Bias Network for LCR Meter

[Smokey]

Cool.  Thanks for the update.  I think I'll give it a shot and see how well it works with the short cal done while the coupling cap is jumped out. 

I saw your BOM is all from LCSC.  What was the approximate BOM cost?  I'll probably find Digikey equivalents.

[mawyatt]

Don't recall the cost, the parts weren't expensive tho. Check the spacing on the BNC, recall they were a few mm off, but was still able to get a good fit by just slightly skewing them when soldering in place. At some point was going to tweak the layout but never got around to it.

If one plans on measuring large caps, then consider installing a bypass momentary push button switch with a 1K series R around the 100K R from the DC bias Banana connector. The idea is to push the switch to quickly get the large cap up to voltage for LCR measurements, as otherwise it can take some time for these large caps to stabilize in DC bias. This is something we'll likely do if/when we build another fixture, one dedicated for high value caps with larger series capacitors (PCB has place & jumper for these electrolytics, see schematic) and probably even a momentary discharge switch to discharge the cap before and after testing.

Best,

[Smokey]

I made a Digikey BOM with what I think should be equivalent parts.  The total is about $41 in parts, with $20 just in BNC connectors.  This is about double what you listed from LCSC, which is sort of what I expected.

| ID | Name | Designator | Quantity | MFG | MFG Part Number |
| 1 | 1N4148 | D4,D5,D8,D9,D10,D13,D14,D15 | 8 | SMC | 1N4148W  |
| 2 | XH-3A_C722863 | HIGHPOT,HIGHCUR,LOWCUR,LOWPOT,BIAS | 5 |  |  |
| 3 | 33u | C4,C5,C11,C12 | 4 | Panasonic | ECA-2EM330  |
| 4 | 10uF | C1,C2 | 2 | Cornell Dubilier | 106MMR100K  |
| 5 | HDR-M-2.54_1x2 | EC | 1 |  |  |
| 6 | 100nF | C3,C6 | 2 | Kemet | MMK5104J100J01L16.5TR18  |
| 7 | C232A475J6SC000 | C7,C8,C9 | 3 | Epcos | B32522C0475J000  |
| 8 | 1uF | C10 | 1 | Panasonic| ECW-FD2W105Q1|
| 9 | 1N4004W | D1 | 1 | SMC | 1N4007FL  |
| 10 | MM1Z10-C128709 | D2,D3,D6,D7 | 4 | Nexperia | PDZ10BGWX  |
| 11 | 1N4148W_C432607 | D11,D12,D16,D17 | 4 | SMC | 1N4148W  |
| 12 | 10k | R1,R3 | 2 | Stackpole | RMCF1206FT10K0  |
| 13 | 100k | R2,R4 | 2 | Stackpole | RMCF1206FT100K  |
| 14 | Header-Male-2.54_1x3 | TEST | 1 | Adam Tech | PH1-03-UA |
| 15 | DOSIN-801-0030 | HC,HCUR,HP,HPOT,LC,LCUR,LP,LPOT | 8 | Molex | 731385033 |

I actually used the cheaper BNC connectors with plastic housings, RF1-01K-D-00-50-BK-HDW.


NOTE1:  This was generated from the V4 design.  I have not confirmed all these parts yet.  Ordered but not yet received or assembled or tested.  No guarantees they are actually fully equivalent yet. 
Note2: Fixed 1uF cap part number.  Previously listed 0.1uF.
Note3: Items 4 and 7 are either/or stuffing options.  Item 3 is qty2 and optional.

[mawyatt]

Typical 2X parts cost from Mouser/DigiKey vs LCSC, however LCSC has a higher shipping charge. The BNCs are somewhat expensive, but we already had/have a bunch as we use these a lot for other projects/tasks, same for most of the "other" parts except the 10uF film types, these we didn't have in stock.

You don't need the 33uF if you are using the 10uF film types, and visa vera. The electrolytics are for measuring high cap values (like other electrolytics) at low frequencies, otherwise use the film types, check out the schematic. Also note that if you plan on going to low frequencies (like our Hioki IM3536 can go to down to 4Hz), you may need to make C10 larger than 1uF, we used another 10uF but had to fold the lead underneath to fit the thru-holes.

This DC bIas Fixture works a treat for our use, so hopefully you get some good use also!!

Now that our main project is heading to a Trade Show 12/7/23 we may have some "free" time to build another with the mentioned switches for fast charge/discharge use with high value DUT caps, and spend more time with the plotting software.

Anyway, hopefully works with the chip-set in the MS5308.

Best,

[mawyatt]

We have some work ahead to measure a bunch of large valued electrolytic capacitors, which may include capacitance change over voltage ranges. So the plan was to use this DC Bias Network with our LCR Bench meter, however recalling that one must wait some time before each measurement to allow the DC bias to stabilize (large RC time constants involved), we looked into providing a means to improve this Bias Network.

Bypassing resistor R2 in the schematic with a momentary Push Button Switch solves the long delay and allows quicker charging (and discharging) of the DUT. The switch was installed on the side of the aluminum housing as shown.

