Cables and connectors for VNA < 3 GHz

[mark03]

I'm researching a low-budget VNA system for hobby use, either a 8753x or another circa-2000 instrument.  Although I took prop&rad and transmission-line theory in school, I'm not an RF expert by any means.  I've learned a great deal already just by reading through old threads here, so thanks to the regular contributors for that.

Much of my interest is at really low frequencies (HF) but I would like to put together parts with adequate performance over the range of the instrument I end up with: 3 GHz for an 8753x or 1.8 GHz for a 4396 combo analyzer.  I think I'm going to make my own cal standards from quality SMA connectors to start with, and perhaps eventually get them characterized on a lab-grade system so I can use IN3OTD's Octave/Matlab script to derive my own cal parameters (https://www.qsl.net/in3otd/electronics/VNA_calkit/calkit.html).

Given this performance level and frequency range, how careful do I need to be in my selection of test cables and adapters?  I will likely need to procure N-to-N cables for the connection between a test set and VNA, SMA-to-SMA cables to the DUT, and a pair of APC-7 to SMA adapters.

I gather there are two distinct benefits of the expensive cables (Gore, Minicircuits)---good transmission parameters (return loss and attenuation) at high frequencies, and phase stability.  Reading between the lines it seems like the former may not be so critical below a few GHz, but I'm not sure about the latter.  Is there some middle ground between the $$$ professional stuff and generic Ebay patch cables?  I'm not averse to spending a bit more for quality, but wonder if lab-grade cables are overkill here.

Are Ebay APC-7 to SMA adapters generally OK if branded, e.g. Amphenol?  I am a bit worried that I may not be able to tell visually if they (or for that matter a used test-set's) have been damaged.

On the SMA side of things, should I be looking for HP's 3.5-mm SMA-compatible connectors which supposedly have better performance, but are susceptible to damage if mated with a low-quality SMA?

[technogeeky]

I am not an expert either, just responding anyway:

1) I think for VNA measurements, the return loss and attenuation are not that important. Both will be calibrated out every time.

2) For the most part, you can get your phase stability by taping down cables and being careful not to disturb them in between measurements.

3) Neither of these factors are anywhere near as important as your calibration kit.

At 3 GHz, even relatively low quality SMA cables are more than sufficient. I don't think there is any need to have 3.5mm connectors at all, this will only drive up cost. If it is forced upon you, then just use a sacrificial connector to convert to standard SMA and don't ever remove it.

Lastly, you may want to ask this in RF. Maybe not.

[DaJMasta]

The phase stability is important, as said, so minimal movement, or better yet, phase stable cables are your best bet.   And yes, calibration gets out a lot of the potential for losses and such, but lower loss then just means higher dynamic range (by a small amount, sure, but you want at least decent cables).

I'd focus on getting something that is durable - good plating and dimensions on the connectors (a good brand name, basically), and something that is flexible, but which isn't too thin or with a small bend radius, and then focus your cable/connector budget into your calibrator.  Shouldn't be too bad at 3GHz or below, but a good SOLT calibrator is key to getting good measurements with a VNA - and for the sake of zeroing out all your connectors - figure out what interface you want to your DUT (probably SMA, but maybe N), and get a calibrator that fits that - not some intermediate cable connector in the system.

If you get a unit without a built in test set, I like short semi-rigid coax for the connections between the source, reference, A, and B connectors.


But yeah, basically what's already been said 

[TheSteve]

This thread has some useful info in it:
https://www.eevblog.com/forum/rf-microwave/must-have-rf-plumbing-for-the-lab/

Speaking of DIY cal kits. I have spent a little time lately trying to derive the cal parameters for a 3.5mm 26.5 GHz cal kit that I have no parameters for. I have decent results to around 18 GHz but my equipment isn't cutting it above that.
Once I have had more time to experiment, and maybe have a friend with better equipment measure the standards for me I'll be posting about it in the rf-microwave section.
At this point I don't think it would be too tough to make a DIY 3 or maybe 6 GHz kit with reasonably priced parts. How repeatable it will be and what the possible phase angle error is has yet to be determined. It is certainly something that interests me though.

btw, from time to time there are good deals on commercially made cal kits on ebay - they sell fast of course. If you have an HP VNA it sure is nice to have a proper HP kit as the parameters will be in the analyzers firmware. One user kit can usually be added to HP VNA's though. Tt is quite easy to do via GPIB, and generally possible but fairly ugly to enter via the keypad.

