binding post to BNC impedance analyzer test fixture?

[coppercone2]

for a dip socket might that look like a disk capacitor/mov made out of solid copper? or would you want a disk that sits in the socket like a small flying saucer landed (like a solid photoresistor)?

[TimFox]

The DIP socket is not ideal:  best is something shaped like the BNC shorting cap with a center conductor connected to a cylindrical shell, grounded all the way around.
If not cylindrical in nature, you have a small inductance that may be acceptable, depending on wire length and frequency.
There is a lot of literature on the inductance of lead and bond wires.

[coppercone2]

it makes me want to braze a staple to a old penny for the ZIF socket

[jonpaul]

in our research for digital audio transmission we use HP4194A with 41952 impedance fixtures

Fine résultats 100k to 500 MHz with genuine HP / Yokogawa test adapters.

Can measure 0. 3 pf

These are all unbalanced, no transformers can be used without affecting the results.

See HP4195A spec, app notes and HP long note on impedance fixtures and measurement

https://www.keysight.com/us/en/assets/7018-06727/brochures/5965-4792.pdf


Jon

[coppercone2]

I have a APC-7 to bnc adapter. I put a single piece old school bnc to banana adapter on it.

If you are curious, with this addition, after calibration at the port, it reads like 6pF at 1MHz

I also got the resistors (those 50 ohm RF leaded ones) finally.

The 16092A fixture, reads picofarads above say.. 10 MHz (under that it reads inductance open circuit). The measured value is ~1pF up to 1 GHz, even if its rated for 500MHz. It gets more resolution as you go up in frequency.  I tried putting those 50 ohm resistors in there before, but its annoying because I don't have the other half of the fixture for through hole parts, and I feel like its going to get damaged. its priced scandalous on ebay. It was sitting for a while now, not cleaned at the test point, when i connected it shorted, it read 300 miliohms at 1MHz. I had it read down to 20 Miliohms before. I have a feeling its the instrument and not the fixture. Maybe I should replace the center pin of the APC-7 connector. I might try a short duration electroclean on the gold stuff  cleaned it very well and recalibrated with everything cleaned. Now its repeatedly giving ~12 miliohms short circuit on the 16092A fixture at 1 MHz. The short circuit value bobs up and down by maybe 2 mili ohms on the fixture. I wonder if its supposed to be dead stable. Maybe its better if the machine stabilizes, this was a test with maybe 10 minutes of on time. I don't want to leave this machine alone since I saw the shuffer cap start smoking in it! ill have to park my ass down there to do a real warmed up test.

If you are interested in deoxit gold, the difference in the test fixture between clean and then swiped with deoxit on a chamois  wipe is not visible.


The APC-7 to BNC adapter by itself reads 3pF, rising to 15pF at 1GHz when open

The APC-7 to BNC with the BNC to Banana (the shorter one that rotates the whole body into position, without a slip joint) reads capacitance to around 625MHz, then it goes inductive. But its alot of capacitance at that point.

When I put the 50 Ohm resistor in it, I get a phase shift of like 0.04 degrees at 1Mhz. At 100MHz the phase shift is 7degree

Waiting for the 93 fixture with banana connector and also the shorting plate that goes under the test fixture.

Just through those measurements might interest people that were interested in what they actually see on the equipment when its used..

The difference between the narrow band load in the APC7 mechanical cal kit and the wide band APC-7 load that was a random purchase is something like 0.2 ohms off from 50. A bit over at low, a bit under at 1GHz.

Is looking at the capacitance of the adapter a good way to determine if there is "too much crap"?

[coppercone2]

if you are curious about electrocleaning the gold fixture stuff with alkali cleaner, basically the discolored smut on the surface of the fixture turns black. If you let it bubble a bit some will come off . Then if you wash it and gently rub it with baking soda on your finger it comes out looking nice and gold

like you can put baking soda paste into it after cleaning an swirl a q tip around in there and the inside comes out looking fresh. no pressure just slurry moving around

the apc-7 connector bit took a few alternations. I bet if you apply any sort of pressure in there, it will also clean up nicely, but it seems kinder to the part to use electricity and just slurry flow rather then hard abrasive scrubbing

[coppercone2]

I got the 125MHz test fixture 16093B and cleaned it as above

If you are curious about what you will see:

At 125MHz (the rating) the OC impedance reading is 250 Ohms.
With the APC-7 to BNC to banana jack kludge, the impedance reaches 250 ohms at ~77 MHz.

