VNA test plate/fixture

[CopperCone]

I want to use my 300MHz vna to measure some custom large inductors I made. Alligator clip BNC cables have not been working well, they are very wonky.

I thought that I could glue copper tape on a piece of wood and solder BNC connectors to it. I would like it to work to 300MHz.

I can't afford to buy anything, but I do have copper tape, copper sheet, wood and BNC connectors.


What should the space between the top and bottom layers be? And what dimensions can I do?

I would like to make one thats maybe 4 inches long and one thats maybe 15 inches long. Should I make two of them? I figured that smaller spacing = bigger capacitor between the ground plane and the contacts, so it would be better to make two different ones

What's reasonable for width and thickness on the two test fixtures? The idea is just to press or solder an inductor to the top of it and measure it. And how far apart should I space the top contacts? If it does not matter I would only prefer to make one big one with a small space between the electrodes so that it fits all types of filters/components.

The volume of things to test on this is probobly very low, so I am ok with using copper tape that is not particularly robust.. but if its considered a good solution I will consider it making out of copper sheet thats epoxied to wood for strength.   

[Andrey_irk]

I would make a PCB with a 50Ohm trace (which matches the VNA's input impedance) and a footprint for a DUT. The PCB is just a two-layer board (virtually any thickness will be OK and the length say 5-10cm), which can be made at home and it will be way more convenient to use than the thing made of wood and copper tape.
And you can even add a trace for calibration (or even a whole cal kit, which is not a must on these frequencies I think).

[CopperCone]

Problem is the footprints are variable, I am winding various hand made coils.

If I designed a 50ohm trace, would the stub left behind from different sized coils or capacitors effect thing significantly?

Some of the coils I have are 15 inches long, some are like 2 inches long. Can a one sized fits all microstrip or w/e be made? Or would I have to trim it out every time (good reason to use tape with a exacto blade).

I don't know why I did not think of the 50 ohm impedance match, long day I guess.

I would still mount it to a wooden board with little rubber feet so its robust and more professional though. I HATE random things made out of a PCB only lol

[Mechatrommer]

i think its best to do custom fixture that resembles closely to where the dut will be operating, if dut will work on top of a gnd plane, then a fixture with gnd plane will it be, if floating in the air, then i guess just connect the dut to sma end on the air will do. ymmv...

[Andrey_irk]

i think its best to do custom fixture that resembles closely to where the dut will be operating, if dut will work on top of a gnd plane, then a fixture with gnd plane will it be, if floating in the air, then i guess just connect the dut to sma end on the air will do. ymmv...

You have to connect it to the VNA with 50Ohm ports (it's not how the DUT works in the real device, is it?). The goal is to measure the impedance of the DUT mounted in a particular way. If it has to have a cutout underneath then consider the cutout as a part of the footprint etc..

 

Problem is the footprints are variable, I am winding various hand made coils.

If I designed a 50ohm trace, would the stub left behind from different sized coils or capacitors effect thing significantly?

Some of the coils I have are 15 inches long, some are like 2 inches long. Can a one sized fits all microstrip or w/e be made? Or would I have to trim it out every time (good reason to use tape with a exacto blade).

I don't know why I did not think of the 50 ohm impedance match, long day I guess.

I would still mount it to a wooden board with little rubber feet so its robust and more professional though. I HATE random things made out of a PCB only lol

1. There is no problem to make a few fixtures if they are PCBs.
2. As for the stubs - I haven't seen your device, but I doubt it will be a big issue. Nevertheless, you can calculate wavelength of the highetst frequency wave propagating in your fixture. If the stub is small in comparison with it - you'll be fine.

Can you post a photo of the coils?

[dmills]

Also, make your test fixture so that it is easy to calibrate out, doing a SOL calibration at the point you will be connecting your test inductors is useful because it means you are measuring at that reference plane.

