Replicating/Copying HP/Agilent/Keysight 16065A High Voltage LCR Meter Fixture

[ZGoode]

I'm working on making a copy of the Keysight 16065A for use with my HP LCR meter and some of my existing fixtures and wanted feedback on my choice of suitable components for this project.  The diodes were relatively straightforward, I think I have the correct ones.  Where I would greatly appreciate feedback would be on my choice of capacitors since nothing is listed in the Keysight manual other than the uF values.  I went ahead and chose caps rated for at least 200V, but wasn't sure if the particular type or tolerance I chose is the best option instead of something else.

C1: 5.6µF https://www.digikey.com/en/products/detail/kemet/C4AQJBU4560M1XJ/13563463
C2: 1µF https://www.digikey.com/en/products/detail/kemet/R75PR410050H3J/12144507
C3,4: 0.001µF https://www.digikey.com/en/products/detail/kemet/PHE426MJ4100JR05/13176743

R1,2: 61.9kΩ https://www.digikey.com/en/products/detail/yageo/MFR-25FTE52-61K9/9140331
R3,4: 38.3kΩ https://www.digikey.com/en/products/detail/yageo/MFR50SFTE52-38K3/9151447
R5,6: 20Ω 20W https://www.digikey.com/en/products/detail/yageo/SQP20AJB-20R/9167197

CR1,2,3,6: BZX55B10 https://www.digikey.com/en/products/detail/vishay-general-semiconductor-diodes-division/BZX55B10-TAP/4826033
CR4,5,7,8: UG56G https://www.digikey.com/en/products/detail/taiwan-semiconductor-corporation/UG56G/7367414
CR9,10,11,12,13,14,15,16: 1N4936 https://www.digikey.com/en/products/detail/diodes-incorporated/1N4936-T/21905

[Kean]

Have you seen the design by Mike Wyatt?

He has documented the component values and capacitor types used in his design
https://www.eevblog.com/forum/projects/bias-network-for-lcr-meter/msg4133314/#msg4133314

Also there is a post with an internal photo of an original 16065A that might help
https://www.eevblog.com/forum/projects/lcr-measurements-at-high-dc-bias-agilent-16065a-dc-bias-fixture-teardown/msg3958007/#msg3958007

[ZGoode]

OOOO, thank you!
I was trying to track down pictures of the actual board itself, but was not having any luck on that.  I'm gonna have to check those posts out.
Based on those two posts, looks like I got pretty close to what they did.  I'm gonna give it a go and see how it performs when the parts come in from Digikey

[Kean]

Please update us on how it goes.  I have one of the PCBs that Mike designed but I have not yet ordered the parts for it.

[mawyatt]

Notes for the OP.

The 5.6uF in the HP design limits the Low Frequency response (our IM3536 goes down to 4Hz) as it's in series with the Force Signal (Hcur), especially important for measuring low Z DUT at lower frequencies as this forms a voltage divider with the DUT and creates a small voltage at the Sense terminal (Hpot) which is more difficult to measure.

In our rendition we included larger film capacitors (lower voltage tho as we don't generally require anything above 100VDC) for general use, and included ability to add back to back high capacitance electrolytics (jumper terminal on PCB for this) for use with low Z DUT devices.

After much use we also added the ability to quickly charge or discharge the DUT with a push botton, this aids when evaluating larger capacitors as they can take a long time to charge to steady state DC bias. This also serves a similar function as the HP interlock, dumping charge into the power resistor for discharging the DUT.

The only issue we've had with our version, which can be directly connected to LCR bench meter with BNC Male to Male adapters and directly connected to standard fixtures (Like SMD Fixtures) without adapters, is that the spacing of the BNCs was slightly off, someone we know well didn't pay attention during PCB layout

We couldn't find anywhere the nice lever arm Male BNC used on the bench LCR meter fixtures, so had to resort to BNC Male to Male adapters for the direct LCR meter interface, of course you can use a standard BNC cable and we have some short ones just for this purpose.

Also we included the ability for the external charge discharge switch with an JST connector (now ours has a PCB wire soldered to a component, sloppy but works). We have an updated version of the PCB and were considering ordered PCBs if interested.

Make sure your version has complete electrical shielding (ground) and is mechanically stable, this is why we chose the extruded aluminum case with aluminum end plates (covered with 3D printed plastic for text and feet), see link Kean mentioned.

Edit: The components you've selected look fine for the HP version, good choices

Anyway, good luck with your implementation, we're here to help if need be.