Our procedure when measuring larger valued capacitors with DC bias is to always start with the DC power + lead shorted to the - return lead "At the Power Supply", then install the DUT, then return the + lead to the PS + Terminal and set the DC voltage. With the Bypass Switch installed, use the same procedure and press the Bypass switch and watch the sense DMM voltage to allow charging to proper level. When removing the DUT, we also short the + lead, wait and then remove the DUT (this discharges the DUT). With the Bypass switch, use same procedure and Push to quickly discharge the DUT before removing.

BTW, changing R1 to 1K also helps.

Anyway, adding the simple Push Button Bypass Switch makes testing/measurement of high valued capacitors with DC bias a little less tiresome!!! We'll create another version of the schematic and PCB someday to reflect this addition, when time permits.

Best,

[scopeman]

Hello Mike,

This looks like a really well done project.

On the BNC spacing there seems to be two standards. I ran into this when I was making some adapter PCB's for some meters I own.

I have a Wayne Kerr 4230 and a Wayne Kerr 4250 and also a XJW 01 and some fixtures to use with them one like the one you depicted for SMT parts and one for radial/axial parts. They are all 22mm spacing.

It seems that the Wayne Kerr is 24mm on the 4250 and 22mm on the 4230 (the 4250 is the older of the two). I believe that the Wayne Kerr 4230 (200KHz capable) was a rebranding of an Asian companies product but at the moment I can't recall which one.

I believe that all modern LCR meters use the 22mm spacing. From measurements on your PCB it looks like you have 22.5mm spacing which accounts for the issue you were having.

I was able to import the Gerber files into my SprintLayout6.0 program. Correcting the spacing for the BNC connectors will be fun! I guess it would be easier if I had the native PCB file.

Again congrats on the nice project.

Sam
W3OHM

[Smokey]

I was playing with the gerbers as well.  I'm no KiCad expert but it looks like it will import Gerbers into an active board layout as lines and holes and it will let you move them around.

I'm going to end up making the matching fixture board for this LCR meter anyway, so I'm just going to match what you did on your board.

[mawyatt]

Think the BNC spacing for modern LCR meters is 7/8" or 22.225mm.

We've updated the schematic and PCB to reflect moving the BNC to the 22.225mm spacing, adding a charging/discharging switch connector, and to "clean up" a few minor things.

The attached Gerber files are for your review, we haven't built this version (yet).

Anyway, hope this helps some. Let us know if you decide to build this and how it works.

BTW make sure to enclose the entire fixture within a Faraday shield as shown, having a stable and repeatable electrical and mechanical structure is important for good repeatable results. Also, the electrolytics shown (C5,6,11,12) can be arranged in various ways, for unipolar use, or different capacitance values, or voltage ratings, or physical sizes, all depending on one's needs.

Edit: Gerber file updated.

Best,

[mawyatt]

CAUTION!!!

Do not use this fixture with any East Tester meters, seems they all have the front panel BNCs reversed from the decades old BNC configurations from HP, and used by Agilent, Keysight, R&S, Hioki, Tonghui, B&K, GW Instek and others.

https://www.eevblog.com/forum/testgear/east-tester-et4510-lcr-meter/

Best,

[mawyatt]

Updated Gerber files for those interested in this DIY adapter/fixture. The OEM ones are indeed expensive!!

Best,

[Smokey]

Finally got around to building up the V4 boards (and making a BNC adapter for the MS5308).

It sorta works, but I'm seeing some strange behaviors.  I got around the short calibration step by bypassing the caps in the HCUR and HPOT lines during that calibration.  I need to measure some reference components without any of this other stuff, and then again with the bias fixture to see how much of an effect that has, but there are bigger issues to deal with first.

First issue is that if I route the LCUR and LPOT lines through the V4Bias board, the MS5308 refuses to measure anything.  If I instead bypass the v4Bias board and hook the LPOT and LCUR directly to the BNCs on the MS5308 breakout then it works fine (with no bias voltage).  This is strange, because apart from the protection diodes those lines are straight through connections on the board.  I confirmed the diodes are all installed correctly and I can pass a signal with my function generator through that board fine.

Second issue is that applying a bias voltage of over a couple volts causes the reading to rise incrementally until it reads overload on the meter.  This never settles out to a stable measurement.

Does that make any sense what's going on?

[mawyatt]

Interesting, wonder if this has something to do with the bypassing the caps during calibration. We tried using it with the DE-5000 as mentioned but the meter wouldn't pass the short test, so abandoned that LCR meter.

However, we have no issues with the Tonghui TH2830 and Hioki IM3536 using this DC Bias adapter since you can bypass the DC short calibration, and we use the adapter often. We've used the DC Adapter with 60 volts DC bias without issue with mentioned Lab type LCR meters.

Also we don't see any significant change is measurements with or without the adapter (with proper calibration) as previously mentioned. For example here's a few measurements from early 2022 comparing the same DUT with and (without) the DC Bias Adapter using the TH2830.

