Author Topic: Testing RF connectors and cables  (Read 7598 times)

0 Members and 1 Guest are viewing this topic.

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #25 on: September 07, 2019, 11:55:56 pm »
I got an APC-7 to N-F adapter from China today.  These are about $22-23 each on ebay from besttrade360 .

I've got 3x used Wiltron APC-7 to N-M and a couple of APC-7 to SMA-M adapters.  So I Calibrated my 8560A using an RG402 N-M to N-F jumper and an N-M to N-M jumper.  I then made up N-M to N-F adapters using various bits.  I  inserted them in the cable and swept them from 1 MHz to 2.9 GHz.

The 8560A is set to 2 dB/div, so I read the ripples as about 0.125 dB.  That seems quite reasonable for a stack of connectors.  For an 8753B/85046A system, it looks "good enough" and I've ordered a 2nd one.

Spectrum for first stack

829566-0

What the connector stacks look like:

829560-1

Second spectrum

829572-2

APC-7 to N-F spectrum

829584-3

Connector stack

829578-4

With typical Chinese prices at $45, used name brand at $100 and new UK made at $135 I thought it worth trying one.

I'll post more detailed TDR results for all the items once I get my 11801 set up again.

Reg
« Last Edit: September 08, 2019, 12:00:43 am by rhb »
 
The following users thanked this post: xrunner, Neomys Sapiens

Offline TheUnnamedNewbie

  • Frequent Contributor
  • **
  • Posts: 935
  • Country: 00
Re: Testing RF connectors and cables
« Reply #26 on: September 08, 2019, 07:43:42 am »
Wonder how they age, because unless the problem is in the aging, these show good performance.

That is nice performance though.
The best part about magic is when it stops being magic and becomes science instead
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 805
  • Country: be
Re: Testing RF connectors and cables
« Reply #27 on: September 08, 2019, 12:24:36 pm »
Wonder how they age, because unless the problem is in the aging, these show good performance.

That is nice performance though.

But even is aging is not so good, typical use for these is to leave them on the unit to act as a connector saver/adaptor, so I think this is a very useful result. Would be interesting to test them up to their rated spec (12Ghz) if someone has the proper gear for this.
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 1092
  • Country: ua
Re: Testing RF connectors and cables
« Reply #28 on: September 08, 2019, 12:47:00 pm »
typical use for these is to leave them on the unit to act as a connector saver/adaptor

$45 for a connector cap? looks like too expensive  :D
 

Offline _Wim_

  • Frequent Contributor
  • **
  • Posts: 805
  • Country: be
Re: Testing RF connectors and cables
« Reply #29 on: September 08, 2019, 01:01:40 pm »
$45 for a connector cap? looks like too expensive  :D

it "only" 22$, so you can put one at both ends  ::)
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #30 on: September 08, 2019, 03:25:46 pm »
Several of you need to understand the post before commenting.  This test is mostly relevant only to owners of 85046A and 85047A S parameter test sets for the 8753x VNA.  I don't know of other gear that uses an APC-7 connector.

The 85046A is a 3 GHz S parameter test set for the 8753x series of VNAs and uses APC-7 connectors.  Unfortunately, APC-7 never caught on so connectors and cables are generally quite hard to find and usually expensive when you do.

As a practical matter, putting APC-7 to N-F adapters on the 85046A & 85047A (6 GHz version) so that standard N cables and connectors can be used is the norm.  The APC-7 is sexless, so wear is not an issue as it is with other connectors.

Because the SOL calibration of the VNA corrects all errors to the reference plane of the calibration port, the little bit of ripple seen in the SA sweep is of no consequence.  And it is likely that the ripple is caused by all the other connectors as much as by the APC-7 to N-F.  That will be easy to determine by TDR with the 11801/SD-24.

