Direct Connect LCR Meter Capacitors

[mawyatt]

We've made a few Direct Connect BNC Capacitors specifically for use with our LCR meters. These use a male panel mount BNC connector mounted to a fixture housing the capacitor, which creates the shortest length for the Hcur, Hpot, Lpot and Lcur paths. The fixture is shielded against external fields.

An interesting topic over here, caused us to look into the Keysight 4980 manuals and they discuss a direct BNC connect reference component and the mention of the effects of cable/fixture length on measurement accuracy. They also mention proper shielding/grounding, and using a known reference component as part of calibration to improve measurement results. Must read for anyone interesting in improving LCR measurements.

https://www.eevblog.com/forum/testgear/keysight-e4980al-precision-lcr-meter-partial-teardown/msg6304548/#msg6304548

Awhile back we created a set of reference caps for our use and also 3 of which have the old Soviet Mica types in the square metal cans. We also included the ability to measure the cap metal can temperature inside the shielded fixture. These are the best caps we've used and we just placed an order for a bunch more!! Also ordered more male panel mount BNC connectors, thermistors, and such.

Hopefully these just ordered old Soviet Micas will be as good as what we have, soon we'll find out

Best

[mawyatt]

Some of the old Soviet select Micas in these Direct Connect BNC Fixtures have shown repeatability better than 6 digits, with D of 0.0000x (Qs >10,000) and Phase lingering around -90.000 degrees!! Having the Short & Open Cal in almost identical fixtures surely helps.

This arrangement also allows LCR measurements to these levels that are quick & easy with no hassles

Waiting on the arrival of a bunch of these Micas, anticipating good results with a few!!

https://www.eevblog.com/forum/projects/direct-lcr-meter-connect-test-subjects/msg6259766/#msg6259766

Best

[mawyatt]

Just received a batch of the old Soviet Micas caps. They are "new" unused, but quite old and in hermetic cases. A quick test indicated some were within the original specs of 0.5%, and the others just out, altho all seemed to be quite stable 

We had prepared a 3D printed Direct Connect Fixture "home" for a few of these "new" micas, and had one ready and waiting. After a few hours of carefully removing the lugs (they are soldered on the feedthru posts) and soldering the mica in place to the BNC adapter front plate, then installing the temp sensor, everything was assembled. Preliminary results are very promising, maybe better than our superb reference 10nF mica, we'll need to wait a few weeks to let everything settle in before attempting a true evaluation tho.

This is better than expected, and our expectations were quite high indeed

Anyway, we'll provide updates along the way for those interested (we've got a bunch of these to evaluate).

Edit: Got another assembled, altho didn't line the box interior with Copper foil, which doesn't seem to be necessary with these larger values (20nF). Took a quick look and this one is showing a "Q" of >13,000, and also quite stable on the TH2830

Best

[mawyatt]

Made some measurements this afternoon of the 2 Soviet 20nF Micas mentioned above, they seem to be stabilizing nicely as the results are within +- 0.1pF from last evening. Now 3D printed another casing and will build up another. The unshielded version (Green, shielded is Magenta) seems fine without a complete shield. This just has Copper foil on the BNC mounting plate which provides a Guard plane close to the BNC wiring, and with the Mica Cap completely shielded within a hermetic metal tub which is attached to the Guard plane, we see no effects of local (hands around fixture for example) influence on readings

Best

[mawyatt]

We're seeing +20~30ppm TC for these larger select (10 & 20nF) Soviet Mica Caps based upon some very crude measurements around cap temperature of ~29C on a few caps. These select caps apparently have little to no temperature induced hysteresis, as they always return to the exact same measurement day after day, and no aging effects (after all, they are really old).

Will probably build up a few more of the select 20nF Micas (have a couple more good ones), as we have a few more BNC connectors available (need to order more later, along with thermistors, brass inserts & SS bolts and filament). PM is interested in acquiring one of these Direct Mount Mica Caps as we'll probably have a few extras.

Best

[mawyatt]

Suspect one of the reason for the extremely good repeatability (often out to 6 places or better!!) is the Open/Short Calibration repeatability. The Calibration Open Fixture is showing less than 2fF and quite repeatable as measured by the Open DUT voltage and current, while the Calibration Short Fixture is somewhat less repeatable usually varying between 50~100uΩ which is due to the contact impedance variations as the Open is less susceptible to such.

