HP 8595E SYTF P/N: 5086-7803 PIN switch problem

[analogRF]

I've got a 8595E which works perfectly fine in the high band but it's about 24-26dB down (almost flat) in the low band.
Long story short, by doing various measurements, I have confirmed that the PIN diode switch inside the SYTF when
switches to the low band is dead (exactly 24-26dB loss over the band) so it does not pass the signal through to the mixer
Everything after the SYTF works fine when I directly inject low band signal. The YIG filter itself is of course fine, too.
It seems to me that the PIN diode in the shunt arm of the switch is shorted perhaps due to excessive input power.
It also overloads the switch voltage that comes from A7 board, instead of -9.8V it becomes -3.6V when that switch is supposed to close. But for the other pole of the switch (high band) it is -9.8V as it should be. If I disconnect the wire, the voltage is back to -9.87V which is correct.

Anyways, since I have never opened a SYTF or any YIG tuned stuff before
I have a couple of questions I was hoping someone on this forum can help me with:

1-what are my chances of repairing the switch and not messing up the YTF?

2-Are the switch and the YTF in two separate compartments? If I attempt to open it up
is there anything that might fall off and mess the whole thing up?

3-is the switch made of wire bonded microcircuit elements? I don't have access to any wire bonding stuff. Only my soldering iron!

I have also posted this on keysight forums and HP/Agilent Group.io
 
thanks

[analogRF]

...Another solution that I am considering is to buy a microwave SPDT switch
and put it between the attenuator and the input of YTF. For the high band the switch will pass the signal to the SYTF and
it does its job perfectly but for the low band the switch will divert the signal directly to the dual mixer bypassing the SYTF
The biggest issues are (1) generating compatible switching and power supply voltages from what is already on the A7 board
for that I will have to put is a small circuit somewhere! and (2) where to mount the switch in the RF compartment.
There are some small spaces here and there and maybe I can squeeze it somewhere...not sure

However, the type of the switch is critical. First I was looking at relay switches for example from Teledyne which have
good specs and relatively cheap but then it struck me that the switching times must be very small compared to the sweep speeds of the SA
Relay switches are rated at 5ms-15ms switching times. That might work OK for most of the time but if it happens that your frequency span
covers the transition between low band and high band and the sweep time is in the range of 10's of ms then it becomes a mess...that's why
the HP designers only used PIN switches in all places in the RF section. Modern analyzers use GaAs FET switches. The switching times for PIN diode switches are around 5-20nsec but of course they can never achieve the insertion loss and isolation specs of relay switches. not even close..

Now the only thing I have been able to find was Mini-Circuits ZYSW-2-50DR and ZYSWA-2-50DR with reasonable specs up to 3GHz and even applicable up to 5GHz but not to 6.5GHz as I need.

Any suggestion about a PIN diode RF switch that I can buy at reasonable price is welcome

[kirill_ka]

I have HP 8561E using the same SYTF. In my case I had low output in the high band. Initially I thought that the switch is not working, but most likely YTF got detuned.
Anyway, I've destroyed the switch during my experiments 

1-what are my chances of repairing the switch and not messing up the YTF?

Besides the switch itself there are two devices that look like protection diodes.
I'd try to find burn marks visually. If the diode is the only part which failed, may try to disconnect it by removing bond wire.

2-Are the switch and the YTF in two separate compartments? If I attempt to open it up
is there anything that might fall off and mess the whole thing up?

I don't think peeling the mesh from the switch side would significantly affect the performance. But also I have no idea how to glue it back in place.

3-is the switch made of wire bonded microcircuit elements? I don't have access to any wire bonding stuff. Only my soldering iron!

Yes, unfortunately. The switch is not a pin diode. Look for HMMC-2006 datasheet (I don't remember if it's the exact part of very similar). It's a GaAs SPDT switch.
The interesting thing is that the switch is connected back to front - switch output is connected to the SYTF input. High band signal is not going through the switch, it's reflected back to the YTF part.

Edit:
"It also overloads the switch voltage that comes from A7 board, instead of -9.8V it becomes -3.6V when that switch is supposed to close. But for the other pole of the switch (high band) it is -9.8V as it should be."
This probably means that the switch itself is dead.

[kirill_ka]

Yesterday I tried too tune the filter. The result was predictable, the sphere snapped off from the rod. I should have used some fixture to control the movements.
But that's only the 3d stage. There are two stages remaining, I must find a convoluted way to destroy them too 
I've already installed a replacement SYTF, so turning attempt is just of curiosity.

