Test results for the e-bay rf bridge in an enclosure

[dazz1]

Hi
I brought one of those reverse RF bridges, directional couplers.
I wondered how it would go in a metal enclosure.  I am also learning to use by Siglent SSA30xx.

Attached are the results of enclosing the coupler within  a standard Hammond die-cast enclosure.  These are far from ideal RF enclosures but I did expect some improvement over immunity to external RF sources.

The test setup was the output of the TG going to a 10dB attenuater.  The bridge DUT and Reference ports were terminated with 50ohms.  The output went to the SA input.

So I didn't do very well on the graphics.  The results show performance degraded below 1.45GHz and improved above.    The graphs are two superimposed.  I haven't yet figured out the multi-trace feature of the SSA.

I am not sure what the spike is at 945MHz.  It coincides with the scientific/medical band.
The peak at 1.45GHz and 2.9GHZ looks like resonance within the bridge.    These peaks change shape depending on whether the lid is off or on.
I changed the length of the cables to see if that made any significant difference.  It didn't.

I then added some ferrites within the enclosure to see if they would damp the resonance.  They did.  The peaks completely disappeared. Calculations showed the cavity resonance should have been above 3GHz but I suspect the resonance was circulating around the top and bottom of the PCB.   

With the exception of the spikes, the rf bridge has better than -43dB isolation up to 3.22GHz.   Over a fairly large chunk of the bandwidth it has better than 45dB isolation.  That is remarkably good for a low budget ebay buy.

The last attachment shows the final curve.  The resonance is gone as a result of adding ferrites within the enclosure.

I would expect further improvements by using a proper RF enclosure. 



 

[tautech]

The test setup was the output of the TG going to a 10dB attenuater. 

Why ? 
The TG level can be set to the amplitude you need.

[dazz1]

Hi
The 10dB attenuator was included to isolate the TG output from the coupler.  The aim was to prevent resonance down the cable resulting in any mismatch between the TG output and the coupler input.

For similar reasons I initially added a 10dB attenuator to the output of the coupler.  This proved to be unnecessary.   

[tautech]

Hi
The 10dB attenuator was included to isolate the TG output from the coupler.  The aim was to prevent resonance down the cable resulting in any mismatch between the TG output and the coupler input.

For similar reasons I initially added a 10dB attenuator to the output of the coupler.  This proved to be unnecessary.   

Is the coupler not 50 ?

Normally you'd manage all this without any external attenuation and just use the SSA UI to adjust signal strength/attenuation for both in and TG.

[dazz1]

Hi
Here is a photo with the ferrites sitting in place. 
The conventional approach to damping cavity resonance is to apply ferrite impregnated material (rubber, plastic etc) to the interior surface of the enclosure.
I didn't have any so tried these ferrites.  They are very effective so I will now glue them in place.

I am going to get a laser printed metal label to stick onto the enclosure.  This will include a small version of the isolation curve. Most importantly, it will make the coupler look like an expensive professional grade version.

Putting one of these bare-bone couplers in a proper RF sealed enclosure should eliminate spikes from external RF sources. 

[hendorog]

Nice one.

What does the directivity look like - i.e. the difference between your trace and a trace with the DUT termination removed?
Also would be good to know what the difference is between the DUT port open and the DUT port shorted?

[dazz1]

Hi
The 10dB attenuator was included to isolate the TG output from the coupler.  The aim was to prevent resonance down the cable resulting in any mismatch between the TG output and the coupler input.

For similar reasons I initially added a 10dB attenuator to the output of the coupler.  This proved to be unnecessary.   

Is the coupler not 50 ?

Normally you'd manage all this without any external attenuation and just use the SSA UI to adjust signal strength/attenuation for both in and TG.

In normal use I wouldn't bother with the attenuator.  This test setup was designed to isolate the coupler from external influences such as the TG output impedance, cable resonance etc.

[hendorog]

Hi
The 10dB attenuator was included to isolate the TG output from the coupler.  The aim was to prevent resonance down the cable resulting in any mismatch between the TG output and the coupler input.

For similar reasons I initially added a 10dB attenuator to the output of the coupler.  This proved to be unnecessary.   

Is the coupler not 50 ?

Normally you'd manage all this without any external attenuation and just use the SSA UI to adjust signal strength/attenuation for both in and TG.

It's just insurance in case the TG output isn't perfectly 50 ohms, and in case that mismatch affects the measurement out of the bridge.
Otherwise he is not measuring the bridge on it's own, and instead is measuring the bridge plus any effect from the TG + cable <-> bridge mismatch.

Eg, if the TG output power is set low then the attenuators inside the TG will switch in and improve the match a bit. If it's set higher then the match may get worse as there a no attenuators in line any more. So it would be easy to get different measurements at different settings without that external attenuator - assuming the bridge is sensitive to mismatch on the input.

[dazz1]

Nice one.

What does the directivity look like - i.e. the difference between your trace and a trace with the DUT termination removed?
Also would be good to know what the difference is between the DUT port open and the DUT port shorted?

