Scaler Coax Cable Test

[Rupunzell]

Following up on the $1000 USD spectrum analyzer discussion, on page 10 mention was made  of what other items are needed for care and feeding of a high performance spectrum analyzer. Beyond using a DC block & input attenuator (as required) input cables and adapters have similar importance to O'scope probes for time domain instruments. To illustrate some various coax cable examples were tested using a Wiltron 610D 10Mhz to 18.5Ghz sweep generator, 50 ohm auto tester, 560A Scaler Analyzer and required precision open, shorts, terminations and adapters.

Wiltron 10Mhz to 18Ghz with 40 db directivity Auto tester (reflectance bridge) and their related bits:


Autotester connected to the 10Mhz to 18Ghz sweep generator output with the detector connected to the test port. The DUT goes between the Autotester out and detector. Autotester measures reflected power, detector measures transmitted power.


Wiltron 560A Scaler analyzer, 0db reference set.


System return loss cal after using a precision open and short, difference subtracted and 50 termination at the Auto tester test port. This is termination measures 25db return loss at 18Ghz, not the best but it will do.


APC 3.5 Detector connected to the Auto tester test port, frequency response subtracted resulting in flat response at 1db/division. Note the return loss of this APC 3.5 detector is about 25db. The system is now ready to test cables.


The cables to be tested:

W.L. Gore cable 1.5 meters long made in the early 1980's. These were among the very first flexible coax cable that worked well up to 18.6 Ghz at a cost of  $1500 USD_early 1980's. They were supplied with SMA connectors and adapters were used as required. This one has a 3.5 APC test connector & N adapter as shown in this image. For the test the N connector was not used.


Test results:
10Mhz to 18Ghz measures an insertion loss about 1db and about 15db return loss. This is a very reasonable microwave test cable to be used with a spectrum analyzer.



A current W.L.Gore test cable with APC7 connectors 1.5 meters long.


Close up of the APC7 connectors and one of the APC 3.5 adapters.



Test results:
10Mhz to 18Ghz measures an insertion loss of 0.5db and return loss about 23db. This is a GOOD microwave cable.



1 meter long Huber-Sunher Sucoflex with SMA connectors on each end.


Test results:
Good, from 10Mhz to 18Ghz insertion loss measures 0.7db and return loss about 22db.



Last a 2 meter long RG188 coax with BNC connectors. This required BNC to SMA adapters for testing in this system.


Test results:


Coax does not do well at these frequencies. The insertion loss measures 25db at 18Ghz and it is already 2.5db at 10Mhz. Return loss is OK at 10Mhz measuring about 15db, at 18Ghz, the return loss of 5db makes this coax mostly useless. Note the big dip at 7.5Ghz.
This illustrated why common coax cables and BNC connectors don't do well at higher frequencies. Knowing this is important when adding a cable to the input of a spectrum analyzer as the increasing insertion loss and return loss can significantly affect the frequency response on the spectrum analyzer display.


Bernice

 

[joeqsmith]

Nice job with this post.   

[TimFox]

Nice results.
Note that spell-check will not help with the difference between "scalar" and "scaler".

[PA0PBZ]

Interesting post. The RG188 result and especially the 7.5 GHz dip makes me think that somehow the connectors or adapters are not playing nice here.

[KJDS]

BNC are only spec'd to a few GHz, so it's unlikely that they'd be playing nicely at 7.5GHz

[Electro Fan]

Any recommendations on a reasonably priced N connector (male to male) cable that is (if possible highly) flexible (~5mm diameter or less) and suitable from about DC to 6.5 GHz in approximately 18" and 24" lengths?

I think what I'm looking for is something that is less than 5mm in diameter and has less than 20dB loss per 100 feet at 3GHz, if there is such a thing.  RG142? 

[Rupunzell]

If this is for Ghz instrumentation like a spectrum analyzer, spend the $ and get a purple W.L Gore cable with the proper N connectors in new or known good condition via ebay. This should cost less then $100 USD for 18" to 36". Good microwave cables are like good probes, they are not cheap and the good ones are expensive and worth the $.

Like probes again, poor cables, poor connectors are going to be a source of grief.

Know the most common connector used for microwave stuff is SMA or APC3.5 with N being mostly next most common. APC7 is found in microwave instrumentation often.

Flexi microwave cables are not easy to make as keeping stable geometry is not easy as flexibility is increased.



Bernice

Any recommendations on a reasonably priced N connector (male to male) cable that is (if possible highly) flexible (~5mm diameter or less) and suitable from about DC to 6.5 GHz in approximately 18" and 24" lengths?

I think what I'm looking for is something that is less than 5mm in diameter and has less than 20dB loss per 100 feet at 3GHz, if there is such a thing.  RG142?