« Last Edit: September 08, 2019, 03:55:09 pm by rhb »
 

Online TheSteve

  • Supporter
  • ****
  • Posts: 3121
  • Country: ca
  • Living the Dream
Re: Testing RF connectors and cables
« Reply #31 on: September 08, 2019, 06:25:35 pm »
I saw those cheaper(under $25) APC7 to N adapters a few months back and considered trying a pair but I had an offered accepted on a pair of Maury ones so I never did buy them. It is nice to see the performance doesn't look bad at all.
I could sweep mine to 26.5 GHz but they would mode before that. It would be nice to see how the cheap ones perform out to 18 GHz. Lower cost N connectors often start to perform much worst past 11 GHz or so.
VE7FM
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #32 on: September 08, 2019, 11:44:12 pm »
I made a deliberate decision to limit most of my gear to 3 GHz.  I did buy an 8566B so I can check for harmonics from my 8648C, but  good RF connectors and cables get so expensive above 3 GHz that I think that was a good place to draw the line.  And there is a *lot* of very fine gear that goes to 3 GHz.
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 1255
  • Country: us
Re: Testing RF connectors and cables
« Reply #33 on: September 09, 2019, 02:19:27 am »
I should note that the SMA PCB connector snapped all 4 pins when I was tightening the short to do a cal using a Suhner torque wrench.  So if you get one, repackage it with more robust connections.

Have Fun!
Reg

I have busted SMA connectors of various kinds.  Then I dawned on me....  these are made of brass....  My Suhner wrench is calibrated for stainless steel torque.  I'd check that to make sure.
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 1255
  • Country: us
Re: Testing RF connectors and cables
« Reply #34 on: September 09, 2019, 02:21:31 am »
One thing I'm very interested in knowing (and I lack proper equipment to conduct the test myself) is sample variation in various makes.  I'd expect well known brand name connectors to be fairly consistent but cheap junk will be all over the place.  But this is only an assumption.  Has anyone done this kind of test?
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #35 on: September 09, 2019, 03:28:47 pm »
The example I posted of SMA-F to N-F adapters was representative of a lot of 24.  One unit had the N thread major diameter  under size by 0.03 mm,  enough that the plastic protector fell off when I put it back on after testing it with the 11801.  However, electrically I did not notice a difference among them. They all had a similar reflection response.

From my testing, BNC connectors are the most variable.  If the male locking collar is not a tight fit on the body, the connectors will give *very* variable results every time they are used or even jiggled in use. I'll post an example later.

I suggest getting a nanoVNA for $30-60 US.  There are 3 different versions on the market, but considerable debate whether there are significant differences among them.  There is a very active mailing list on groups.io with lots of interesting things going on.  You could  use one of those to test RF connectors to 900 MHz.  There is a python script to do TDR using it.  I'm hoping to modify the FW to do SWR and TDR only on the STM32F072CBT6 based units.  I think if I delete all the other features that there will be room.    A completely portable unit  would be very  handy for antenna and feedline work, especially for field day operations.

There is a console USB serial port so you can get magnitude and phase data back as ASCII text.  That would make it very easy to scan a bunch of connectors in the frequency domain and plot the results with gnuplot.  You could also calculate mean and standard deviation at each frequency.

Because of the very high cost of RF connectors I decided that I'd buy cheap units and simply test them all.  I'm building an RF switching system to automate full annual calibrations of my test bench.  The cheap SMA-F to N-F would not be satisfactory for that, but for ham radio work below 1 GHz they are fine.  So I'll use them for a remote antenna switch.

I plan to continue this thread for a long time.  But I only have so much time.  I'd like to urge you to get a nanoVNA, order a bunch of connectors suited to your needs, test them and post your results along with a link to the seller.  That would help a lot of people.

Have Fun!
Reg
 
The following users thanked this post: _Wim_, Neomys Sapiens

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 1255
  • Country: us
Re: Testing RF connectors and cables
« Reply #36 on: September 09, 2019, 05:50:14 pm »
Great suggestion!

TDR has been on my list for some time.   I'll have to make one soon.  I think there is an already made module with very fast rise pulse - which name escapes me.  I do the same on connectors.  I buy bulk cheap.  What's obviously not good goes to garbage.  My main frustration is thin-ness of gold plating.  Use few times and I can see corrosion forming.  Oh well.

Those mini-vna looks fun.  I need little higher frequency but for that kind of money, it'll be nice to play with. 