By going with that metric, that is... what is the maximum impedance you can measure at what frequency, the performance is roughly doubled by using HP test fixture.

also beware the spring clip test fixture. alot of them on ebay have broken acrylic bushing. the APC-7 center pin is loose then. thats why my short circuit impedance readings were shit, the seller deceived me with creative backwards pictures. I used the APC-7 gauge and it was giving me odd readings that were inconsistent. the plastic bit is smashed and it was shifting  . What a fucking dick wad

[jonpaul]

Coppercone, we had a lot of experience with HP 4195A and the impedance test sets and fixtures.

You can occationally find these off epay, at costs perhasp $100..500.

The original vinatage HP/Yokogawa Japan fixtures are APC and interface to other analyzers besdies the 4195A.

The SMD, binding post and otyher fixtures are the best.

We used the system since 1992 in our transformer developent to 500 MHz.

Enjoy,

Jon

[coppercone2]

did you do any calibration at the fixture save for electrical length setting?

I.e. use calibration grade 50 ohm SMD/leaded resistor, some kind of short circuit (golden block), open circuit (acrylic block)?

For calibrating a 50 ohm on the banana / screw terminal jacks, you would need a much longer 50 ohm resistor. I have seen these in a old HP amplifier. Like a thin long resistor that basically fits between the jacks with no leads. Or some kind of clever resistor soldered to low impedance bus bar to put into the through hole fixture.


I.e. do you have a part # for a calibration grade RF SMD 50 ohm resistor?

I got RF resistors that say they are good to 3 GHz (sprial cut) and 1% accuracy @ 50 ohms. They have leads. I kind of wonder what accuracy I am getting if I perform a calibration at the screw terminals with custom standards (say a copper strip, the resistor and nothing) to get extended bandwidth from the fixture.

[coppercone2]

also have you used the temperature test fixture ever?


I don't understand the high temp fixture HP has. At first I thought the ceramic looking thing might be a heat conductor, but it looks like the fixture itself is expected to go into a oven or something? I don't see any power connections or heaters in the diagram.

What I thought might work is to have a glass tube covered in a fiberglass braid blowing hot air from a hot air station on a through hole part. Then to some how measure the part. I assume this would need to be a thermal camera. I thought that maybe you can put a tiny dot of black paint on it, then use a thermal camera with a zoom lens to zoom in on the painted spot to measure temp. I have the seek thermal, which I built a stand for that has a remote cable. That way I can put a i phone on top of the IA so I can get a temp readout while I am looking at the instrument readout.

I thought this might give a reasonable temperature test for a leaded component.

[jonpaul]

Cu: 

This topic could be moved to from RF to Measuements group. I would also post to Groups.io>>HP forum.

So,  what is the C/L/R min/max you seek to measure .... what is your objective?

We have  HP 4195A (similar to 4395A) and 4275A RLC, as well as Wayne Kerr 3245 inductnce analyzer. HP 4195A may be similar in some test fix and CAL/
See complete app notes and manual on the analyzers, test sets and fixtures for CAL info.

Best gen ref is the classic HP/Keysight Impedance Measurement Handbook A Guide to Measurement Technology and Techniques 6th Edition
https://www.keysight.com/us/en/assets/7018-06840/application-notes/5950-3000.pdf

1. Every Impedance/Network/RLC   requires a  full CAL for specified accuracy over the band.

Open Ckt
Short Ckt
Zo 50 Ohm

2. Each test fixture kit includes  an APC-7 or other calibration open/short/Zo standards.

3. After full CAL the analyzer will process the CAL for each freq.
We leave the Zo 50 Ohm in place and check for  flat 50 Ohm -  j0 across the band eg 100 kHz..500 MHz.

4. All DIY CAL attempts at open/SC/Zo will be OK at LF but have some L or C >  5 MHz.

5. We get pure Cu sheet 10..32 mil and small flats, have then  Au plated to make DIY SC eg for SMD fixt.

6. We have never used the HP high temp test fixtures, as our transformers have little change with temp.

7. Some LCR bridges have a lead legnth comp or switch eg 0.5 m or 1 m lead comp.

8. With HP 4195A,   full CAL at 100k..500 MHz we  easily measure  0.3 pF cap Pri..Sec in a guarded shielded transformer.

At these low C, a ridgid setup is required for repeatibility.

Hope this info is interesting.

