AnsweredAssumed Answered

HP 85047A test set coupler (5086-7489) - disassembly

Question asked by wb0gaz. on Jun 28, 2019
Latest reply on Aug 6, 2019 by wb0gaz.

Like • Show 0 Likes0
Comment • 9

HP 85047A contains HP 5086-7489 test port coupler which grossly fails simple directivity test (terminate with known-good APC-7 50 ohm load, and almost all source drive signal appears on the coupled port save for a notch/dip at mid-span.

As the history of the test set is not known, abuse is suspected.

Visual inspection inside the RF coupler subassembly shows no evidence of physical damage to any component or connection, nor absence of any component.

Question - is there any practical path to troubleshooting the coupler at the component level (with understanding that this would obviously negate any internal factory adjustments that might have been made at time of manufacture)?

Question - can the front APC connector assembly be removed from the coupler subassembly? If so, are there any torque limits I should be aware of (if there is a coaxial connection between the external APC connector and the coupler housing, I don't wish to damage that.)

Question - is the reference coupler (5086-7488) similar to the test port coupler (5086-7489) such that I could replace the coupler subassembly with a 5086-7488 (and ideally, transfer the external APC connector from the failed 5086-7489)?

Thanks very much for any advice!

Dr_joel

Jul 1, 2019 9:03 AM

Correct Answer

So, I designed that coupler (actually a bridge) back in 1986. We'll see how good my memory is 33 years later.

you can test continuity from the bias tee to the output, it should be DC coupled. You can test continuity from the coupled port to ground, it should be 50 ohms.

If the bridge is over powered, then the R1 resistor or the R2 resistor can open up. There is a picture of the bridge on page 82 of my book and a representative schematic on page 81. If you open the bridge you can probe the R1 resistor (from the center conductor to the plate on the mono-block capacitor) and it should be 265 ohm. Occasionally the R3 resistor can fail for over power, and if that happens it is very difficult to repair. The R3 is under the coax outer conductor and sits in a pocket on a second small thin-film circuit under the main thin file. Those are saphire substrates so they are a bit fragile. Sometimes we see a crack in the upper substrate (it floats above ground). If you find that you can bridge it with some conductive epoxy.

The R channel is identical in body, but it might be hard to get the test port connector out. This is where my memory is a little fuzzy: the glass to metal seal at the test port I think is the same between the two bridges. So you can back out the threaded SMA connector and then take off the APC7 from the failed bridge and add them to the R channel body. It's a bit tricky to put in the pin-bead assembly (its easy to have if offset and miss the glass seal center conductor) but I've done it lots of times. But I don't think the R channel bridge has any DC feeds so the bias tee won't work.

See the reply in context

No one else had this question

Outcomes

Dr_joel

Jul 1, 2019 9:03 AM

So, I designed that coupler (actually a bridge) back in 1986. We'll see how good my memory is 33 years later.

you can test continuity from the bias tee to the output, it should be DC coupled. You can test continuity from the coupled port to ground, it should be 50 ohms.

Actions

wb0gaz. @ Dr_joel on Jul 3, 2019 8:18 PM

Close inspection under microscope comparing the bridge between the failed unit and the known-working R coupler. R1 (as described above) appears to have mostly vanished. The bridge resistors (at least the two visible on the substrate leading to opposite sides of the capacitor) appear to be film deposited on the substrate, and R1 (longer and thinner, presumably to get to the 265 ohm design value) on the failed unit is uneven (just an outline around the perimeter of the rectangle making up the resistor - the area within the perimeter is white as if the substrate itself is seen.) The other resistor (not sure which designator, as the capacitor isn't shown on the schematic in your book) looks like it's OK (matches the reference coupler I compared it to.)

There is a small bit of black material (about 3mm long, 1mm wide, at an angle) attached to the test port center conductor. Is it some kind of tuning element? It doesn't get in the way of seeing R1. I don't see any such object in the photo in your book.