Thanks for your effort and also a very nice reply.
 

Offline jadew

  • Frequent Contributor
  • **
  • Posts: 456
  • Country: ro
Re: Testing RF connectors and cables
« Reply #37 on: September 09, 2019, 08:54:58 pm »
One thing I'm very interested in knowing (and I lack proper equipment to conduct the test myself) is sample variation in various makes.  I'd expect well known brand name connectors to be fairly consistent but cheap junk will be all over the place.  But this is only an assumption.  Has anyone done this kind of test?

I have tested lots of SMA, BNC, TNC and N-type cables and connector assemblies, because I wanted to find a balance between price and performance for a product I'm making. What I found was that both connectors and cables (of the same make and PN) have usually very little variation in the specified bandwidth, regardless of price.

Of course the specified bandwidth varies with price, but generally speaking, within that bandwidth they seemed to be roughly the same.

After the specified bandwidth it's a different story: you'll start seeing a bit more variation, as well as intermittent flaws, and all this gets worse the cheaper the connector is.

What I found interesting was that those flaws were also consistent. They were always there for that particular type of connector.



I made a deliberate decision to limit most of my gear to 3 GHz.  I did buy an 8566B so I can check for harmonics from my 8648C, but  good RF connectors and cables get so expensive above 3 GHz that I think that was a good place to draw the line.  And there is a *lot* of very fine gear that goes to 3 GHz.

That is indeed a good place to draw the line. You can do 3-4 GHz on the cheap, but after that you can't rely on your measurements if it didn't hurt you a little when you bought those connectors. The higher the frequency, the more it has to hurt.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #38 on: September 09, 2019, 09:32:07 pm »
Great suggestion!

TDR has been on my list for some time.   I'll have to make one soon.  I think there is an already made module with very fast rise pulse - which name escapes me.  I do the same on connectors.  I buy bulk cheap.  What's obviously not good goes to garbage.  My main frustration is thin-ness of gold plating.  Use few times and I can see corrosion forming.  Oh well.

Those mini-vna looks fun.  I need little higher frequency but for that kind of money, it'll be nice to play with. 

Thanks for your effort and also a very nice reply.

Leo Bodnar sells an excellent fast rise pulse generator at leobodnar.com.  The standard unit is a 10 MHz square wave, however, Leo will provide a unit with a 1 MHz square wave which works better for TDR.  I'll post some screen shots using a 200 MHz DSO.  I have three,  a 1 MHz & 10 MHz square wave and a 100 ps pulse version.

The nanoVNA *should* do better for TDR than the pulser unless you have a 1 GHz scope or design a low cost sampling scope.  That would be a very cool OSSW/OSHW project.  Use a GPSDO for timing and an MCU with a 12-14 bit ADC.  It would be a serious challenge to design the sampling strobe circuit layout.

But for *real* TDR fun nothing beats a Tek 11801 or CSA803 and SD-24 TDR head.  20 GHz BW and femtosecond time resolution.  I got an 11801 with a bad NVRAM for $145 delivered.  It was $40 for the replacement NVRAM and I picked up a pair of SD-22, 12.5 GHz dual channel heads for $150 delivered.  The 11801 calibrator has a19-20 ps rise time.   I recently got an SD-24 for $250 from Tektronix via ebay which has a pair of the same pulsers.

The downside of the 11801/CSA803 is there is no component level service data available.  So other than replacing bad NVRAM chips, all you can do is swap boards from a parts mule.

Reg
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 1255
  • Country: us
Re: Testing RF connectors and cables
« Reply #39 on: September 10, 2019, 12:03:39 am »
"Leo Boner":  That's the name I was looking for!

I have a CSA602A with 1 GHz plugins.  Or....  I could use a sweep generator up to 26.5GHz and SpecAn, or a detector and an XY scope.  I'll have to do some research.  On later, I tried it recently.  It was rather difficult to assert problems because flatness is not quite there.  Since I have to use MaxHold, I can't use normalize function either. 
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #40 on: September 10, 2019, 02:16:08 am »
"Leo Boner":  That's the name I was looking for!

