HP 8594E failing confidence test

[hackwell]

Hi everybody

I recently got a HP 8594E spectrum ( not logic  ) analyzer off ebay from south Korea at a reasonable price. Needless to say it was sold as defective with no further infomation.

The front panel isn't in good nick , the brightness knob is broke , many scratches and faded inscriptions and a missing screen filter.

But it powers up , the YIG osc is fine and displays the calibration signal at -30 dB instead of -20dB. The freq calibration passes but  the amplitude calibration fails with a RES BW AMP FAIL error message.

After some RTFM I found out how to cablibrate the A11 and  A13 BW filters. So I made three XTAL shorts with 10nF and 90 ohms in series and carried out the procedure. It's very easy to do and take 30 minutes to do it properly.

After that the AMPD calibration passes every time 

Buuuut not the confidence test...

And here is my problem : the conf test fails at the very begining with a POS PK FAIL message. The analyzer seems to be working fine tho.
According to the SM the processor board or analogue interface is defective.

Did some measurements with my scope at the input of the DAC on the processor board and the signal looks fine. the 3rd converter DAC's correction value is 199. A bit high but no problem whatsoever.
Then i noticed that if I set the ref level to -58dBm bad things happen ( see photo ) same thing at -68dBm and -78dBm. everything is fine below -57dBm

The default doen't appear on my scope tho.

I don't have a swap processor board but I swapped the two A11 and A13 BW filters 3rd converter board and  the A12 attenuator board and the failure is still there even with the 321.4 Mhz input disconnected.

so I suspect the positive peak detector on the processor assy but I can't figure out how it actually works (got the CLIP also )

Does someone have an idea abt how to solve this ? Or give me some informations on the peak detector ? What does it do exactly?

Regards
Alex

[Miti]

You can't fix it. Can you sell me the display?

Seriously now...
HP8594E is a spectrum analyzer, not a logic analyzer. Interesting that it shows something above the Ref level. Mine seems to clamp at the Ref level.

[MarkL]

Make sure the attenuator is operating properly.  You could have a failing step.

[hackwell]

 my TLA714 was just in front of me hence the mistake  corrected now
your display seems better than mine BTW 

I'll try with the same settings and see if mine clamps as yours does. Had a look at the video signal and the noise floor stops increasing whenit reaches 2V which is normal I think. And there's no clipping.

[hackwell]

I already swapped the A12 amplitude control with a known good one. Same result. And the signal increases smoothly all the way up to 2V.
The signal looks good at the DAC input that's what puzzles me

[hackwell]

@ Miti

You were right about the signal clamping at to top of the screen.

 Checked with my other SA (8595E with a bad YTF , but that's another story) and your photos and the 8594 doesn't clamp at all.
So my guess it that's the purpose of the positive peak detector. It's a quite complex circuit involving a differential amplifier and various reference voltages.

If someone has informations about how it works it would help me a lot !

I'm aslo looking for a YIG oscillator for my 8595E , they become hard to find apparently

[hackwell]

Problem solved.

After some head scratching and tons of measurements and RTFM I eventually tracked the failure down to the log amplifier.
It was impossible to calibrate as with all the adjustement set to their min or max the signal was still 4 divisions above what it shd be and le RF amplitude needed to align the LIN DC voltage to 1Vdc was -10dBm , instead of -14 dBm. Clearly a lack of gain somewhere.

Luckily I have a known good log amplifier so I measured the amplitude in log an lin mode at each test point in the signal path on both of them.
If someone needs these values I can post them , let me know.
The problem was located in the 6th and 7th stages. The amplitudes were good in log mode but totally out of range in lin mode.( almost by a half)

I first checked every component auround and swapped some with no result. Then I noticed that R93 and R101 were factory selected and bingo ! the ones on the bad board were 4.64 K as stated in the schematics but were 1.5 K on the good board.

So it seems that those boards have different gains in lin mode depending on which model of SA they're fitted into. And no mention of this in the SM.

After I replaced those 2 resistors I was able to calibrate everything within specs 
 I also had to readjust R88 and R21 which are factory settings too so there's no procedure for that either. Apparently it's related to the temperature compensation but how to calibrate these properly?

So be careful when replacing a complete board , they might be different in some factory adjustments.

Anyway the 8594E passes freq and amptd calibration and the confidence test and all values are spot on

But for some reason the signal doesn't clamp at the top of the screen as Miti's one does. Maybe a matter of firmware version? Or yet another failure ?

[MarkL]

Problem solved.
...
But for some reason the signal doesn't clamp at the top of the screen as Miti's one does. Maybe a matter of firmware version? Or yet another failure ?

Glad you found the problem!

Check that the video limiter is on.  If it's off it can cause this behavior on the screen.  It's in the CAL menus on the 4th page: "VID LIM".

[hackwell]

that's intersting thanks a lot

I'll check this and let you know 

[Miti]

Check that the video limiter is on.  If it's off it can cause this behavior on the screen.  It's in the CAL menus on the 4th page: "VID LIM".

Thanks Mark, I learned something today!
That setting is the culprit and it is persistent. It doesn't change if you power cycle or press PRESET. Smarter every day... 

[hackwell]

+1

just turned on the video limiter and it clamps now.

Thanks Mark 

[hackwell]

Now I have a dumb question.
why do I get a tone at 3.9 MHz 1.19 dBm even with or without input signal?