Bon Soiree,

Jon

[coppercone2]

I have the 16092 and 16093 fixtures and the APC7 mechanical calibration kit with all the things in it. But the machine only calibrates at the connector with the APC-7 open short load, and it accepts electrical length for the compensation, and you can use the little grounding spring/bonnet.

the 3 terminal binding post fixture is 125MHz max. The two terminal one is 250 MHz (don't have it), and the SMD one is 500MHz.

What I am curious about is with the correct standards, can the maximum frequency of the fixtures be increased. A short and open for the SMD test fixture seems doable, it would just be a metal block like a SMD resistor and a acrylic/ceramic block of the right size. But then the load would require a calibration grade SMD part. Does this exist? Are there other parameters of the fixture, besides lack of standards, that limit the frequency?

I.e. if you got metal and dielectric blocks the same size as a 1210 component, and a really good 1210 resistor, could the maximum frequency of the fixture be pushed past 500MHz?


And also curious, if you did the APC-7 load calibration, but did the open and short calibration inside of the fixture (with custom standards the size of the component you are measuring), could this some how increase accuracy, or will it just get messed up? I assume it won't work, but I don't know exactly how it works (very complicated). That way your open and short is nearly perfect, but the load (due to lack of a calibration resistor) is done conventionally, without the test fixture installed. And if someone knows, it would be interesting to know what the machine actually calculates if you did the open and short in the fixture but the load with just the connector. Even if its totally worse, I just wanna know what happens if you do this.  I assume part of the problem is the electrical length correction, which would be applied to all 3 terms, but it would only be required for the load, not the open or short? i.e. if you did O/S cal in fixture, it would still assume the electrical length is 0.34, even though its pretty close to 0 for O and S, but actually 0.34 for load, since its not part of the fixture if done that way.?

[jonpaul]

Cu: AHA! Now understand you points better. Please reread  my edited last response   link.

1/ The BP and SMD adapters indeed have only a SC   included.

2/ CoilCraft has done extensive work in exactly this area with the same test sets and fiztures.
See their excellent notes on standards for CAL.

3/ Our 4195A is speced to 500 MHz max and no work can be abouve that freq.

Other similar like 4395A may go higher.

4/ The test fixtures for impedance have transformers, bias nets and other components that are limirted in freq.

41952A is 100 kHz..500 MHz only.

Jon

PS: Excellent Coilcraft app notes may have some relivance to your issues on calibration:

https://www.coilcraft.com/en-us/resources/application-notes/

[coppercone2]

the next SMD test fixture came in, this one also has damaged acrylic. Instead of letting it fall apart like the other one i dripped about 0.075 ml of weldon #4 acrylic cement on top of the cracking acrylic with a syringe.

Hopefully it seals any damage. In a week I will test with APC-7 gauge to see if pin did not shift.


This acrylic APC-7 test fixture dielectrics seem like a time bomb. I wonder if you can replace it with 3d printer.

[coppercone2]

Well I got a question, it says the spring thingy could be used to cancel residual L+R. Determined I should not wait a week because I am so bored

When I put the fixture together and short it out with the spring gold elbow thing, the inductance at say 50MHz reads something like 0.6nH

When I put on the grounding skirt/terminator, the inductance actually reads higher, at ~1nH

Is this correct or is something messed up?

I figured when you screw the skirt on, the inductance would be lower then the fixture shorting itself. But it turns out that if I did cancellation, I would get negative inductance.

Is it only for use with a part and the methadology breaks down with a self short circuited fixture?

Like, if I store the ground terminator values, then actually measure a part, it will be more accurate, even if it does not agree with the fixture being shorted out by the gold spring loaded elbow piece?

At 500MHz
1) test fixture shorted by itself : 1.1nH + 100-120 mili ohms resistance (varies)
2) test fixture shorted by grounding terminator skirt : 1.58 nH + 75 miliohms (very repeatable down to maybe 2 miliohms).

So the L is higher but the R is lower on the L + R measurement.

is this right? why the hell is residual inductance higher with the special shorting terminator then with the shorted fixture. Is shorting the fixture by itself some how bogus and or has a error?

Like just looking at it, when I see the fixture gold angle block pushed up against the post with the spring, it seems to me like the inductance has to be lower then with the skirt. The skirt is like big, it fits nicely but its still the size of a coin. I don't think the measurement is incorrect. Is it just a general case? Or is it only meant to be used with the 2x clip, rather then the plunger?'

https://www.picclickimg.com/ZWQAAOSw7K5kqcHM/HP-Agilent-16092A-Spring-Clip-Fixture.webp

I just can't see the grounding skirt having a lower inductance then the shorted plunger?! I could understand that the skirt has a lower resistance, since there is like 30 points of contact, rather then some angle piece being wedged against a post.. but inductance just must be higher no matter how wide you make it because its damn big. You could make it 1/2 as big and it would still work. Its over sized as far as I can tell.

And yes its the correct one, I know there is a different skirt for the binding post fixture. This is the skirt for the 92A, not the 93. Infact I would call this a ball room gown from the 1300's of spain, not a skirt. This skirt would be more then modest even in the 1800s. Explain this to me. Why is it so big. its the size of a quarter but you can probobly make it the size of a dime. 

And the plunger/gold angle is fine, I have a few of them now because of failed aquisitions. They all read very similar when I swap em.

Is the grounding skirt only for the 92A fixture configured with the big top slide piece and NOT the plunger, or should it be used anyway for real parts measurements?!

[jonpaul]

Dear Cu: fine questions, I cannot help....
I Suggest you  pose this question with pix at
https://groups.io/g/HP-Agilent-Keysight-equipment/topics

 That forum has  HP afficiandos and ex HP retirees. They can help....mention the analyzer (4195A?) and test fix #s.
As I have  4195A and   different fixtures, I cannot assist.

In practical transformer deveopment to 500 MHz,  inductance >> 1 nH
 Our C meas. is down to to 0.3 pF.

 
Bon chance,


Jon

[coppercone2]

lol that is the email box killer website

Hopefully someone here measured the skirt ground on the 4191A.

I am guessing its because if there is a part, it will be high up pinned against the center post, so the whole inductance will be factored in, and when the angle piece is wedged against it, its near the base, bypassing the majority of the expected inductance.

Like it might be 2-3 mm off the ground, but the shorted fixture might be making contact 0.1mm off the ground. Unrealistic conditions for most parts save maybe <0201

[jonpaul]

Cu:" lol that is the email box killer website" (Groups.io)

have sued many forums in Groups.io for years no issues with spam etc.

Like any forum or site (incl EEVblog) one must tweak comm parameters as the defaults often have a lot of useless nagsa sent via email.

Easy also to hide your real email adr.

Found many great connections and deals on groups.io

j

[coppercone2]

I wonder if there is a neon element that I can plug in that would function at 40VDC to measure impedance of it

[coppercone2]

Cu: AHA! Now understand you points better. Please reread  my edited last response   link.

1/ The BP and SMD adapters indeed have only a SC   included.

2/ CoilCraft has done extensive work in exactly this area with the same test sets and fiztures.
See their excellent notes on standards for CAL.

3/ Our 4195A is speced to 500 MHz max and no work can be abouve that freq.

Other similar like 4395A may go higher.

4/ The test fixtures for impedance have transformers, bias nets and other components that are limirted in freq.

41952A is 100 kHz..500 MHz only.

Jon

PS: Excellent Coilcraft app notes may have some relivance to your issues on calibration:

https://www.coilcraft.com/en-us/resources/application-notes/

What is the limiting element on the spring clip test fixture? I took it apart, there is just gold plated things.

Also, I got a HP balun thing, for measuring differential impedance. I see there is alot of options available for higher frequencies. I guess it might answer those burning questions I had about twisted pairs though. I just can't sleep at night thinking about whats going on when you twist some wire together. How could anyone given all the shenanigans around wire insulation properties and how neat your twist is. 

[coppercone2]

got the transformer. I tried a twisted pair of 18 gauge banana alligator clip cables that I loosely twisted.

Short circuit was 8.4 ohms, open circuit was 412 ohms. I believe its (OCR*SCR)^(1/2)

I got 58.8 ohms @ 1MHz

Is this right? why the hell do I need to know the impedance of a twisted pair anyway?! . What the hell am I going to do with this information?

But actually, how do I do this right? The balanced impedance is 100ohm, the unbalanced impedance is 50 ohm.

I am guessing I do the OSL calibration with 100 ohm instead of 50 ohms on the load?

When I measure 100 ohms, it reads as about 50. So my cable impedance is like 58.8*2, or 117.6 . Maybe I will go through seeing how calibration changes things when I construct a good 100 ohm load. who else is super excited about cable impedance?