I have a CSA602A with 1 GHz plugins.  Or....  I could use a sweep generator up to 26.5GHz and SpecAn, or a detector and an XY scope.  I'll have to do some research.  On later, I tried it recently.  It was rather difficult to assert problems because flatness is not quite there.  Since I have to use MaxHold, I can't use normalize function either.

It's "Bodnar".  He's a great guy with a fantastic product at a bargain price.

To do TDR from frequency domain data you have to have the phase information. So I can't think of any way to do TDR from an SA & TG.

An impulse at any time delay is constant amplitude in frequency.  The only thing that changes is the exp(i*2*pi*f*t) term) for the shift of the spike in time.  That is a pure phase shift.

Get a nanoVNA and join the party.  It's a lot of fun and there is a lot of stuff that needs to be developed.  The more the merrier.  It's the RF equivalent of the $20 LCR tester.  I predict that 1-2 years from now *everyone* who is serious about RF will have one even if they have a $20K+ VNA.  It fills the "no money for T&M kit" niche and the "need something convenient for field work"  niche.  David Kirkby talked me into getting one and I'm glad he did.  It is really neat.

Have Fun!
Reg
 

Online OwO

  • Super Contributor
  • ***
  • Posts: 1080
  • Country: cn
  • RF Engineer @ OwOComm. Discord: スメグマ#2236
Re: Testing RF connectors and cables
« Reply #41 on: September 10, 2019, 05:16:43 am »
Or wait for V2 to get 3GHz upper frequency range and reduced noise because of no longer using the third harmonic. I just got info that V2 will be priced the same as V1. It will use ADF4350 synthesizers in addition to the si5351 so in theory it can go up to 4.4GHz.
Discord: スメグマ#2236
Email: OwOwOwOwO123@outlook.com
GitHub: gabriel-tenma-white
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #42 on: September 10, 2019, 07:21:18 pm »
There will still be a slew of this version that people have.  If someone gets a V 2, the V 1 can go in the field day kit.

What really matters is software which makes full use of the capabilities of either model, user documentation, etc.

edy555 and hugen79 have provided us with a very nice instrument which almost any radio amateur can afford.  There are a lot of better units, but this one is about $35 delivered from China.  My xaVNA, VNWA and 8753B cost many times more.

Things like python scripts which test RF connectors across all the ham bands or other ranges and compute statistics about uniformity, losses, etc have long term value.  Most of the work is in creating a good presentation of the information.

And if, like me, you have no experience using a VNA, the nanoVNA is an excellent learning tool.  Set yourself a task such as testing feedlines and antennas and figure out how to do it so that it is entirely in the nanoVNA FW.  Treat it as a device with a changeable personality and develop different personalities for different use cases.  But most of all.....

Have Fun!
Reg
 
The following users thanked this post: Neomys Sapiens

Offline FriedLogic

  • Regular Contributor
  • *
  • Posts: 77
  • Country: gb
Re: Testing RF connectors and cables
« Reply #43 on: September 14, 2019, 07:53:18 am »
From my testing, BNC connectors are the most variable.  If the male locking collar is not a tight fit on the body, the connectors will give *very* variable results every time they are used or even jiggled in use.

I think that the shield connection is one of the main contributors to how well they work, and is something that many cheap connectors and adaptors get wrong:

https://www.eevblog.com/forum/metrology/distributing-lab-standards/msg2670798/#msg2670798

I've had less problems with the BNC(f) connectors, but they might still damage whatever is plugged into them if the dimensions are wrong or the machining is bad. I tried some BNC(f) to SMA(m) adaptors on one of Leo Bodnar's BNC pulsers which had about a 32ps rise time, and the cheap one worked not too bad, although it had the slowest rise time.

https://www.eevblog.com/forum/projects/yet-another-fast-edge-pulse-generator/msg1372481/#msg1372481

Unfortunately the scope I was using limited the rise time a bit.
 

