Poor performance of DIY SMA-M Precision Load

[Randy222]

thats all like reinventing the wheel..

Not really. The only reason I selected a PCB in the footprint of the bulkhead, was to make soldering the tiny SMD's a tad easier.

I am now working on a 50 x 1 on a modified board.

[szoftveres]

I see that everyone here is using the LiteVNA cal kit as the reference - is it really that good? How does it compare to an actual reference cal kit?

[joeqsmith]

I see that everyone here is using the LiteVNA cal kit as the reference - is it really that good? How does it compare to an actual reference cal kit?

Appears yours is the first mention of the LiteVNA cal kit in this thread.  Define good?  Wear?  Mate cycles?  Cost?  Repeatable?  Characterized?   

[szoftveres]

Appears yours is the first mention of the LiteVNA cal kit in this thread.  Define good?  Wear?  Mate cycles?  Cost?  Repeatable?  Characterized?   

Hmm, did I use the word "everyone", when actually there was only one mention in this thread of using the LiteVNA cal kit as a reference? I apologize for that!

[joeqsmith]

Appears yours is the first mention of the LiteVNA cal kit in this thread.  Define good?  Wear?  Mate cycles?  Cost?  Repeatable?  Characterized?   

Hmm, did I use the word "everyone", when actually there was only one mention in this thread of using the LiteVNA cal kit as a reference? I apologize for that!

None of the standards supplied with any of my low cost VNAs were characterized.   

All of these low cost standards have all been SMA, not a precision 3.5mm.

None of the low cost short or open standards allowed a wrench to be used to lock the center conductor.  Some of the thru standards did not provide a flat for a wrench.  The thrus supplied with my LiteVNA64s have all had a flat.   Short and open standards were modified. 

All of the low cost standards I have received were shipped in a bag loose, free to bounce around.  I've seen many a ham treat connectors this way.   

The difference between the a virgin set of open and short standards supplied with the LiteVNA64 matches very close to the ones supplied with my V2Plus4.  See the following:
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5666353/#msg5666353

All the low cost loads I measured had poor return loss.  I used one of them to create a low cost T-Check.  The load I use for calibration was a hand selected terminator from Mini-Circuits. 

Characterizing all of that mess with my PNA (against an ECAL that was calibrated to a set of known Agilent standards).  Using METAS VNA tools to derive the polynomial coefficients.  Using Agilent's cal kit editor to create  a new user cal for them.  Then using these standards to calibrate the PNA,  it would pass a T-Check within 10% over the entire frequency range (300k -9G).  See:
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5676077/#msg5676077

Using these same standards with the LiteVNA64 yields poor results.  It does not matter much if I use the METAS fit data, or the databased S-parameters.  The same math can be ran with the Agilent PNA, it it works fine.   My only conclusion is that if you were to put together such a set of standards, they would not be the limiting factor when making measurements with the LiteVNA64. 


How they wear over time, assuming you took very good care of them, hard to say.  I doubt they are rated for very many mate cycles.  The LiteVNA64 is a great system for the cost and very useful.  But I don't think all of my low cost VNAs combined = the cost of one good short standard from Keysight.   

[Randy222]

The LiteVNA64 is a great system for the cost and very useful.  But I don't think all of my low cost VNAs combined = the cost of one good short standard from Keysight.

True.
People buy what they can within reason, who would of thought a decent 4ch 200MHz 12bit digi o-scope can be had for $275(GBP). 20yrs ago that would have been $4k(GBP).

Here's my next test, 50 x 1 done two times. Item "B" is better than "A", but both are fairly wavy on the VNA. I soldered a grounding band to the bulkhead.

I cannot get "flat" reponse past 3.2GHz.

The 100 x 2 setup was the best of the bunch.

0603's are a challenge to solder, so I might try bigger SMD's.

[coppercone2]

The steel ones might be easier to solder with the right flux, because they transfer heat less.

I thought about how to do this properly alot. copper spot welding does kind of work. I thought to spot weld a copper shim to the body, so it can be bent into a U shape laying on the side, so you can solder a part to it with only a reasonable amount of heat, and minimum extra inductance, and after you spot weld it to the shim, bend and squash it into place along the center conductor. Spot welding copper to copper sheet does seem to work, but I have not tried thin to thick bonds. Or, spot weld a nickel strip, and then pen plate it with copper.

It seems like the assembly of these loads is the hardest part, there is alot of options, but everything solders like crap.

Another option is to maybe bolt a fabricated fairly thick copper shim with a hole in the middle to the conductor with alot of tiny bolts, and solder a resistor to it, so when you bolt it down you only need to solder it to the center pin.

I imagine in almost all the cases, the resistors are severely stressed by the huge thermal mass of the connector, its not getting anything close to a proper solder profile. I wonder if it has a adverse effect on the RF properties of the part, like reduced stability.


Or indium. 


The fact that the soldering profile is so out of whack makes me feel like all of the DIY standards are flawed compared to whatever magic they use for proper ones, but I think there is some solution that will solve this problem and let you confidently solder a expensive RF resistor down without violating its soldering spec.

And if this was figured out, you could try to use whatever connector you please, not the one you have to find does not start spewing plastic during the soldering. For the higher frquencies, I think teflon might even be disturbed because of the giant thermal mass of the chassis connector, which is designed to be screwed in.

[Randy222]

I have a decent spot welder for doing the nickel tapes to batts.
Maybe I can flow solder to pad, and then try to spot weld a SMD to the pad.

Some 0603's don't have a wrap-around tin end. The 50ohms I have have a sliver of tin on just one side, so that makes it hard to spot weld into place.
I am also working on laser spot welding.

But you gave me an idea for the other SMD's I have.