[mark03]

Thanks.  That is more or less what I suspected re: cables.  The link in the other thread to reasonably priced cables at Mouser is helpful.  Not that I think inexpensive brand-name is necessarily better than Ebay no-name, but it does stack the odds in one's favor.

Regarding cal kits, Claudio's web site (linked in my first post) shows a way to derive the "user" cal-kit parameters from accurate S-param data, if you can get access to a properly calibrated lab for measuring your own SOL standards.  In effect, this provides a way to transfer calibration from a proper cal kit to a DIY one.  If I am reading correctly, his and G0HZU's results both suggest that this is perfectly serviceable up to 2 or 3 GHz, which is enough for me.  I find the learning aspect of this process also quite attractive.  But if I'm missing something, please say so.

[TheSteve]

Yes, that is the process I am currently working on using the software you linked and/or Metas VNA tools. I have been able to calculate values that work up to 18 GHz for an unknown 26.5 GHz kit. The limitation being my ability to measure the data from the unknown kit. I have a few 26.5 GHz cal kits I am using as a reference.
Making a kit good to 3/6 GHz shouldn't be too hard, but again how repeatable it might be is the question. It really depends on how good the SMA connectors we use are. They still need to be(hopefully) name brand with proper center insulation.

[OwO]

Below 4GHz even without characterization a "home made" calibration kit like the one shown here are good enough: http://hforsten.com/img/vna/xcal_kit.jpg.pagespeed.ic.VwORbnbUqN.webp

I made an exact cal kit like that and have no problems measuring 2 port S parameters with good accuracy. I can measure insertion loss of a cable to ~0.05dB accuracy and reflection to <0.1dB accuracy, and this is with a cheap USB VNA.

[mark03]

So I'm finding that besides the usual sources for cable assemblies (Mouser, etc.) there's quite a selection of surplus fancy microwave cables on ebay, some of them quite inexpensive.  HOWEVER, what's *really* rare and pricey are those silly N-to-N jumpers between the VNA tx/rx and an S-parameter test set.  This is clearly where the demand is, and it shows: $100/set of HP 8120-xxxx if you're patient).  So how picky should I be with these?  I see mention of phase-matching between cables, but it seems like that would be easily calibrated out.  Is this snake oil?  New generic sets of four (LMR-240, RG-214, or similar with crimped connectors) would be $120 at Mouser or $50 for "US assembled" on ebay.

@DaJMasta: I haven't seen any semi-rigid N-to-N in short lengths on ebay; guess I need to keep looking.

Edit:  Found this from Joel Dunsmore on the Keysight community forum, https://community.keysight.com/thread/17355

Originally these cables were specified to be phase matched, as the orginal 8753A with the 87046A test set had specified uncorrected performance than included phase linearity as well as loss...
[...]
In practice, it makes no difference.  And with the user of error correction, there is really no benefit at all.  So use any cables which are stable and pass the frequencies needed. 

[virtualparticles]

I have found these cables to extremely good for the price. The best value in terms of amplitude and phase stability.

https://www.minicircuits.com/WebStore/dashboard.html?model=CBL-2FT-SMNM%2B

Some say that the cable is calibrated out. But if the phase changes when you move the cables then the calibration goes out the window. You can end up with ripples in the S21 measurement because the S21 forward tracking error is phasing in and out with respect to the value which was recorded during the calibration process. Phase stability has value. I've tested these and haven't seen anything better for the price.