Online TheSteve

  • Supporter
  • ****
  • Posts: 3121
  • Country: ca
  • Living the Dream
Re: Testing RF connectors and cables
« Reply #44 on: September 17, 2019, 04:57:56 pm »
I see APC7 to N out of China for 16.99 each shipped now. For anyone who need some at what can only be described as cheap check out ebay.
VE7FM
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #45 on: September 17, 2019, 10:42:01 pm »
Sigh...  I already ordered the 2nd one I needed for my 85046A S parameter set for around $23.

Might be worth getting a couple extra though at that price.    I just put the one I have on my 11801 with an SD-24.  It looks very good.

First scope image is the reflection from an Astrolab SMA-F to N-F and an Anritsu SOL load.  This is a very good quality RG402 SMA cable and adapter.

838371-0

838377-1

Second image is the reflection from a stack of the Astrolab SMA-F to N-F  to a Wiltron N-M to APC-7 to the Chinese APC-7 to N-F to the Anritsu load.  The SD-24 pulser is a 250 mV step.

838383-2

838389-3

The cheap Chinese APC-7 to N-F adapter is not quite as good as the used ebay Wiltron APC-7 to N-M adapter, but it's not bad and very acceptable for the price.  It's certainly good out past 7 GHz.

Have Fun!
Reg

FYI search ebay with "RF Coaxial APC7 to N Female Connector Adapter" to find the $17 ones.  I just bought 2.
« Last Edit: September 18, 2019, 05:25:00 pm by rhb »
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #46 on: September 20, 2019, 12:57:36 am »
Here's some more TDR porn.

I've set up so I'm looking at the rise time of the transmitted step from the SD-24 as it passes through a 30 cm Chinese RG402 SMA-M jumper, Astrolab SMA-F/N-F, Wiltron N-M/APC-7, Chinese APC-7/N-F, N-M/N-M,Astrolab N-F/SMA-F and another 30 cm Chinese RG402.  The N-M/N-M doesn't have a name, just model designation, but is silverplated and in NOS condition.

839650-0

The highlighted trace is the through step which has a 35 ps rise time.  The other trace is the reflection trace.  The large bumps at far left and right are the Astrolab SMA-F/N-F adapters.

839646-1

If I substitute another Wiltron APC-7/M in place of the Chinese APC-&/N-F and  N-M/N-M I also get 35 ps rise time, but much smaller reflections.   al the visible reflections are from the SMA-F/N-F adapters.

839642-2

839638-3

 A cheap ebay SMA-F/SMA-F bulkhead connector has a 33 ps rise time and a single 30 cm piece of RG402 has a 31 ps rise time.  A 3.5 mm short at the input has a 28 ps rise and fall.
« Last Edit: September 20, 2019, 12:59:39 am by rhb »
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 2807
  • Country: us
Re: Testing RF connectors and cables
« Reply #47 on: September 20, 2019, 01:09:02 am »
Here's the SMA-F/SMA-F adapter that came with my nanoVNA.

First hand tightened as recommended to me after I broke the PCB side ground pins of the SMA port on the nanoVNA which had not been soldered on both sides.  I soldered bulkhead mount SMA-F on in place of the originals.

839656-0

Now here's what it looks like after tightening with my Suhner torque wrench.

839660-1

So tell me, which do you think is better?

Have Fun!
Reg
 
The following users thanked this post: xrunner, _Wim_

Offline 0culus

  • Super Contributor
  • ***
  • Posts: 1985
  • Country: us
  • Electronics, RF, and TEA Hobbyist
Re: Testing RF connectors and cables
« Reply #48 on: September 20, 2019, 01:15:55 am »
Definitely needs to be torqued haha. I'm considering getting one, as a guy at work got one and I'm super impressed with what it does for the form factor and price.  :-DD
 

Offline xrunner

  • Super Contributor
  • ***
  • Posts: 4674
  • Country: us
  • hp>Agilent>Keysight>?
Re: Testing RF connectors and cables
« Reply #49 on: September 20, 2019, 01:37:26 am »
Clear results.  ;)

I guess another tool I need to buy.  :clap:
I am a Test Equipment Addict (TEA) - by virtue of this forum signature, I have now faced my addiction
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf