Homebrew Return Loss Bridge

[KD0RC]

I recently bought a Rigol DSA 815 TG, and am very happy with it.  As a non-engineer hobbyist, this is the ultimate price/performance gizmo for me!  The first thing I did was build a Return Loss Bridge from the ARRL site from an article circa 1997.  I built it using leaded parts in a point to point configuration similar to Zach Lau's unit pictured in the article.  I used an on-hand, unknown material ferrite core for the detector transformer.  The darn thing seems to work pretty well for a $5.00 tool...  Here is a plot of it from 0 - 500 MHz with a non-precision 50 ohm resistive load.  I did not set a marker at 2.5 MHz, but that seems to be the lowest practical frequency.  The sweet spot is around 62 MHz.  I seems quite usable to 250 MHz, but pretty marginal by 500 MHz.  For my ham radio purposes, it seems good to go for HF - UHF, even if the ulitmate accuracy is suspect at the higher frequencies.

So...   Am I fooling myself with this thing?  Is it really worse than I am making it out to be?  How does it compare to a Mini-Circuits or other $100 class RLB?

Thanks,
Len

[KD0RC]

Here is the bridge in action with a slightly mis-tuned (4MHz low...) 2 Meter J-Pole (end-fed half wave with a quarter wave matching section). It is showing a VSWR of about 1.1:1, which if true, is excellent for my purposes.

Len

[sacherjj]

I haven't had a chance to build a bridge for VSWR.  It is really cool that I can unlock the radio based stuff for hobby use, not just the EMI pre-compliance stuff that is the real reason work bought this DSA for me. 

I've seen some designs with high frequency rigid line that I want to make, to get a bridge closer to the capabilities of the full frequency range of the DSA.   

[guipoletto]

@KD0RC

Hi! Could you post some pictures??  I'm in the need of a vswr meter, and have read the Arrl paper, but i didn't quite fell for it.

I've been digging for a couple weeks about SWR measurements, and found a lot of designs.
As of now i found the one with a (Bi)Directional coupler the most ellegant of them. The thing is that i could not find any dual output couplers on ebay, and i don't have the equipment to caracterize one if i make it.

On the other hand there are some minicircuits splitters on aliexpress that got me thinkink, there should be possible to cascade them, or not?
http://www.aliexpress.com/item/MINI-ZEDC-15-2-b-1-1000-m-SMA-15-db-broadband-directional-coupler-15-db/1474178485.html

What do you guys think?

[KD0RC]

@guipoletto
Hi, here are some pictures of the RLB.  I used paralleled pairs of 100 ohm .1% resistors (leaded) instead of single 49.9 ohm 1% resistors.  My ferrite core is unknown and slightly larger than the one in the QST article.  I built mine like Zack Lau's - in a piece of 3/4 inch (19 mm) square aluminum tubing.

So this is not a VSWR bridge (but it can be used to calculate VSWR), it is a Return Loss Bridge.  A directional coupler is normally used with a meter to directly measure VSWR, and may be more like what you are looking for.  If you have a spectrum analyzer the RLB is easier to use.  The QST article shows how to do it using a diode detector and a meter, but that is the hard way to do it...

I am not sure what you mean by cascading the Mini-Circuits units.  Normally a bridge is used by itself.

[kg4arn]

So...   Am I fooling myself with this thing?  Is it really worse than I am making it out to be?  How does it compare to a Mini-Circuits or other $100 class RLB?

Pardon these comments if this is something that you already know...

A figure of merit for the return loss bridge is the bridge directivity.  Basically directivity is a measure of how much energy is present at the detector port when the DUT port is terminated with a "perfect" 50 ohm load (for a 50 ohm system).  These two papers by Sam Wetterlin explain it pretty well.  Open and Short load performance are also important.


http://www.wetterlin.org/sam/Reflection/ManualReturnLoss.pdf
http://www.wetterlin.org/sam/Reflection/Bridge_BalunPlusBeads.pdf

The plot shows that for a 30dB bridge, if your measured return loss is 20dB, the actual return loss could be as good as 24dB or as poor as 18dB.