In the spirit of "nothing left to lose", I'm wondering if I could clean off the remaining material from R1 (how? scrape it away?) then use conductive exoxy to attach a small (0402?) SMD 265 ohm in it's place (assuming I can find such a part - nothing immediately seen in distribution stock.)

Actions

Dr_joel

@ wb0gaz. on Jul 7, 2019 1:07 PM

Yes, you can do exactly that. The R1 will sometimes disappear due to over power condition. During the development we tested it to distruction, and it withstands +38 dBm for 24 hours at 55C. So some high power definitely went into it. Also check that the capacitor is not shorted on top. If it were shorted, it won't affect the RF performance but could provide a DC path to ground that would allow bias on the port to run through the 265 ohm resistor and burn it up.

The resistor value doesn't matter very much. If you are within 1% of 265 it won't affect the directivity by very much at all. More significant will be the parasitic shunt C from the large physical size. There is likely a little gold tuning ribbon we use to tweak up the directivity at the higher end. If there is black epoxy on the trace it might be a left-over from the beads. they are also tuned to tweak up directivity at 6 GHz and glued in place with a low temperature epoxy. Don't mess with them, it's easy to break the connection between the coax and R3.

Oh, and the blocking cap in the bridge is in the R1 leg.

Actions

wb0gaz. Jul 7, 2019 3:07 PM

All very helpful - thanks for so much detail!

I will check the blocking capacitor for continuity before any next step, understood re a DC path affecting R1.

One question still unsure about - can/should I scrape away the residual material from R1? If so, any recommended tool? I don't have any sense of the substrate's physical properties, and don't want to risk fracturing the substrate.

Why I ask about scraping away R1 --- its residue looks like an ellipse (as if the center area of R1 evaporated, but the outer perimeter survived, and may have collected a bit of material from the center.) Anyway, I haven't yet tried measure DC resistance across the residue, but I'm thinking it should be full open before I try a substitute.

Once I get R1 out of the picture, my next thought is to put an 0402 267 ohm resistor (267 ohm is closest standard value I can find) on end of toothpick (like a tuning stick capacitor) and press it into position (assuming there is room as the black epoxy thing attached to the center conductor is very close to R1's position), and see if the directivity seems reasonable (currently there is no directivity.)

Thanks again!

Dave

Actions

Dr_joel

@ wb0gaz. on Jul 12, 2019 6:55 PM

Don't try to scrape it off. Just measure the resistance across it. Then add an external resistor in parallel to get close to 265. the substrate is pretty fragile and if you break it, you are out of luck.

Actions

wb0gaz. @ Dr_joel on Jul 21, 2019 9:32 AM

Continuing forward with this project - some follow-up information:

1. Measuring R at the coupled port with DC shows 37.5 ohms. Didn't expect this (figured that any other associated damage would have resulted in other open circuit thin film resistors in the bridge.)

2. Measuring R at the test port with DC shows open circuit.

3. Measuring R across R1 with DC (between the test port path and the side of the SMD capacitor leading towards the coupled port) shows open circuit (even though some thin film material is still visible in the path.)

4. Attached an 0402 resistor (not a useful value for R1, just serving as a physical geometry sample) to the end of a toothpick with a very tiny amount of epoxy. Used this to lower the 0402 resistor into position over the residue from R1. There is a good fit (about the correct length in that the contact points of the 0402 package will land on the two traces supporting R1); width is somewhat greater than necessary, but there's at least 50% overlap on the trace leading to the SMD capacitor going to the coupled port. Anyway, this verifies that it would be physically possible to use an 0402 part over R1.

The main puzzler at this point is the unexpected resistance at the coupled port. As the 264 ohm resistors aren't due here for a long time (they weren't stock), there's plenty of time for further exploration.

Actions

wb0gaz. Aug 6, 2019 8:31 AM

First experiment showed some interesting behavior, encouraging further experimentation.

I built a "tuning stick" style device out of a toothpick, a tiny amount of epoxy, and a 267-ohm 0402 resistor (264 ohm resistors are on order but won't be here until October; 265 ohm resistors are unobtainium.)

Pressed the resistor into place over the residue of R1 (which DC measures open) and swept S21 1-1000 MHz to watch coupled port output and compare the sweep without the resistor in place. I used a VNWA and it's Rosenberger cal kit components (SMA open, short, load vna an APC/SMA adapter) with 1M LMR195 test cables and "master" calibration (meaning the loss of the test cables is present in the sweep.) Obviously this only sees a subset of the coupler's passband, but the purpose of today's experiment was just to see if it was worth putting more effort into the project.

In a nutshell, adding the "tuning stick" 267 ohm resistor across R1 residue changed coupled port response in each case in a way that appears closer to normal/typical behavior. There is no image for short, resistor because my tuning stick failed before I could re-run the sweep and save the image, but the result was removal of the pronounced dip around 850 MHz.

DC resistance at the coupled port (which showed 37.5 ohms with only R1 residue in place) was not measured this time before the tuning stick failed, and will be measured once a new tuning stick is built.

Actions

wb0gaz. Aug 6, 2019 10:01 AM

While waiting for a new tuning stick epoxy to cure (after two failed attempts) I re-measured DC resistance cross the coupled port with all other ports open, and this time saw 51.5 ohms. Not sure what lead to the previous (incorrect 37.5 ohm) measurement, but this looks like what Dr. Joel indicated should be seen.

Next steps (once tuning stick cured) will be to re-do the "short with resistor" trace to supply the image missing above, and also measure port DC resistances with 267 ohms temporarily across R1.

Actions

wb0gaz. Aug 6, 2019 1:42 PM

Below: test port connected to short, source port connected to VWNA, coupled port connected to VNWA, 267 ohm resistor pressed into position of R1.

Presence/absence of R1 did not affect DC resistance across coupled port or test port. Verified DC continuity between test port and bias input jumper (per Dr. Joel original recommendation.)

While waiting for 264 ohm thin-film 0402 resistors, will experiment with and practice methods to install the resistor permanently (there is no room to accommodate a soldering iron tip without causing mass destruction, so plan to experiment with conductive epoxies and how to handle the part for installation. Very interested in any advice, comments or suggestions as to how to go about this.

Actions

9 Replies

Dr_joel Jul 1, 2019 9:03 AM
Unmark Correct
Correct Answer
So, I designed that coupler (actually a bridge) back in 1986. We'll see how good my memory is 33 years later.

you can test continuity from the bias tee to the output, it should be DC coupled. You can test continuity from the coupled port to ground, it should be 50 ohms.

If the bridge is over powered, then the R1 resistor or the R2 resistor can open up. There is a picture of the bridge on page 82 of my book and a representative schematic on page 81. If you open the bridge you can probe the R1 resistor (from the center conductor to the plate on the mono-block capacitor) and it should be 265 ohm. Occasionally the R3 resistor can fail for over power, and if that happens it is very difficult to repair. The R3 is under the coax outer conductor and sits in a pocket on a second small thin-film circuit under the main thin file. Those are saphire substrates so they are a bit fragile. Sometimes we see a crack in the upper substrate (it floats above ground). If you find that you can bridge it with some conductive epoxy.