My favourite trick for this is just to own a handful of SMA sockets and a short male to male coupler, I have short, open and a 50 ohm load made up on some spare end launch connectors, so I can just solder the device under test to a spare socket then cal the VNA, remove the calibration load and connect my test device to the same connector. 

Regards, Dan.

[CopperCone]

Can I make long 50ohm strip lines then make a mechanism that slides ontop of them for contact points and short it out for normalization at that point?

Im imagining a 15 inch long stripline with a aligator clip soldered on vertically on one side (small one) and a metal moving bit attached to a plastic screw to set the gap distance with another alligator clip soldered to it so i can adjust the gap distance, then do the normalization with a copper sheet thats cut appropriately at the right lenght.

My inductors are rougly 15x2 inches, 3x2 inches, and 4x5 inches, wound with 2.65mm wire on pvc pipes with wooden support structures

I wanna make something like this.

http://www.dpaonthenet.net/article/143367/Test-fixtures-for-vector-network-analyser-measurements.aspx

Keep in mind thats for 40ghz, i just wanna go up to 300MHz

I see that 300MHz is roughly 1 meter and a 15 inch stubb would be 40 percent or so.

Whats the max stub 50 ohm proportion thats acceptable?

[dmills]

If your stub is 50 ohms, just cal the analyser at the end of the stub where you are connecting your inductor! The concept of a VNA reference plane is useful here. 

Note that an inductor includes any loop you need to get to its contacts, so if the ends are a foot apart then you have an extra foot of wire in the loop and the inductance will depend a bit on the resulting loop area.

Regards, Dan.

[CopperCone]

Well I want to make the contacts mobile, on a fixed stripline, by a screw mechanism. I think I will just try it.

Regardless of how bad it is, the normalization should take care of the problems right?

However, if there is particular resonances/etc in the test fixture, then even normalized (digital process?), the dynamic range of the measurement there will be reduced right? Or do VNA typically dynamically reduce signal power to maintain dynamic range during normalization? I have a E5100A I got real cheap, otherwise I would not have a VNA.

Like say by some weird property, the test fixture has a 80db peak (at the top of the measurement range) at 250MHz, and this is flattened by normalization. During the sweep does the sweep signal get attenuated by a dynamic gain stage to prevent saturation, or is it purely a digital offset? (in the case that the test object makes it even worse, and adds another 40db of gain at 250MHz, which in the digital case would cause saturation/gain compression(correct term?)).

I assume this could cause additional errors not cancelable just by correction though right? Like complex interaction between the test fixture impedance mismatch and the part being tested impedance mismatch, antenna effects (small?) and the mutual inductance due to geometry of the clamping system? I assume you can't get perfect results with a complete bullshit setup... or can you??

Can you clarify what factors are perfectly nulled out and which ones cannot be (if any)?

[Bud]

Passive circuits such as a fixtures or coils do not have gain. You may want to read a bit about VNAs before  jumping to using one.

[CopperCone]

Yes, I did measurements just now with some BNC connectors. Without some kind of special setup I found foil capacitors rather difficult to catagorize.


I think alot of it has to do with the 50ohm impedance requirement for the VNA. I think I want a impedance analyzer to do single components.

[rhb]

I suggest that you make separate stripline connectors for each end that have a common mounting arrangement for attaching them to the coils.  Arrange so that you can secure the connectors to a a small sheet of plastic with good dielectric properties.

For example, a small piece of copper with a half hole soldered to the trace.  Take two identical pieces, clamp them together, drill a hole the correct size along the contact surface.   File one end so that it has a taper at the BNC end when soldered to the stripline to minimize transition reflections.  This is just to ensure good contact between the wire and the fixture.  Make a couple of larger clamps in similar fashion from polystyrene and then use some nylon screws to clamp things in place.  Make one for each wire size you're using.

I bought a VNWA  3E and shall be doing this sort of thing myself, but I've got lots of projects going, so when is unknown.  But you resolved my question of how I was going to do this.  I'd pondered the problem, but did not yet have an answer.

Have Fun!
Reg