In either case a proper Open/Short Calibration reduces/removes these effects, and all the better since the Open & Short Fixtures are almost identical electrically and mechanically to the actual DUT Mica capacitors fixtures.

Best

[mawyatt]

We've built up a few more of the Direct Connect LCR Fixtures with the old Soviet Mica 20nF caps (these are old new stock) and now have 4 to work with (need more BNCs!!).

This has precipitated the dive into Test Controller (TC) to get our IM3536 and TH2830 working with such. With the help of HKJ (TC author) and Josh we were able to get both LCR meters up and running under TC. This is quite a useful tool for logging and plotting various measurement parameters, highly recommended for folks to spend the time to get familiar with TC.

https://www.eevblog.com/forum/testgear/program-that-can-log-from-many-multimeters/

These Soviet Mica based Reference Caps have a Thermistor attached to the metal case that allows remote device temp measurements with a DMM. We've found that the IM3536 causes ~2C temperature rise at the DUT, and the TH2830 ~3C rise, by conducting heat thru the 4 BNC connectors.

With TC up and running we began to plot capacitance (and other parameters) along with DUT temperature for these new old stock Soviet Micas including others we have. This allows one to get a crude temperature coefficient of the DUT around room temperature by allowing the LCR meter to warm up without a DUT connected, then connecting the DUT and observing the temperature rise due to the heat conduction thru the BNC connectors while plotting the various DUT parameters (capacitance for example).

The general "time constant" of the LCR Meter and DUT setup can also be observed. Another benefit of the TC application is to observe the LCR Meter thermal "time constant". This is accomplished by using a very low temp coefficient DUT (we have one select Soviet Mica with ~+10ppm temp coeff) connected to the LCR Meter powered Off that has been allowed to stabilize to room temp. Power the LCR Meter ON and start TC and plot the DUT capacitance and DUT temperature. The result should show a classic exponential temp rise where one can extract the thermal time constant of the DUT and LCR Meter. For a finer result one can utilize the RSS result of the DUT time constant and LCR Meter time constant since they are partially uncorrelated by allowing the LCR Meter to stabilize (~1hr) then attach the DUT and start TC, this should show ~ the DUT time constant result. Then from above one has the combo time constant and the DUT time constant, then solve for LCR Meter time constant using RSS method.

Anyway, this is an ongoing experiment (takes time!!) that we thought others might be interested in, and here's some examples of Direct Connect Soviet Mica Fixtures.

Best

[KungFuJosh]

That's fantastic! It very clearly illustrates the DUT temperature sensitivity over time. 

[jonpaul]

We use GenRAd precision caps, have a stock of old USA made micas that arte calibrated.

The best are the 4 term GenRad coaxials.

Beware the BNC connectors use teflon or nylon insulators with their own cap and TC.

The most accurate bridges all need a guard to have sub pF accuracy.


j

[mawyatt]

Some of these old Soviet Micas and others are housed within a Copper foil shielded (Faraday) box which makes contact thru the BNC case to the LCR meter BNCs.  The ones we have are Teflon, they aren't the best connector (N or SMA better), but generally OK at these lower frequencies.

Here's an example of an ongoing Test Controller Log with our best Soviet Silvered Mica (20nF), note the SD of 8.7fF!! Someone indicated that the old Soviet Micas were made with some of the best Mica available, think we may have one

Best

[mawyatt]

We've been doing some testing as we become familiar with Test Controller. We have a few old Soviet Mica caps mounted in these Direct BNC Connect fixtures, and a few that haven't been mounted.

These unmounted are all old new stock 20nF Mica in square metal cans (heretically sealed). We decide to hot glue a thermistor to the metal case and place the case between a pair of cermet power resistors to act as heat sources, held in place with a plastic clamp. Not wanting to run high above ambient (25-35C is where we are interested in the capacitance and TC measurements) we didn't use much power for heating the resistors.

As you can see these 4 20nF Micas are pretty good with temperature around 28~38C range showing all positive 8~12 ppm/C TC

Best

[mawyatt]

The above caused us to retrofit some of the Soviet Micas with the ability to heat the DUT metal case. We mounted a power resistor (10Ω) on the case and brought the two leads out in another 2 pin header arrangement. We used a 4 pin heater for the thermistor and the heating resistor.

Here's a few images and Test Controller Plots that show the results.

Anyway, we tried reversing the resistor bias polarity to cool the DUT below ambient (electrons going the other way to pull the heat out), that didn't work    So we put the fixture into the freezer for a few minutes, and that worked

Best