I don't have a shorting adapter but I do have U.FL adapter.  This has a physically small output to minimise fringing effects so it makes a good open circuit port. 
I have just run the test.  Attached is the plot.

I am going to see if I can cobble a shorting adapter.

[dazz1]

OK so I have made up a shorting adapter.  It is not perfect and it requires another inline adapter in between the short and the coupler.    It should be good enough to give a reasonable measurement.  It gives results no more than 3db from the open circuit plot.  Some of this is likely to be the due to the imperfect shorting adapter.

[dazz1]

In conclusion

• Given the rock bottom price of these things, the performance is good.  I know there are better couplers out there but I'm not going to spend the $ to get them.
• Fitting these couplers in an enclosure does affect performance.  The performance is enhanced/degraded at different frequency ranges.
• Cavity resonance matters.  The conventional solution would be to lay ferrite material on the lid or base of the enclosure to impede surface currents in the metal on at least one major surface.  In my case some bulk ferrites did the job.
• Hammond diecast enclosures are crap RF enclosures that allow stray RF to leak into the enclosure.  I knew this before I started, but I didn't want to invest in making a full blown RF enclosure only to find that the coupler was a dud. 

The external interference leaking into the enclosure isn't sufficiently important to justify building another coupler.  The Hammond enclosure is good enough.  It is unlikely that a suitable RF enclosure is available off-the-shelf at any sort of reasonable price.  Making a decent one requires a milling machine, something most people won't have access to. 

[OldDogSleeping]

you can buy these already in a milled enclosure;

https://www.ebay.co.uk/itm/RF-bridge-1-3000-MHz-VNA-Return-Loss-SWR-reflection-bridge-antenna-cased/332640254339?hash=item4d72e77983:g:1mAAAOSwrptbKT2w

I'm a bit of a RF newbie and I don't really understand what your test shows.  I thought the important measurement was directivity, and response with an open DUT port needed to be subtracted from the response with a 50ohm load to show this.

When I try this with my unit it's good under 500Mhz ( >30db), directivity rapidly reduces to less that 15db at 1.5ghz. So I didn;t consider it much use above 500Mhz.

[rhb]

You don't need a mill.

Make a wooden model of the bridge which is slightly over size.  Place a thin sheet of copper on it and hammer the copper down against the wooden form.  Just to be safe you might want to put a pressure adhesive plastic sheet on the inside faces. Then solder the sheets to the SMA ground connections.  Copper foil on an insulating plastic sheet should work almost as well.  Or apply a conformal coating and then press adhesive copper foil over that.   Another option would be to vacuum form a clear plastic take out container to shape and apply  copper foil to that.

Here's a list of dielectric constants:

https://www.kabusa.com/Dilectric-Constants.pdf

The new 60dBm.com bridges in the milled enclosures will still have significant resonances.

[dazz1]

you can buy these already in a milled enclosure;

https://www.ebay.co.uk/itm/RF-bridge-1-3000-MHz-VNA-Return-Loss-SWR-reflection-bridge-antenna-cased/332640254339?hash=item4d72e77983:g:1mAAAOSwrptbKT2w
...

I hadn't seen this version before.  The bridge looks physically smaller with fewer ferrites than the version I have.

[hendorog]

I'm thinking about buying a few of those boards and experimenting with the ferrites. AFAIK the ferrites are all exactly the same.

I reckon substituting in some higher frequency ones closer to the input end would improve it quite a bit.

[dazz1]

Hi
I did a directivity test today.
I did this by normalising the Spec An with the coupler in-circuit and with an open DUT port.

The output of the TG went to a 10dB attenuator.  This is directly connected to the coupler input.  The reference is a 50ohm termination.  The DUT is a U.FL adapter (open circuit load).
The coupler output is connected to the Spec An input.

I normalised the Spec An.

I then replaced the DUT with a 50ohm termination.

Now the Spec An shows the difference between a 50ohm DUT load with near zero reflection and a open circuit DUT load with near 100% reflection.  This shows the couplers ability to discriminate between a matched and unmatched load.

This test was done with the coupler enclosed in the Hammond case and with ferrites inside to suppress cavity resonance.  Your results may vary.
The plot show no indication of any resonance or external RF interference.  There are no anomalous spikes or dips.  Within the limits of sensitivity, the measurements can be regarded as reliable.

The plot shows quite good sensitivity (for a low budget device) at the lower frequency end.  At 350MHz, the sensitivity is -30dB.     The -20dB point is at 1.77GHz which roughly defines the limit of usefulness.  You definitely wouldn't want to be doing measurements at 3GHz where the sensitivity falls to just -6db.

This e-bay coupler is no substitute for a certified lab grade device but $/dB  is still good.  It is a usable piece of kit.

I set up a log period antenna with a bandwidth of 400MHz to 1000MHz.  I left the 10dB attenuator on the coupler input for consistency.  The result is the second attachment.
The claimed bandwidth is slightly wider than the measured bandwidth.  The plot looks believable up to 3.2GHz but study of the previous tests will show that above 2GHz, the results are unreliable.

