30 dB loss on 8595E spectrum analyzer

[ezalys]

I'm trying to track down 30 dB of loss on an 8595E spectrum analyzer I got. I've tested the frontend attenuator with a VNA... it appears to always have 10 dB of loss. Is this normal? If so, then the fault is isolated to the SYTF... which is pretty depressing. How fixable are these? I do, fortunately, have access to a wedge bonder, which I hear is useful for fixing this sort of thing. What's curious is that the fault appears across band zero and band one... so that would imply maybe that the switch portion of the SYTF is blown rather than the yig bit. What say you guys?

[Armadillo]

Have you watch this;




I hope is not the YIG. You should test it as per video above.

[ezalys]

That analyzer has no YTF.

[Armadillo]

That analyzer has no YTF.

That is a good response.
Enjoy. 

[Bud]

Did you test with DC coupling?

[ezalys]

Okay I've traced it to the SYTF. This sucks. I probed the control signals and it's (0V,-15V) for low band, and (-15V,0V) for high band. Does this sound like a PIN switch, a BJT switch, or a FET switch? What do I have to be careful about when opening the SYTF? I can work on it in a cleanroom if necessary.

[gab_au]

Hi,
before de-lidding the switched yig filter assy, a few questions and some thoughts:

1) You indicate the initial issue was low readings. Are there any error messages?
2) I assume you tried a cal, and that failed, I expect that would have thrown up some errors?
3) By saying it's the switched filter I assume then it works on the 0-2.9Ghz band but not on the other?
4) I assume you've injected a signal after the yig filter on the high band and it reads near enough to the correct level?

If the answer to 3 is YES and 4 is YES then first I'd check to see if the filter is switching out of the 0-2.9GHz path when it's supposed to.

If the answer is YES it switches away from the low band then I'd check to see if the filter is off frequency due to a fault in the driver. Use your VNA to sweep across the filter when the spec an is set to some arbitrary frequency in the high band and set to zero span. Then sweep the vna across the whole high band +/- some to see if there is a passband anywhere (obviously with a number of frequency points to give a freq step which is  much less than the yig bandwidth - perhaps you may need to do the whole  high band in 3 or 4  or more chunks if the number of points in the vna sweep is not sufficient.)

If there's no passband observed then I'd check the yig driver signal is getting a ramp during the sweep (span is over the full high band).  If it is sweeping then I'd check to see if the yig drive current is following the ramp (also check the resistance of the coil - perhaps 5 -20 ohm?). If it is, then I'd check the yig heater (I assume the filter is wideband, say 20Mhz or so as I don't see any pll around the filter). If the heater is drawing current (you could also check the resistance of the heater) and the tuning coil is drawing current (perhaps anywhere from 100mA to 500mA? dependant on frequency set point)....

Once that's all done then you're in a position to know if the switched yig filter assy should be opened and then you can start thinking about fault finding the pin switches and then perhaps think about de-lidding the yig filter.

like I said - just some questions and thoughts and I expect some or all of this was done as you've come to the conclusion it is the switched yig filter assy.

Greg

[MarkL]

I'm trying to track down 30 dB of loss on an 8595E spectrum analyzer I got. I've tested the frontend attenuator with a VNA... it appears to always have 10 dB of loss. Is this normal?
...

I realize you're focusing on the SYTF now, but to answer this question, no, the attenuator should not have a loss of 10dB.

As a safety feature, note that the SA won't automatically set the attenuator to less than 10dB when adjusting the amplitude.  You can force 0dB attenuation by setting ATTEN to MAN, and then typing in 0dB on the keypad (it's in the AMPLITUDE menu).  Perhaps that was the issue?

FYI, here's the inside of a 8595E SYTF.  I wouldn't open it unless you've proven it to be at fault.

[ezalys]

gab_au:

No no. The SYTF assembly is a combination YTF and SPDT switch. If the switch is in the low-band position, signal goes in the input port, and out the low-band output port. If the switch is in the high-band position, signal goes in the input port, through a YTF, and then out the high-band output port. It is the low-band signal path that I see 30 dB of attenuation on. The high-band port I have not yet checked, as I don't have a high enough frequency VNA.

MarkL: Ahhh nice catch! Yeah. I didn't know that there was still a lower position you could place the attenuator in. Indeed the value shown under "ATTEN" matched the attenuation of the attenuator assembly, as verified by a VNA. So this part is fine.

[MarkL]

How about a simple test?  It doesn't involve the VNA, unless you want to use it as a signal generator.  I assume you have the front end partially disassembled and accessible at this point.

- Lock the SA on Band 0 (in the FREQUENCY menu).
- Feed a known amplitude test signal into the SA input, say 300MHz @ -20dBm.  It should be shown low by 30dB, as you report.
- Disconnect the input and low band coax lines from the SYTF and use an SMA F-F jumper to route the signal completely around the SYTF (connect the input coax to the low band output coax).
- Does the test signal still have a significant amplituide loss, or is it more or less correct now?

This is what I would do, at any rate.  But maybe you're beyond this point already.

[gab_au]

Hi,
on that basis - I wouldn't be taking the lid of anything until you have isolated the problem. You at least need to inject a low band signal after the switched filter to ensure it's not an issue with a local oscillator , mixer, or IF stage.

Then you'll need to find a signal source above 3 GHz to check the high band in the same way and go from there.

Greg

[ezalys]

Does anyone know how these things typically die? Do they blow at the wire bond or do the junctions themselves blow through?

[ezalys]

MarkL

Would you mind explaining a little more about what’s going on in the picture? The coax on the left seems to be input. Then the blue patch is maybe decoupling. The large die looks like a switch, where the input is on the bottom, the control lines hit the pads at the top, and the outputs seem to be on the left and right. How does the signal leave for the low band? The high band seems to leave the switch on the right where it’s coupled into the ytf. What are the two small black square at the bottom? Is it a limiter diode pair? Why would you stick a limiter diode pair right before the YTF instead of before the switch?

[ezalys]

Or wait is the input on the top? Why are both sides of the switch activated then? What is that blue patch? I’m so confused.

[ezalys]

If perhaps the top coax corresponds to the input, then the two squares maybe correspond to limiter diodes on the input which makes sense. I then don't understand why you'd leave half the SPDT switch disconnected even if you're driving it. I just don't understand what happens to the reverse signal flows. I've drawn my understanding of the signal flows in the attached images... maybe a MMIC and YIG expert has some insights as to what's going on?

[MarkL]

I need to defer to others who are familiar with this type of design.

I found this page which may be helpful:

  https://microwaves101.com/encyclopedias/yig-technology

It's the same SYTF, and you have the signals identified correctly (low band output is closest to the switch control inputs).

If your YIG filter is still working and it's just your switch that's dead, you might want to consider patching together an external switch solution rather than risk damage to the YIG filter given the close proximity.  If the YIG gets ruined, it may not be worth the stupid ebay pricing for a replacement SYTF to fix the 8595E.  That's where I am right now with one of my units.

[ezalys]

I think I'm just going to sell this thing for parts, given that I'm not a huge fan of this analyzer to begin with. I've noticed that the return loss on the input port changes depending on the orientation of the SYTF, so I think it's hosed. Unfortunate