The R channel is identical in body, but it might be hard to get the test port connector out. This is where my memory is a little fuzzy: the glass to metal seal at the test port I think is the same between the two bridges. So you can back out the threaded SMA connector and then take off the APC7 from the failed bridge and add them to the R channel body. It's a bit tricky to put in the pin-bead assembly (its easy to have if offset and miss the glass seal center conductor) but I've done it lots of times. But I don't think the R channel bridge has any DC feeds so the bias tee won't work.
Like • Show 0 Likes0
Actions
- wb0gaz. @ Dr_joel on Jul 3, 2019 8:18 PM
  Mark Correct
  Correct Answer
  Close inspection under microscope comparing the bridge between the failed unit and the known-working R coupler. R1 (as described above) appears to have mostly vanished. The bridge resistors (at least the two visible on the substrate leading to opposite sides of the capacitor) appear to be film deposited on the substrate, and R1 (longer and thinner, presumably to get to the 265 ohm design value) on the failed unit is uneven (just an outline around the perimeter of the rectangle making up the resistor - the area within the perimeter is white as if the substrate itself is seen.) The other resistor (not sure which designator, as the capacitor isn't shown on the schematic in your book) looks like it's OK (matches the reference coupler I compared it to.)
  
  There is a small bit of black material (about 3mm long, 1mm wide, at an angle) attached to the test port center conductor. Is it some kind of tuning element? It doesn't get in the way of seeing R1. I don't see any such object in the photo in your book.
  
  In the spirit of "nothing left to lose", I'm wondering if I could clean off the remaining material from R1 (how? scrape it away?) then use conductive exoxy to attach a small (0402?) SMD 265 ohm in it's place (assuming I can find such a part - nothing immediately seen in distribution stock.)
  Like • Show 0 Likes0
  Actions
  - Dr_joel @ wb0gaz. on Jul 7, 2019 1:07 PM
    Mark Correct
    Correct Answer
    Yes, you can do exactly that. The R1 will sometimes disappear due to over power condition. During the development we tested it to distruction, and it withstands +38 dBm for 24 hours at 55C. So some high power definitely went into it. Also check that the capacitor is not shorted on top. If it were shorted, it won't affect the RF performance but could provide a DC path to ground that would allow bias on the port to run through the 265 ohm resistor and burn it up.
    
    The resistor value doesn't matter very much. If you are within 1% of 265 it won't affect the directivity by very much at all. More significant will be the parasitic shunt C from the large physical size. There is likely a little gold tuning ribbon we use to tweak up the directivity at the higher end. If there is black epoxy on the trace it might be a left-over from the beads. they are also tuned to tweak up directivity at 6 GHz and glued in place with a low temperature epoxy. Don't mess with them, it's easy to break the connection between the coax and R3.
    
    Oh, and the blocking cap in the bridge is in the R1 leg.
    Like • Show 0 Likes0
    Actions
wb0gaz. Jul 7, 2019 3:07 PM
Mark Correct
Correct Answer
All very helpful - thanks for so much detail!

I will check the blocking capacitor for continuity before any next step, understood re a DC path affecting R1.

One question still unsure about - can/should I scrape away the residual material from R1? If so, any recommended tool? I don't have any sense of the substrate's physical properties, and don't want to risk fracturing the substrate.

Why I ask about scraping away R1 --- its residue looks like an ellipse (as if the center area of R1 evaporated, but the outer perimeter survived, and may have collected a bit of material from the center.) Anyway, I haven't yet tried measure DC resistance across the residue, but I'm thinking it should be full open before I try a substitute.

Once I get R1 out of the picture, my next thought is to put an 0402 267 ohm resistor (267 ohm is closest standard value I can find) on end of toothpick (like a tuning stick capacitor) and press it into position (assuming there is room as the black epoxy thing attached to the center conductor is very close to R1's position), and see if the directivity seems reasonable (currently there is no directivity.)

Thanks again!

Dave
Like • Show 0 Likes0
Actions
- Dr_joel @ wb0gaz. on Jul 12, 2019 6:55 PM
  Mark Correct
  Correct Answer
  Don't try to scrape it off. Just measure the resistance across it. Then add an external resistor in parallel to get close to 265. the substrate is pretty fragile and if you break it, you are out of luck.
  Like • Show 0 Likes0
  Actions
  - wb0gaz. @ Dr_joel on Jul 21, 2019 9:32 AM
    Mark Correct
    Correct Answer
    Continuing forward with this project - some follow-up information:
    
    1. Measuring R at the coupled port with DC shows 37.5 ohms. Didn't expect this (figured that any other associated damage would have resulted in other open circuit thin film resistors in the bridge.)
    