[dazz1]

Hi
For comparison, attached is the directivity graph for a Mini-circuits BDCG-15-272 coupler.  It does have better performance than the ebay bridge but it is also more expensive, especially if you add the cost of the PCB, connectors and enclosure needed to convert it from a component to a useful instrument.

[hendorog]

Check this thread out, not as good as yours at low frequencies, but it goes have reasonable performance all the way to 6GHz.

https://www.eevblog.com/forum/rf-microwave/open-hardware-microwave-vector-network-analyzer/msg1374172/#msg1374172

I think a combo of the ferrites in that bridge, and the ones in yours, and the marchand balun design might result in the best of both worlds.

[dazz1]

Check this thread out, not as good as yours at low frequencies, but it goes have reasonable performance all the way to 6GHz.

https://www.eevblog.com/forum/rf-microwave/open-hardware-microwave-vector-network-analyzer/msg1374172/#msg1374172

I think a combo of the ferrites in that bridge, and the ones in yours, and the marchand balun design might result in the best of both worlds.

Building your own VNA is an impressive goal.  It looks like he is doing a good job.  It might help to explain that my goal is far more modest.  I am aiming to build, test and calibrate biconical antennas for pre-compliance testing.    Calibrated EMC antennas are hideously expensive so I decided to build my own.  I brought the e-bay coupler for that purpose alone so the working frequency range is more than I need.  The cost of full compliance testing here is so expensive, I have set up my own pre-compliance test lab (including a Siglent SSA) for much less than the cost of a single compliance test. 

In my case, I don't need the best of any world.  I just need something good enough for the purpose.  This thread is actually a by-product of me getting to know the spec an.  I used the coupler as a learning exercise and along the way I got measured performance data for the coupler that I have shared here.  My old Professor used to say "if it isn't measured, it isn't true".  Having measured the performance of the coupler, I now know it is good enough for my purpose.  I also know its limitations.  In addition I got to know the Siglent a little better.

[hendorog]

Check this thread out, not as good as yours at low frequencies, but it goes have reasonable performance all the way to 6GHz.

https://www.eevblog.com/forum/rf-microwave/open-hardware-microwave-vector-network-analyzer/msg1374172/#msg1374172

I think a combo of the ferrites in that bridge, and the ones in yours, and the marchand balun design might result in the best of both worlds.

Building your own VNA is an impressive goal.  It looks like he is doing a good job.  It might help to explain that my goal is far more modest.  I am aiming to build, test and calibrate biconical antennas for pre-compliance testing.    Calibrated EMC antennas are hideously expensive so I decided to build my own.  I brought the e-bay coupler for that purpose alone so the working frequency range is more than I need.  The cost of full compliance testing here is so expensive, I have set up my own pre-compliance test lab (including a Siglent SSA) for much less than the cost of a single compliance test. 

In my case, I don't need the best of any world.  I just need something good enough for the purpose.  This thread is actually a by-product of me getting to know the spec an.  I used the coupler as a learning exercise and along the way I got measured performance data for the coupler that I have shared here.  My old Professor used to say "if it isn't measured, it isn't true".  Having measured the performance of the coupler, I now know it is good enough for my purpose.  I also know its limitations.  In addition I got to know the Siglent a little better.

Of course and that is all good. But bear in mind that AFAIK yours are the first measurements of this bridge. The information that you have provided might help someone else reading this come up with something better. I think that is pretty cool.

[dazz1]

Hi
Now that the coupler is fully housed with the glue dried holding ferrites, I thought I would do a final test by repeating the first test.
Both the Reference and DUT ports are fitted with 50ohm loads.
Between the TG output and the coupler input is a 10dB attenuator.

The only significant difference between the first test and this one is the TG level is down -10dBm.  That shouldn't make a difference on a purely passive instrument.

The plot is significantly different to the first results.  So different that I repeated this latest test to make sure it was right.  Notably all the RF interference has gone.   There is a hint of cavity resonance at the marker but not enough to cause any concern.    The plot starts off at the implausibly good value of -65db.  I was deeply suspicious of the reliability of the measurements.  I repeated the test and got the same results.

I repeated the test again with a TG level 0dBm to replicate the first tests.   The results shown in the 2nd attachment are the same.
I suspect there must have been a fault with the first test setup.  Most likely a loose connector.  I have re-checked the latest test setup and it is faultless.  I normalised before each test.

What should happen now is to go back and replicate the original tests on a bare unenclosed coupler to see what effect the enclosure has. The coupler was installed in a way that would make it difficult to remove.  I will leave it for someone else to repeat the test.  I have intentionally provided enough information on all the test setups to allow others to validate the results. 

The fact that the rf interference has gone is evidence that the enclosure is good enough to block it.  Conversely the rf interference shown in plot of the bare coupler is evidence that it is susceptible  to rfi.  No surprises there.