[analogRF]

Yesterday I tried too tune the filter. The result was predictable, the sphere snapped off from the rod. I should have used some fixture to control the movements.
But that's only the 3d stage. There are two stages remaining, I must find a convoluted way to destroy them too 
I've already installed a replacement SYTF, so turning attempt is just of curiosity.

thanks for the info. So it seems, as I feared, the chances of fixing the switch is pretty close to zero.
But I thought it was PIN switch because I remember somewhere in the service manual is says so...oh, well, GaAs FET or PIN does not matter now...
I think I will have to go down the path of installing a new extra switch if that's possible but I don't know which switch is a good choice

[kirill_ka]

but I don't know which switch is a good choice

Yes, that's the problem. You can't install a solid state switch in the same configuration as it is in the SYTF.
If you use a relay, you'll probably need an external band switch.

It might be possible to use the other half of the switch if it's alive. But again, it would require wire bonding or soldering a microscopic wire...

[MarkL]

If it helps, here is the inside of a 5086-7803 SYTF from an 8595E.

The high band was dead on this one so I decided to open it up.  You can see the YIG sphere in the lower right which had fallen off its post and was lodged between the two coupling loops.  The YIG sphere later rolled away and could not be found.  Bye, bye, high band.

EDIT: Added close-up.

[analogRF]

If it helps, here is the inside of a 5086-7803 SYTF from an 8595E.

The high band was dead on this one so I decided to open it up.  You can see the YIG sphere in the lower right which had fallen off its post and was lodged between the two coupling loops.  The YIG sphere later rolled away and could not be found.  Bye, bye, high band.

EDIT: Added close-up.

great pictures! So the switch is actually right next to the YTF side by side. Well, thats not what I can repair...at least the filter part works perfectly fine

[madao]

I has a  8595E with  fault in band 1.  massive loss.


I could repair him with soft power on YIG sphere-holder (plastic bloc) and its work again.
But i dont guess, it is again in specification, but self-cal: passed.

best regards
Matt

[hackwell]

Hi

got a 8595E off Ebay recently , showing two issues.

first the keyboard was not responsive at all, the sweep refreshes every minute of so. after some measurement on the YIG , it appeared to be dead.
Installed a replacement from a 8594E and after a factory cal reload the operation was back to normal.

the second issue is the same as described above. Spot on below 2.9GHZ but no signal in the high band. After taking the YTF filter appart guess what?
yeah... misaligned spheres again.

I managed to realign two of them but infortunately the one on the high band side had snapped off it's rod and my attempt to glue it back into position was a complete failure 

My guess is that when the sphere is detuned to the point it touches the loop it snaps off shortly after  I thing the analyzer was used in low band until the YTO failed then was decommissionned.

the rods are made of glass apparently so be very careful when touching them.

I wonder what kind of glue they used  in the first place ?

Some interesting info on wire bonding on this page : https://www.qsl.net/ct1dmk/wbond_ex.html

some shops offer YIG rebuilding as well but the  prices are... hum...

 EDIT : surprisingly enough the 8595E passed all self calibrations and confidence test with a dead YTF. Weird 

[Bicurico]

Questions (for my own satisfaction):

1) Is the YIG sphere the small metallic sphere on the tip of what looks like plastic pens?
2) Are there 3 of them? Why?
3) You guys try to glue the YIG sphere back to the tip?
4) Isn't the YIG highly toxic/radioactive/whatever = meaning: never touch it?
5) If you lost one YIG sphere, what is the resulting danger, if it is lurking unprotected under your table? What if your cat/dog eats it?
6) How exactly does the YIG work? I read about it, but it sounded like black magic...

Regards,
Vitor

[hackwell]

Questions (for my own satisfaction):

1) Is the YIG sphere the small metallic sphere on the tip of what looks like plastic pens?

Yes indeed. the rods seem to be glass BTW

2) Are there 3 of them? Why?

The three cells are used as filters to separate low band up to 2.9GHZ and the high band from 2.9GHZ to 6.5GHZ. The benefit of using YIG filters is the response speed allowing a continuous sweep across all the bandwidth without drop. Much precise and faster than switches.
This is achieved by applying a magnetic field focused on the spheres. this adjust the frequency response accordingly.

3) You guys try to glue the YIG sphere back to the tip?