[Neomys Sapiens]

I totally dislike the Gore cables that came with the R&S VNA in my company lab.
1. they are too stiff
2. the N connectors on them are quite shitty for that price (don't engage easy, hard to actuate, slight cable torque loosens them too often)
3. one did go wacky with no external sign
(they are of a type with a black&violet braid cover)

Get H&S cables instead (Sucoflex as baseline, Sucotest or Boaflex = the real nice ones)

[hendorog]

Here is a procedure for how to test and classify VNA cables - from Dr Joel Dunsmore.


Step 1:  Put a load on the cable, put data->mem, then Data-Mem (NOT Data/mem), display LogMag.  A perfect cable will be like -80 dB. Then bend the cable.  Note the hightest peak, this will be the return loss or directivity stability of the cable.  If there is a bad spot in the cable, that causes the match to change with bending, it will become apparent.

A very good cable will be below -50 dB, a cable worse than -30 dB should not be used.

Step 2: Put a short on the cable,  put data->mem, then Data-Mem (NOT Data/mem) display LogMag.    A perfect cable will be like -80 dB. Then bend the cable.  Note the hightest peak, this will be the source match stability, and an indication of the phase stability.  Because the short causes a large reflection, which we subtract with data-mem, any change in the cable loss or cable phase will show up as a big return loss (in dB, you are seeing 20 log (ref-test)).  Thus, you are taking the difference of big numbers and you will see any defect in either phase, or return loss.
A good cable will be similar to above (-50) but because this tests both phase stability and reflection stability at the same time, it will show more errors and typically be worse than above.

Step 3: same as 2, but put data/mem and display phase.  Bend the cable and see the phase stability.  This will be the reflection tracking stability, and will be 2x the transmission stability.  If you display logmag, this will be the magnitude stability of the reflection and transmission.

https://community.keysight.com/message/2595

[hendorog]

For test cables, I STRONGLY recommend TFlex405. This is among half a dozen cable types I tried, the only one that does not show a significant response upon being twisted or bent.

I got mine made by OnlineCables, their factory (Applied Interconnect) supplies cables for F22 jets.

My 40" SMA-M to SMA-M TFlex405 cable costed me ~$40 each. I ordered three for VNA and SA.

This is the load test. There are also open and short tests on my channel.

Can you do the test I posted above? That is quantifiable - it is difficult to tell what is going on from a smith chart.

[Neomys Sapiens]

For test cables, I STRONGLY recommend TFlex405. This is among half a dozen cable types I tried, the only one that does not show a significant response upon being twisted or bent.

I got mine made by OnlineCables, their factory (Applied Interconnect) supplies cables for F22 jets.

My 40" SMA-M to SMA-M TFlex405 cable costed me ~$40 each. I ordered three for VNA and SA.



This is the load test. There are also open and short tests on my channel.

If they hold up to your promise, that would be sensational. At that price, you get MAYBE used/tested Sucoflex.

Another remark: I can't say if those Gore cables were endorsed or even sold by R&S or whether a former member of my company decided on them.

[Neomys Sapiens]

I have found these cables to extremely good for the price. The best value in terms of amplitude and phase stability.

https://www.minicircuits.com/WebStore/dashboard.html?model=CBL-2FT-SMNM%2B

Some say that the cable is calibrated out. But if the phase changes when you move the cables then the calibration goes out the window. You can end up with ripples in the S21 measurement because the S21 forward tracking error is phasing in and out with respect to the value which was recorded during the calibration process. Phase stability has value. I've tested these and haven't seen anything better for the price.

Your observation matches mine so far. And it is not the high-profile experiment when a new technology or product is coming to be, but the everyday 'could you do a quick check on this coupler' measurements which  are most affected, as you can not fix everything down and arrange as elaborately as in the other case. Then, suddenly, you can not say anymore if the filter you have on your desk has gone bad or if your cable acts up.

[rjyousey]

FYI, we have sold several of the CeYear cal-kits in the USA and I've had them inspected by at least one expert.  I interested email me for specifics at ry@custom-cal.com