开云体育

My adventure fixing up my workshop's RF gear brings me to the last repair, the heaviest unit (94 pounds, and I'm older now than when I put it in position!)

The HP 8673D synthesizer uses a 3.5mm male panel connector (used on some other HP units supporting K-band). On this 8673D, at some point someone glued (?) a SMA female-female adapter into the 3.5mm male connector, and subsequently the adapter was sheared off, leaving half it's corpse in the 3.5mm male connector. I tried holding a spare SMA jack in place with a copper pipe cap, however, the RF path is impaired enough with this scheme that return losses badly confuse the leveling loop (otherwise the unit appears to function without errors.)

I have a replacement connector (08673-60040 is the HP part number) in hand.
I have service manual in hand (likewise quite large/heavy).

The procedure for replacing the (A4J1) connector isn't specifically called out, however, it appears the normal course of action is to remove the front panel assembly, gaining access to the back side of the connector. Removing the front panel, like most significant service procedures described in the service manual, begins by "separate the two units". I've attached photo of the relevant start of service sheet A pertaining to disassembly for front panel removal.

My idea is this - the front panel (A4) may (hopefully) only need to be moved slightly forwards (enough to get access to the back side of the RF output connector.) It appears that removal of the four internal screws that hold the front panel casting to the top/bottom internal structure (runs the length of the left and right sides) would allow the front panel to be moved forward away from the body of the unit (I don't know if there is a screw along the bottom front of the top section rendering this idea moot).

Although there are a number of connections (referring to service manual service sheet A), only one appears to be the semi-rigid output coax, and it can be removed from it's top connection (attenuator) easily (the cable cable would then come away with the front panel pulled forward.)

So, looking for any comments, suggestions, observations (and of course can supply any additional images that might be helpful.)

Dave

Hi Dave,

I own an 8673D and I'm very satisfied with this generator; it performs quite
well.
I was able to get it back in exchange for another working instrument,
however, this 8673D was considered DEAD.
After several repairs and... several weeks later, I was facing the same
problem as you: separating the two stages to gain access to the 3.5mm
connector I wanted to replace.

Unfortunately, I wasn't able to separate them.
Let's just say I didn't try too hard, but I still removed or freed all the
cables and semi-rigid cables connecting them, and I couldn't find a way to
separate the two front panels.
I thought it was the central screws that were blocking everything.
But according to the procedure you attached, it seems the central screws
weren't involved in the disassembly.
So, I don't know what was preventing the separation.
All the metal support clips were removed.

In my case, it was the external part of the 3.5mm connector that was
slightly damaged.
I had corrected the connector with an exacto, but it was by installing a
3.5mm connector extension that I resolved the problem.
This 3.5mm connector came from a defective (non-repairable) module that
fitted an 83480A.
Since then, it has remained stable and meets specifications.

If you ever find the correct procedure, I would be interested to know it.

Yves

Thank you, Yves - I appreciate your information and agree that the unit is very nice tool (although very heavy!). Fortunately, this unit appears to still work internally, as the problem of leveling/output seems isolated to the periphery of the system, not the internals.

I do not directly see the need to separate the assemblies, provided that I can physically move the upper unit front panel forward slightly to gain access (working inside the gap between the bottom of the relocated upper unit, and the fixed lower unit, it seems access to the mounting nut for the connector should be accessible.

Do you remember what path you followed to reach the connector nut behind the upper unit front panel?

Thank you again!

(Yves - I now see from the picture - you were able restore operation by adding component to the existing connector? In my case, the internals of the connector are damaged so my attempt to add component was not successful - very poor RF path resulted. Sorry for the mis-understanding.)

Decided "go for broke" --- gently lowered the unit onto the table in front of the shelving where it lives (and onto a piece of carpet material), turned it on it's side (3.5mm front panel connector towards bottom - a better choice would have been 3.5mm front panel connector towards top), and continued fastener removal (in many small labeled bags!)

Got to the point where the two halves are split, but not physically separated.

There are numerous wires (towards the front only) connecting the two halves; only two are UT141-type semi-rigid line (RF to and from the lower unit, I believe.) There appears to be enough slack in the wires (except for the two UT141-type semi-rigid lines) to allow about 100mm of separation without unplugging or unsoldering wires (which represents most of the steps involved in separating the two units by service manual.)

Getting to this stage required removing fasteners along the left and right sides that appeared likely to hold the front panel of the upper section in place, as well as fasteners along the sides that join the two halves. The (diagonal, right side) upper fastener at the front was partly obstructed by a U shaped aluminum section (about 5mm radius). I used small diagonal cutters to chew away some of that until I could remove the fastener. The material turns out to be part of the internal cover that isolates the two units, except for an open area near the center/front. I do not think my small cutting will cause problem.

At this point, the front panel of the upper section would still not separate from the body. Along the back of the unit (between the two units) was a span of thin metal (like top and bottom covers, same type of finish, decorative style) with two screws; I removed that piece (exposing the gap between the two units at the rear), and with this, the two units began to move independently, so I focused back to the front panel separation.

On the next step, I removed the decorative trim from the top of the upper front panel (thin screwdriver to pry away the plastic trim piece); under there were three more fasteners that I removed. Once these were removed, the front panel showed some more possibility of independent movement but still was stuck in position.

At this point (with the rear mid-point cover removed and the two units beginning to move independently), I began separating the two units, but discovered that the two semi-rigid (UT141) hardlines connecting the two halves was preventing much movement. I removed the 3.5mm male connectors from the upper unit termination (one of the two switches alongside the output attenuator). This required temporarily removing one end of two of the local (terminating both ends between the two switches) UT141 hardlines so that I could get better access to one of the two top-to-bottom hardline connectors. Once both of those connectors were removed from their switches (on the top side; no attention to the bottom side of the two cables), I was able separate further the two units (requiring careful steps to navigate the two now-free male hardline connectors.) Although I took some pictures, I will consult the service manual during re-assembly because it is possible to reverse the two top-to-bottom UT141 hardlines.