I did try that to no avail. used bi component epoxy glue but curiously enough it would'nt harden even after a whole night. and as the sphere is approx 0.5 mm diameter it's quite hard to put the right amount of glue at the good place and get the sphere in exact position.
Anyway it is useless to glue it back in place as it is factory tuned by rotating the supporting rod and the chances are slim to get into specs after such a repair. I dunno if someone managed that already ?

4) Isn't the YIG highly toxic/radioactive/whatever = meaning: never touch it?

Nope they are maily gems. Costs a fortune but not toxic at all

5) If you lost one YIG sphere, what is the resulting danger, if it is lurking unprotected under your table? What if your cat/dog eats it?

The main danger is to have to find another working YTF on the bay for less than $800 

6) How exactly does the YIG work? I read about it, but it sounded like black magic...

As I explained before it is a tunable filter. The center frequency is adjusted by a magnetic field. The ytrium is magnetic sensitive and acts as a tunable XTAL. In the YTF case the speres are used as filters the separate the two bands. But in a YIG oscillator it is used as a high frequency tunable oscillator

[MarkL]

Here's a decent intro to YIG tuned filters:

  https://www.microlambdawireless.com/resources/ytfdefinitions2.pdf

[Bicurico]

hackwell and MarkL: Thanks!



The article is great and reinforces my opinion that the YIG is some magical material! How on earth did physics came up with such properties?

Regards,
Vitor

[madao]

 EDIT : surprisingly enough the 8595E passed all self calibrations and confidence test with a dead YTF. Weird 

Uuuh,  it means, my sucess with SYTF is not sure.  I has only one  SHF-Signalgenerator, it is  from 1960s. Very heavy  Rohde Schwarz gears carry up to  four etage..no mood !

[hackwell]

I don't know how the Rhode & Schwartz self calibrate. Used one for years when I worked in a HAM radio repair shop but we had a calibration service to care about that. Don't touch it little boy 
Sure enough all the 85xx HP series self calibrate with the 300 MHz calibration signal. So what happens when the high band fails ?
 
EDIT : the YTF calibration passes as well. even with -16 dB loss.

[hackwell]

I did some more testing on my 8595E . injected a 4GHz -16 dB  signal into the hi band stage after the SYTF and the result seems coherent.
So I opened the SYTF in search of spheres misalignment and I found the input one slightly off.
I managed to put if back into place quite easily , as the rods weren't glued  weird !
But this changed absolutely nothing.
I'm now quite sure the problem is in the FET switch that turns on and off the low band. It appears to let the low band signal pass through when selected but I think the FET that grounds the the input loop when high band is selected doesn' work anymore.
It's becoming tricky ! I don't know if I even could probe that without destroying the switch
If anybody has and idea about what I could try ?

[kirill_ka]

http://www.simonsdialogs.com/2018/12/hp-5086-7803-yig-tuned-filter-and-switch-sytf-details-and-adjustment/ 

[kirill_ka]

It appears to let the low band signal pass through when selected but I think the FET that grounds the the input loop when high band is selected doesn' work anymore.

Not likely. The switch doesn't affect the HI band path significantly.

[charlyd]

hello with a message on a 8595E amplitude out of spec > 3ghz
is that someting to do with the YIG.

[analogRF]

hello with a message on a 8595E amplitude out of spec > 3ghz
is that someting to do with the YIG.

In all likelihood, yes, it's the SYTF fault. There is a small chance that the driver of the SYTF is faulty too but
in all cases I have seen (8595E, 8596E, or 8563E which is more complex because it also has a mixer in it) it is always the YIG filter
or the switch that is integrated in the device.

a bit more info could be helpful though, if the "out of spec" means it's almost dead for the high band like >10-15dB off
then refer to what I explained above but if it is like 2-3dB off then it could be the calibration which is pretty easy to do.

[charlyd]

@analogRF  may i come back to you when i know more..it is still coming

[analogRF]

@analogRF  may i come back to you when i know more..it is still coming

absolutely, i hope i can help. you can pm me.

[charlyd]

can i test this with a cal  ( cal cable loop) i am new to SA , sorry.

[analogRF]

can i test this with a cal  ( cal cable loop) i am new to SA , sorry.

if the problem is the high band (>2.9GHz) no, not really. Only the last 5-6 harmonics of the cal are visible  at the beginning of the high band
and they are falling off quickly so it will not give you an indication of what (if any) is wrong with the high band