[KJDS]

That graph assumes perfect source match. In practice it never is perfect and at poor return loss levels tends to dominate the uncertainty.

[guipoletto]

I am not sure what you mean by cascading the Mini-Circuits units.  Normally a bridge is used by itself.

I could not find any bidirectional couplers out there. It looks like they are all the same tough, but the ones in the market have the return path terminated inside the case. Hence (almost) every bridge i could find will just give-me forward coupling, and i need reflected(reverse) aswell.

I think i will need to blindly build my own coupler(since i'm not the lucky owner of a DSO or scope).

[kg4arn]

That graph assumes perfect source match. In practice it never is perfect and at poor return loss levels tends to dominate the uncertainty.

Yes. I believe the graph is intended to show the effect of bridge directivity and does not address source mismatch.

[KD0RC]

Pardon these comments if this is something that you already know...

A figure of merit for the return loss bridge is the bridge directivity.  Basically directivity is a measure of how much energy is present at the detector port when the DUT port is terminated with a "perfect" 50 ohm load (for a 50 ohm system).  These two papers by Sam Wetterlin explain it pretty well.  Open and Short load performance are also important.

The plot shows that for a 30dB bridge, if your measured return loss is 20dB, the actual return loss could be as good as 24dB or as poor as 18dB.

KG4ARN, thanks for this!!  I am learning this from scratch, so this is eaxctly the kind of paper that I needed.  I intuitively put a 50 ohm resistive load on the DUT port to see how flat the loss curve would be, but now I have a better feel for why I did it and what it really tells me.  The 815 has a 'Cal Open' function on the VSWR function, so it does not take the short into consideration, only the open.  I will try shorting it later this evening when I get time to compare results with the open condition. 

I don't know how much calibration advantage I get from the Cal Open function, but I did check a 2 meter antenna with an MFJ antenna analyzer and got the same VSWR at the same frequency as reported by the 815.  My analyzer only goes to 170 MHz, so I don't have a way to cross verify the SWR measurements at UHF.  Even if the absolute accuracy is not good up there, the relative indication of resonance is still quite helpful.  I ran it on a couple of UHF antennas and got results in the ballpark that I expected, so (non-scientifically...) I think it is working OK up there.

[krenzo]

I could not find any bidirectional couplers out there. It looks like they are all the same tough, but the ones in the market have the return path terminated inside the case. Hence (almost) every bridge i could find will just give-me forward coupling, and i need reflected(reverse) aswell.

Switch your input and output ports to get reflected power from the coupled port.

[guipoletto]

I could not find any bidirectional couplers out there. It looks like they are all the same tough, but the ones in the market have the return path terminated inside the case. Hence (almost) every bridge i could find will just give-me forward coupling, and i need reflected(reverse) aswell.

Switch your input and output ports to get reflected power from the coupled port.

Aren't those in reality bidirectional internally terminated? it would be too bad if i chained two of those mini-circuit ones facing each other to take both measurements at the same time?

It looks like they can handle 50W at 450mhz..  couldn't find anything in the Datasheet tough..  Getting a pair to play with! 18 bucks a piece seem good.

What specs should i be looking in the diodes?
 

[KJDS]

What specs should i be looking in the diodes?

Use BAT63s, use one for RF detection and the other with an LTC1051 or similar low offset op amp for temperature correction.
https://www.infineon.com/dgdl/bat63series.pdf?folderId=db3a304314dca389011518104e5d0df2&fileId=db3a304314dca38901151817843c0df4

So for the RF input, a series dc blocking cap and then one half of the dual BAT63 shunt to ground, put the cathode to ground as bias current from an op-amp generally flows out of it*

Then just put the other half of the diode pair in the feedback path to give temperature stability, otherwise you'll have something that's a better temperature detector than it is for RF power. I'd then use another LTC1051 to provide a gain of 100 or so to provide a useful output voltage.

Having typed that out, I've just checked and found the TLC2652 is still available, I thought it had gone EOL years ago and that gives better results than the LTC1051. Use the A version if you can get hold of it.