Once the two halves were separated enough to allow access to the bottom front of the upper section, I found no decorative trim, but three more fasteners of the same type, which I removed. I also saw several threaded (like machine screw end) visible along the same bottom front area, but fortunately these do not impede separation.

At this point the front panel assembly can move freely (still constrained by numerous wires) from the upper unit body, maybe 50mm clearance possible for access to the failed connector through what would be the bottom and side of the unit nearest the connector. The attachment of the internal connector is a simple hex nut, like used on old style potentiometers.

I believe (tomorrow, hopefully) I will be able replace the 3.5mm male connector from the front panel (remove both ends of the UT141 hardline connecting to the output attenuator.

Two photos are attached, one of the partially separated units, and other close-up of the 3.5mm-type switch connectors that are temporarily removed.

Successfully swapped the front panel connector (it's replacement was a used pull.) The main challenge is that the front panel PC board comes fairly close to the back side of the front panel connector, making it a little difficult to get a decent grip on the rear nut. Turns out a 1/4" drive, 1/2" thin wall hex socket on a 1/4" screwdriver-style socket driver handle was sufficient once the front panel assembly was rotated sufficiently around to accommodate the handle.

Now begins reassembly (reverse of disassembly, as the HP manual says.....) - the only steps that should take any special care are getting the top-to-bottom semirigid lines (2) back into position, and avoiding trapping any of the top/bottom section interconnect wires while bringing the two halves back together.

8673D is making RF and working as expected. No more leveling problems. Fan (after first minute or so) sounds OK.

Remaining problem (which existed before, and I will address in future) is very low output in the 10 MHz - 2 GHz range (which uses the lower section.) This problem has existed for the time I've had the unit, and I haven't wanted to tackle repair (as I also have HP 8660 series generator and nowadays a TinySA can reasonably suffice).

Now that I have experience with the unit's (partial) disassembly (and some tools to help move the literal boat anchor around in the workshop), I'll revisit the 10 MHz - 2 GHz issue at some point and report on that adventure.

Thanks for the bandwidth!

Hi,

Glad to hear you were able to successfully replace the connector. I kept
your notes for disassembly; I'll probably have to replace my connector as
well. Where did you find the replacement?

Regarding your low-frequency output problem (10MHz-2GHz), I've had the same
problem twice: no more loss at 10MHz and less and less loss as I approach
higher frequencies. It's probably the Watkins Johnson 6203-044 amplifier (HP
0955-0215). The internal screws make the ground contact. I tightened the
screws, and everything returned to normal. A few months later, the problem
returned. I then soldered the ground contacts instead of relying solely on
the screws. Since then, the problem has never returned...it's been two years
now.
This amplifier is located in the lower part; simply open the bottom panel,
and you have access to the amplifier.

Here's a picture showing the specifications, the amplifier itself, and the
results after the repair.

Attached are some tests I performed once I had completed all the repairs to
ensure optimal operation.

The 8673D, the excellence of this generator is matched only by its weight :)

Yves

Thank you, Yves, about working on lower section.

I now wish I had made photographs of top-of-bottom, and bottom-of-top areas. There is a panel separating the two sections, with ~8cm x ~15cm opening near front for wires to pass top-to-bottom. I wish HP had used flexible cable for RF between the sections (two cables), this would make separating the sides easier.

Fortunately, when I was re-assembling, the two UT141 section connectors arrived very close to original position (I had not applied stress during dis-assembly), so there was no worry about reversing the two connectors (which go to switches next to the output step attenuator.)

After repair of 8673D, I tested output using my Tektronix 2755AP analyzer (I have had since ~2015, also old machine almost ready for scrap pile) to look at very narrow span (down to 100 Hz/div) for spectral purity. Tektronix 2755AP analyzer is providing error messages (about IF2 frequency control, it is strange "frequency locked loop" under control of firmware, and it is not PLL!) so it is next big machine for repair here. Adjustments are made with many many many small potentiometers, guided by primitive firmware. New adventure!

Some flat head screws inside (which are in many places along both sides of HP 8673D) are M4 x 4mm. I believe HP used many metric fasteners. I ordered some for spares, because of course it is easy to permanently lose specialty fasteners!

On the output connector (HP part number 08673-60040), it is used in other units, including HP 83595A and 83594A plug-ins for HP 8350A/B, it is used in HP 5348A counter/power meter (which is strange animal, like power meter with no option to set correction factors, and frequency counter, in same box, but with zero integration except shared LCD digits (????), and also connector is used in HP 5343A frequency counter, HP 8563E Opt 026 spectrum analyzer (and other/newer models). From the part number, I believe HP 8673 series was first requirement for front panel 3.5mm connector.

I searched for long time to find spare 08673-60040 connector, it is not so pretty, but it is working OK. Now there is one for sale on auction site (by supplier in israel), but cost is very high. Perhaps create search trigger for part number, after some time I think another will appear.

There is other kind of 3.5mm panel connector from HP, 08513-20017. It is I think used in modular plug-in microwave test equipment (HP 7000 mainframe), it is much lower cost connector on used market. I do not know if bulkhead hole diameter is compatible, but it is possibility. The original connector type is more rugged against damage (but as you found and I found, damage is still common.)

So, thank you for dialog about HP 8673D service adventure, and it is nice machine, almost weighs as much as adult person!

Dave