Best,

[ZGoode]

At this point, I would say revision 1 of the draft is complete.  Based on the components I've selected, I feel pretty confident in saying this is probably safe to around 500VDC+, although I probably will never use it past 100-200VDC.  Yes, I am aware there are technically 3 grounds on the board that do not appear to be connected.  My plan was to connect them through the aluminum enclosure using the screw mounts I have added to the PCB to keep the board from being too crowded with ground traces all over the place.  I'd love some feedbacks on my thoughts here, since I'm not sure if this is a good method or not.  In my second revision I might change that, since with this method the low side would technically always be connected to earth ground.  Maybe instead connect them through a separate copper or aluminum plate isolated from the outer enclosure?

I'm also a little bit worried about how the reed relays might hold up over time, but it is surprisingly hard to find higher voltage relays that are default normally closed.

edit: I'm probably going to swap out the 1N4936 for 1N4007 and UG56G for 1N5408G due to higher voltage ratings

[mawyatt]

Don't think relying on a screw to complete the ground connections between PCB layers is a good idea. Believe connecting the grounds directly on the PCB and then having the screw(s) make contact with the shield is a preferred method.

Not sure what the frequency range of the old HP fixture is, but the 5.6uF and 1uF might limit the low end. This is why we chose the larger values in our fixture.

Mechanical and electrical stability is key to making accurate and repeatable measurements with these fixtures. How are the input/output BNCs connected? Are the BNC outer shells connected to the shielding enclosure?

Best,

[ZGoode]

I guess I could use the top layer as a ground plane, since I didn't route any traces there.  I was thinking of doing it that way to try and minimize capacitance, but you're probably right.  I might try and reorganize the components to see if I can connect everything together without running any signal traces directly under a component or another trace.

As far as the frequency range, according to the Keysight accessories catalog it should be good from 50Hz up to 2MHz (I've attached the chart below), although I have not tested that claim, so I do not know how well it functions at the extremes of either end of the range.  I'll have to take a look at the values you chose for your design.  For what frequency range does yours appear to be effective in?
So if I am reading your schematic correctly, you went with a larger value than the 5.6uF, but still used a 1uF for the second cap.

My plan for the BNC connectors was to directly mount them to the aluminum enclosure as you described.  I'm using bulkhead BNC to coax on the other side.

For connecting the BNC connectors to the PCB I was planning on using spade connectors.  I feel like since this is relatively low frequency that should be an acceptable solution.  Or should I go through the hassle of just soldering the inner core of the coax directly to the PCB?

[mawyatt]

Recall we used multiple 10uF metalized mylar film types, 2 for the Hcur and 1 for the Hpot, also provisions for larger electrolytic types.

Spend some time over on the link Kean pointed too, there's lots of details there.

https://www.eevblog.com/forum/projects/bias-network-for-lcr-meter/msg4133314/#msg4133314

Best,

[ZGoode]

I think my plan for revision A will be to use the capacitance values as is so I have a reference base point, and then update that in a future revision if the performance isn't as desired.  I'll probably go ahead and order various capacitance values of the same footprint so I can swap them out and see what the performance difference is like (ie 5.6uF, 4.7uF, 10uF, 22uF, etc).

As for the other considerations you mentioned such as the ground plane, I'll look at rerouting the traces and components to see if I can further improve on it.  Ideally, I'd like to keep it 1 layer, but if it is unavoidable I'll turn the second layer into a ground plane with cutouts under critical components to try and minimize stray capacitance between layers.

For the case, I'm planning on cutting and bending some aluminum sheeting in a similar shape to the old HP one and then powder coating it (leaving bare spots where connectors are supposed to attach and ground to it)

Your's and Keen's feedback has definitely been helpful here.  I'll let y'all know how it behaves after I get it all setup and tested.  Currently, I don't have any way to test the low frequency ranges since I only have a 4277A.  I might be able to use one of the E4980A's on campus if there aren't any projects currently relying on them.

[mawyatt]

Mentally trace out the DUT current path, that's the important traces as the DUT current travels thru the DUT, and the ground is for shielding as the DUT current travels from the Hcur to Lcur with very little bleed off to ground by capacitance or leakage if kept reasonably low.

Keep Potential lines thin and Current lines thicker.

Open Calibration removes most of the parasitic shunt capacitance effects like cables/connectors, PCB and shields, while Short calibration removes most of the PCB trace, cable/connector and the added series components (discrete capacitors) resistance/inductance effects.

Note for the Calibrations to be effective, the mechanical and electrical environments must be stable, a common source of error is the cables (if utilized).

Best,

[ZGoode]

I've gone ahead and mostly implemented your suggestions.  I've laid out a ground plane on the top/bottom of the board that avoids going under important components.  I've attached a PDF and gerber files below.  As far as the capacitor values, I am going to go the route I mentioned earlier and order capacitors of various values to test the effective working frequency of.