Advantest R3465 Spectrum Analyzer Repair Help : Input Attenuator

[Mechatrommer]

currently trying to repair Advantest R3465 Spectrum Analyzer. calibration process is failure due Input Att. after dismantling what i believe as the input attenuator, it seems nothing wrong with it. all switches tested to function normally, attenuation also spot on (except wobbling on high frequency (4 - 6GHz) range but i blame on my VNA impedance mistmatch issue). when reassembled, calibration still report the same issue, i can hear switches activated/deactivated properly during operation so it ruled out 24V power supply failure. so seeking help what other possibility on the failure. attached are pictures of the attenuator, datasheet and some SA RF assembly to get the big picture. help and thank you.

datasheet:
https://www.hirose.co.kr/Upload/Product/Catalog/e36100124.pdf

[edpalmer42]

I noticed that on my R3465 the internal calibration would sometimes fail if the circuit was too far out of adjustment  The automatic procedure didn't seem to be able to reach the correct value - even though that value was within the adjustment range.

Instead of doing the automatic calibration that tests and sets everything, I did each calibration test seperately in a logical order.  i.e. you have to calibrate THIS before you calibrate THAT.  I didn't keep detailed notes so I can't be more precise.  Even then, you might have to do the test a couple of times before it passes.

I also seem to remember adding an external attenuator to the input to help it get closer to calibration.  I think in that case it was claiming that the reference level was too high.

Did you confirm that the 30 MHz reference is on frequency and at the correct level?

Ed

[Mechatrommer]

i did try to calibrate individually. when i tried IF Step Amp it also report Input Att failure. the other item will report HS ADC failure (the problem becomes worse). i did Total gain cal it success without warning but...

Did you confirm that the 30 MHz reference is on frequency and at the correct level?

i checked with my KC901V in SA mode, the 30MHz CAL out is spot on at -10dB... currently R3465 reads about -5dB lower than that. with failure to complete all calibration, the R3465 reading is a bit off. and unluckily LPRO101 Rubidium that i bought from a member here also playing up, cannot locked when i need it most

i believe i need to check further downstream, next to Att is YIG BPF model Advantest TOP2302X, not sure what this one does, i can find in the net stating its YIG Filter, but no datasheet. band pass filter of what? is this IF stage? what should i expect on the output? see attached.

[edpalmer42]

The TOP2302X is a combination of a YIG bandpass filter and a switch for low band vs. high band.  The YIG is only used on the high band.  Its purpose is to track the sweep and thereby reduce interference from off-frequency signals.  It would be worthwhile to measure the loss from the input to the low frequency output.  Then remove the TOP2302X by jumpering from the attenuator output to the low frequency input on the RF deck.  Include a variable attenuator in the link.  10dB should be enough.  Reduce this attenuator and see if you can nudge the autocal routine in the right direction.  Confirm that you can get the autocal to pass, even if it requires an unrealistic input level.  You want to confirm that there isn't a fault with the autocal process before looking for problems in the RF deck.

I had to disconnect my TOP2302X permanently.  One of the heaters in the YIG fried and cracked.  The YIG sphere is now out of position and there's basically no signal throughput.  I have to move a cable to go between bands, but I rarely use the high band so it isn't much of a problem.  Besides, I got the unit really cheap! 

Note:  In my unit, I found that many of the surface mount electrolytics had gone bad - high ESR.  I'm in the middle of a total replacement for two or three values.  It's not possible to be sure, but maybe a bad one is confusing the autocal process.  I remember having a trace that showed noise about half a screen high.  Replacing one capacitor gave me the slightly noisy line I was expecting.  None of the other replacements have had such a dramatic effect.

Ed

[scopeman]

Did you fix this thing?

Sam
W3OHM

[Mechatrommer]

nope. its still there just like that, i'm too lazy aka not enough time. it can be used just that i cant self calibrate it.

[scopeman]

I am wondering if the attenuator is the same Weinschel attenuator that is used in the R3271A which I believe is either made for Weinschel by HP or was licensed by them. If so it probably has the famous dampener o-ring failures that this design is noted for.

I have an R3271A that I am currently refurbishing. I just ordered all of the electrolytic caps for it as several in the power supply were marginal. As big an effort as it is to take this unit apart caps are cheap and it has a plethora of SMT electrolytics in it that are known leakage problems.

Mine self cal'd ok the first time I powered it up and then the next day and thereafter did not. It would fail saying that the self cal 25MHz signal could not be found even though it was there (and on the screen) and close to the correct level. So I am just going to go thorough it in a methodical fashion and see if I can get it cleaned up.

73,
Sam
W3OHM

[Mechatrommer]

I am wondering if the attenuator is the same Weinschel attenuator that is used in the R3271A which I believe is either made for Weinschel by HP or was licensed by them. If so it probably has the famous dampener o-ring failures that this design is noted for.

i did test all the switches electrically in the attenuator, they all seems (clicked) fine. i also checked with VNA the attenuation of each switch activation, they are all seems correct. so beyond this electrical test i dont have any more idea, in later post above, i moved to the Yig Osc and thats where i'm stucked. i'm not sure how a o-ring damper failure can be detected by self calibration, please do explain, thanks.

[scopeman]

Perhaps contact bounce. Just a guess here as I don't know for sure if that is a real issue. I just know that the lack of good O rings can cause the switches to fail to make. Once I get mine back together I will know more.

I will probably do the attenuator last depending on how much I have to put back together before I can get the instrument into a state where I can re-run the previously failing self test.

If the attenuator design is what I think it is since the analyzer was built in the early 1990's I would bet that it is going to need them.

Sam
W3OHM

[Mechatrommer]

i wish you success and report back in this thread how it goes.