectivity. Note how the Plunger Mechanism is attached and operates, and the 3D custom tip and standoff which places the Plunger base above the PCB Surface.

One should study the various LCR Meter threads related to this special Custom Split-Kelvin SMD Fixture and it's use, as this provides a good vision of the usefulness and capabilities.

Anyway, hope some folks find this unique SMD fixture worthwhile, we're here to help if needed.

BTW plans are to publish this unique fixture soon, stay tuned!!

Best,

[KungFuJosh]

The nylon screw was a success. No modifications, and the screw is flexible enough I can safely position it on 805 SMT stuff with no trouble.

I did order the V5 PCB since I like the contact design better.

Thanks,
Josh

[mawyatt]

Nice work Josh

That setup sure looks like it belongs on the IM3570

What size nylon screw did you get?

Best,

[KungFuJosh]

Nice work Josh

That setup sure looks like it belongs on the IM3570

What size nylon screw did you get?

Thanks! It is nice. Now that I have the nylon screw, it's very simple to use.

I bought a 6-32 1" nylon screw set. This one: https://www.amazon.com/dp/B00NQR0GAI

Bought that one because next day delivery and my impatience get along well enough. 😉

You would want a longer one if you're keeping the 10mm spacer.

Thanks,
Josh

[KungFuJosh]

Today I built version 5A of Mike's SMT fixture. It looks great, I really like the new contact pad design.




This time, instead of the cheap aliexpress clamp, I used the Adafruit 2459, and it's significantly better. The 2459 is smooth and easy to move, and holds its accuracy better.

With a single finger and very little force, the 2459 can easily move through its full range. The other one stops pretty short before a second finger is needed to convince it to move further:



With the Adafruit clamp, I used a 3mm spacer instead of a 6mm spacer.

Thanks for the awesome design, Mike!

[mawyatt]

Looks great, did you get a gold plated PCB?

The Adafruit Lever Arm does look good, as the nylon bolts...good job

Best,

[KungFuJosh]

Thanks! Yes to gold plate.

I'm glad I bought that extra fixture from that link you sent me. I bought the kelvin clips and the slightly cheaper tweezers. The 5A is on the tweezer fixture.

[Randy222]

The test fixtures look nice.

Just for educational purposes, does the measurements using compression forces (smd to fixture, banana leads pushed into bnc adapter, bnc insertions, etc) show the same results vs soldering the ends of test leads directly to the DUT?

[mawyatt]

Our experience suggest any type soldering causes DUT small changes due to applied heat. With this fixture the light pressure required doesn't seem to have any effect, likely due to the actual measurement technique since DUT contact impedance effects are highly attenuated, thus high DUT pressure not required.

WRT the BNC and/or Banana connectors, these don't have any affect we can detect assuming quality connectors of course

Best