    2. Measuring R at the test port with DC shows open circuit.
    
    3. Measuring R across R1 with DC (between the test port path and the side of the SMD capacitor leading towards the coupled port) shows open circuit (even though some thin film material is still visible in the path.)
    
    4. Attached an 0402 resistor (not a useful value for R1, just serving as a physical geometry sample) to the end of a toothpick with a very tiny amount of epoxy. Used this to lower the 0402 resistor into position over the residue from R1. There is a good fit (about the correct length in that the contact points of the 0402 package will land on the two traces supporting R1); width is somewhat greater than necessary, but there's at least 50% overlap on the trace leading to the SMD capacitor going to the coupled port. Anyway, this verifies that it would be physically possible to use an 0402 part over R1.
    
    The main puzzler at this point is the unexpected resistance at the coupled port. As the 264 ohm resistors aren't due here for a long time (they weren't stock), there's plenty of time for further exploration.
    Like • Show 0 Likes0
    Actions
wb0gaz. Aug 6, 2019 8:31 AM
Mark Correct
Correct Answer
First experiment showed some interesting behavior, encouraging further experimentation.

I built a "tuning stick" style device out of a toothpick, a tiny amount of epoxy, and a 267-ohm 0402 resistor (264 ohm resistors are on order but won't be here until October; 265 ohm resistors are unobtainium.)

Pressed the resistor into place over the residue of R1 (which DC measures open) and swept S21 1-1000 MHz to watch coupled port output and compare the sweep without the resistor in place. I used a VNWA and it's Rosenberger cal kit components (SMA open, short, load vna an APC/SMA adapter) with 1M LMR195 test cables and "master" calibration (meaning the loss of the test cables is present in the sweep.) Obviously this only sees a subset of the coupler's passband, but the purpose of today's experiment was just to see if it was worth putting more effort into the project.

In a nutshell, adding the "tuning stick" 267 ohm resistor across R1 residue changed coupled port response in each case in a way that appears closer to normal/typical behavior. There is no image for short, resistor because my tuning stick failed before I could re-run the sweep and save the image, but the result was removal of the pronounced dip around 850 MHz.

DC resistance at the coupled port (which showed 37.5 ohms with only R1 residue in place) was not measured this time before the tuning stick failed, and will be measured once a new tuning stick is built.
Like • Show 0 Likes0
Actions
wb0gaz. Aug 6, 2019 10:01 AM
Mark Correct
Correct Answer
While waiting for a new tuning stick epoxy to cure (after two failed attempts) I re-measured DC resistance cross the coupled port with all other ports open, and this time saw 51.5 ohms. Not sure what lead to the previous (incorrect 37.5 ohm) measurement, but this looks like what Dr. Joel indicated should be seen.

Next steps (once tuning stick cured) will be to re-do the "short with resistor" trace to supply the image missing above, and also measure port DC resistances with 267 ohms temporarily across R1.
Like • Show 0 Likes0
Actions
wb0gaz. Aug 6, 2019 1:42 PM
Mark Correct
Correct Answer
Below: test port connected to short, source port connected to VWNA, coupled port connected to VNWA, 267 ohm resistor pressed into position of R1.

Presence/absence of R1 did not affect DC resistance across coupled port or test port. Verified DC continuity between test port and bias input jumper (per Dr. Joel original recommendation.)

While waiting for 264 ohm thin-film 0402 resistors, will experiment with and practice methods to install the resistor permanently (there is no room to accommodate a soldering iron tip without causing mass destruction, so plan to experiment with conductive epoxies and how to handle the part for installation. Very interested in any advice, comments or suggestions as to how to go about this.
Like • Show 0 Likes0
Actions

HP 85047A test set coupler (5086-7489) - disassembly

Outcomes

Related Content

Recommended Content