[Electro Fan]

If this is for Ghz instrumentation like a spectrum analyzer, spend the $ and get a purple W.L Gore cable with the proper N connectors in new or known good condition via ebay. This should cost less then $100 USD for 18" to 36". Good microwave cables are like good probes, they are not cheap and the good ones are expensive and worth the $.

Like probes again, poor cables, poor connectors are going to be a source of grief.

Know the most common connector used for microwave stuff is SMA or APC3.5 with N being mostly next most common. APC7 is found in microwave instrumentation often.

Flexi microwave cables are not easy to make as keeping stable geometry is not easy as flexibility is increased.



Bernice

Any recommendations on a reasonably priced N connector (male to male) cable that is (if possible highly) flexible (~5mm diameter or less) and suitable from about DC to 6.5 GHz in approximately 18" and 24" lengths?

I think what I'm looking for is something that is less than 5mm in diameter and has less than 20dB loss per 100 feet at 3GHz, if there is such a thing.  RG142?

Ok, thanks for the recommendation on Gore.

If SMA is even more popular than N - but I happen to have N connectors on my spectrum analyzer and signal generator - what would you think about as a strategy - to put a good N to SMA adapter on the SA and Sig Gen - and then use a Gore SMA to SMA cable.  This would add the extra loss of the two adapters but it would protect the threads on the test equipment and allow standardizing on SMA to SMA cables (for example on another sig gen with BNC connectors I could use a BNC to SMA adapter).

(If Gore is your preferred cable, who makes your preferred connector adapters?)

Thx

[Rupunzell]

This is basically the purple Gore cable. It is SMA on both ends, to get the N connector, Gore originally provided an SMA to N adapter.

A return loss of 20 db is often good enough for this kind of RF work as long as the attenuation with rising frequency is linear. Where this does not work at all is when adapters are used with a network analyzer or good scaler analyzer. This is where high quality adapters, every connection, every length of transmission line matters.

As for what brands for adapters, the good APC7 (limit to 18 Ghz), APC-3.5, K and others made by Anritsu/WIltron, Maury Microwave, hp-Agilent-Keysight, Amphenol (not always that good) are often of good quality. Good adapters often have 40 db or better return loss for their entire rated frequency range.

GR connectors can work surprisingly well too including the GR900 which evolved to become APC7 (circa late 1960's).
Omni/Spectra made good connectors at one point in history.

For BNC's Trompeter Electronics makes good BNC connectors, as did Selectro, Kings and Amphenol.  Most all of the made in China BNC's are gross in so many ways. Factoid, BNC-TNC-N share much the same internal mechanical dimensions, the BNC does not work well above 4 Ghz due to mechanical stability where the TNC and N does due to the threaded connection that provides much improved mechanical stability over the bayonet quick connect.

One of the most common RF connectors is the SMA.

Here is an on-line guide.

http://www.microwaves101.com/encyclopedias/microwave-connectors

*I'll do a post on 50 ohm  terminations in the near future as an info-post. Terminations are not the same as a load.


Bernice

Ok, thanks for the recommendation on Gore.

If SMA is even more popular than N - but I happen to have N connectors on my spectrum analyzer and signal generator - what would you think about as a strategy - to put a good N to SMA adapter on the SA and Sig Gen - and then use a Gore SMA to SMA cable.  This would add the extra loss of the two adapters but it would protect the threads on the test equipment and allow standardizing on SMA to SMA cables (for example on another sig gen with BNC connectors I could use a BNC to SMA adapter).

(If Gore is your preferred cable, who makes your preferred connector adapters?)

Thx

[G0HZU]

Interesting thread
However, some of the plot data looks a bit odd to me. I wasn't expecting to see just 0.5dB loss at 18GHz from a 1.5m cable.  I was expecting just over 1dB for that Gore cable?

Also, 2.5dB loss at 10MHz for 2m of RG188 seems very high. I've never used RG188 before but I can't see how it can be that lossy?

The other (obvious) thing to be wary of if you buy used cables on ebay is that the cable may no longer meet its specifications. They wear out and also they get abused in the form of kinking or twisting or stretching.

I have several microwave cables here that look fine but test out poorly. They are worn out ex works rejects and were originally made by Suhner or Mini Circuits.

When the cables begin to fail, the return loss degrades and also the cable loses its phase stability when flexed slightly. At work we throw cables out like this quite regularly.

eg I have a poorly MiniCircuits uW SMA Test Cable here: SMA-Male/SMA-Male CBL-1.5FT-SMSM+

http://194.75.38.69/pages/s-params/CBL-1.5FT-SMSM+_GRAPHS.pdf

Mine is a wornout or damaged reject and does not produce the same return loss performance to 3GHz. But it still looks and feels perfect. I'd post up the plots but it wouldn't be fair on MiniCircuits as these cables normally test out really well.

I also have some rejected Suhner 18GHz cables here that are even worse with just 18dB return loss at 3GHz. But also I have one (identical) healthy Suhner cable that has 32dB return loss or better up to 3GHz.