4.6872pF (4.6830pF) C0G
993.498pF (993.468pF) C0G
101.654uF 0.706 ohm (101.783,0.691) Tant
21.319uf, 1.43 ohm (21.359,1.42) Tant
46.981uF, 0.793 ohm (47.007,0.748) Tant

Maybe if you can locate another Lab LCR meter, give that a try.

If you want to ship to us (with return shipping), we'll test the adapter with our Lab LCR meters.

Best,

[Smokey]

Thanks for the reply.  I didn't mean to say there is anything wrong with the adapter.  It's most likely either a quirk of the front end of the MS5308 (which is based on the same chipset as the DE-5000) or something else I'm doing with the setup.

If I get motivated I'll take apart the meter and try to trace out the front end schematic and see what I can see.

The strange thing is that with no bias voltage, and the LPOT and LCUR lines connected directly to the meter, it actually looks like it's doing a good job of making a stable measurement across multiple ranges and frequencies, all through the bias fixture caps. 

[KungFuJosh]

Is version 5A the latest? Do you have a BOM and schematic for that version also?

Thanks,
Josh

[mawyatt]

Is version 5A the latest? Do you have a BOM and schematic for that version also?

Thanks,
Josh

There's a 5B and 6A versions now. The 5B version has the ability to select the large Film or Electrolytic coupling capacitors, also the bypass switch connector to speed up charging large value capacitors. The 6A version is tailored for higher frequency use than 5B and doesn't have the electrolytic selection option, but also has the bypass switch connector.

Here's the Gerber and Schematic files, along with images of the assembled 6A version and PCBs for 5B and 6A. Note we had to use two 2.2uF Film caps paralleled for C2 since we didn't have a 4.7uF in house.

Study the schematics to get an idea of how these DC Bias Adapters works (distilled from old HP fixture and Hioki reference), basically providing DC Isolation by means of series coupling capacitors in the Hcur stimulus line and another in the Hpot measurement line. There's no need for DC Isolation on Lcur or Lpot as they are normally around ground potential even with applied External DC Bias.

The optional protection Diodes and Zeners add a layer of protection for the LCR Meter, some modern quality Lab LCR Meters have built-in protection, +-42VDC is common, however these optional components provide some added meter protection.

Edit: Updated Schematic for 6A

Also, none of the component values are critical nor needs to be matched, just about any good quality, stable components should work. For the V6A pay attention to the Coupling Capacitors ESR and ESL, here is where the quality 0.1uF C0G helps! 

Best,

[mawyatt]

Note updated with corrected schematic V6A, see above.

A little more detail on the workings of these adapters. In V5B the CHG-SWT 2 pin connector is for a Push Button Momentary Contact Switch, in V6A this SWITCH 3 pin connector allows a Charge and Discharge capability depending on the Switch type (3 position momentary vs 2 position momentary).

C1 and C2 selection are based upon desired maximum DC Bias Voltage, Low Frequency Range, Highest DUT Capacitance Range and type LCR Meter used. We've found with our intended usage and available LCR Meters (TH2830 & IM3526) that C2 is ~ 1/2 of C1 value works. We used 10uF 100VDC for C1 as this was available and had to parallel two 2.2uF 100VDC Capacitors (~4.4uF) for C2. The IM3536 is more sensitive to C2's value than the TH2830 which works fine with C2 as 1uF. Suggest trying different values for C1 and C2 your intended LCR Meter use and component range.

C8 and C13 are for 0.1uF C0G types with low ESR and ESL (these act as "bypasses" at higher frequency), C5 and C10 allow additional paralleling with C8 and C13 which we didn't find necessary and omitted.

The reason for the long resistor strings from the BIAS Connector is to decrease the effective capacitance "seen" from the DUT HC BNC connector, and also decreases the voltage across a single resistor, which is likely overkill unless one envisions very high voltage use. 

One very important consideration is the shielding and mechanical stability of the PCB enclosure. Use a sturdy metal enclosure (we used an extruded Aluminum section as shown in the images) with good all around shielding connected to the BNC ground and you will be rewarded with highly repeatable results that agree quite well with measurements obtained directly from the LCR Meter, we know we've confirmed this multiple times at pF levels

Anyway, we have a couple extra PCB sets, PM if interested.

Edit: As a sanity check we compared some results {with} and without the V6A Bias Fixture @ 1MHz on the IM3536 at ~26.6C ~44% RH.

One interesting note is from a measurement over 2 years ago with a 10pF Mica & 10nF Polystyrene*.

1nF Polystyrene       998.125pF       {999.756pF}
1nF Mica                 1.00688nF       {1.00808nF}
10nF Mica               9.99980nF       {10.0707nF}
2pF Mica                 1.9762pF         {1.9766pF}
10pF Mica*             9.9304pF*        {9.9247pF}
10nF Polystyrene*  10.3343nF 93.96mΩ*    {10.3252nF 98.15mΩ}
* Measurement result without Bias Fixture from over 2 years ago!

Best