So my advice to everyone is to be very careful when buying used cables like this. However, you can get some good deals if you shop wisely. I bought some nice Gore cables from KJDS a while back and these give fabulous performance and are very stable in terms of phase when moved.

[Rupunzell]

The surprise was that purple W.L Gore cable. To see the insertion loss being that low really surprised me. It does have APC7 connectors on each end, new in the packaged when it arrived via ebay.

RG188 is basically a PTFE version of RG174, according to the specs, they should be nearly identical. The most likely explanation for the severe loss (2.5db) at 10Mhz is a bad cable/connector.

These results illustrate the problems cable and connectors can have with increasing frequency and other problems due to connectors, cables wearing out from flexing, bad crimps/termination and such.

The Mini Circuits cable is not too bad at all, Mini Circuits is know for their cost-vs-performnace trade offs for quite a long time. They have come a long ways from their early years as a microwave component supplier. Then again, many uW companies come and go over the years.


Bernice

Interesting thread
However, some of the plot data looks a bit odd to me. I wasn't expecting to see just 0.5dB loss at 18GHz from a 1.5m cable.  I was expecting just over 1dB for that Gore cable?

Also, 2.5dB loss at 10MHz for 2m of RG188 seems very high. I've never used RG188 before but I can't see how it can be that lossy?

The other (obvious) thing to be wary of if you buy used cables on ebay is that the cable may no longer meet its specifications. They wear out and also they get abused in the form of kinking or twisting or stretching.

I have several microwave cables here that look fine but test out poorly. They are worn out ex works rejects and were originally made by Suhner or Mini Circuits.

When the cables begin to fail, the return loss degrades and also the cable loses its phase stability when flexed slightly. At work we throw cables out like this quite regularly.

eg I have a poorly MiniCircuits uW SMA Test Cable here: SMA-Male/SMA-Male CBL-1.5FT-SMSM+

http://194.75.38.69/pages/s-params/CBL-1.5FT-SMSM+_GRAPHS.pdf

Mine is a wornout or damaged reject and does not produce the same return loss performance to 3GHz. But it still looks and feels perfect. I'd post up the plots but it wouldn't be fair on MiniCircuits as these cables normally test out really well.

I also have some rejected Suhner 18GHz cables here that are even worse with just 18dB return loss at 3GHz. But also I have one (identical) healthy Suhner cable that has 32dB return loss or better up to 3GHz.

So my advice to everyone is to be very careful when buying used cables like this. However, you can get some good deals if you shop wisely. I bought some nice Gore cables from KJDS a while back and these give fabulous performance and are very stable in terms of phase when moved.

[G0HZU]

The Mini Circuits CBL cables are normally excellent performers The only reason I have this one here at home is because it has been abused in some way which has resulted in it failing a return loss test at work. I saved it from going in the bin.

http://194.75.38.69/products/test_cbl.shtml

http://194.75.38.69/pdfs/CBL-1.5FT-SMSM+.pdf

These MiniCircuits CBL cables are probably the best SMA cables we have at at work in terms of performance vs reliability. We have them in loads of different lengths and they look and feel like they have been designed to give long life. By contrast we have loads of Suhner cables from the ST18 range like these:

http://www.farnell.com/datasheets/63779.pdf

Also the 'black' braided SucoTest Suhner cables rated to 13GHz.

The Suhner cables give very good performance when new but they tend to degrade over time. This is mainly due to the hard life they lead in a busy RF lab. They don't see a lot of sympathy and get treated the same as basic test leads. They usually keep working but the return loss degrades over time/use. The black Sucotest cables are known for losing phase stability over time and eventually become quite erratic when flexed. I think we only have a couple of these left at work now and they aren't popular...

The MiniCircuits cables seem a lot more robust with much chunkier strain relief boots and I'm surprised to see this one fail the return loss test. It's still a very useful cable (to me at least) so I won't be throwing it out any time soon.

Seen in the picture below is the MiniCircuits cable alongside one of my Gore cables. There shouldn't be that much difference in performance (up to 10GHz) if they were both new and I try and keep the Gore cables for critical testing only. I don't want to see the Gore cables lose their fabulous phase stability if they begin to wear out.

Realistically, if I wanted to buy some decent (new) cables for use at home I'd be looking at the CBL range from MiniCircuits. They aren't exactly cheap but they are still cheaper than the opposition and seem to last longer if subjected to lots of use.

I'd post up performance plots of this one vs the Gore cable but it isn't fair on the MiniCircuits CBL cable as this one is a  damaged reject. However, it still gives good but not excellent return loss and the insertion loss is still very low up to 8GHz. I can't reliably measure the loss above 8GHz here at home...

[thewyliestcoyote]

Awesome work!
and great post!

It makes me real happy to use a vector network analyser to do these kinds of measurements.