AnsweredAssumed Answered

8590A Spurious signal on display

Question asked by phil_r on Nov 22, 2010
Latest reply on Dec 23, 2010 by tabbott

Like • Show 0 Likes0
Comment • 3

Hi All,
I am slowly bringing an old 8590A Spectrum analyser back to life. It is generally OK. Displayed signal levels are down by about 2dB (after self calibration), but that could be my signal generator.

I have a strong spurious response appearing on the display (at about 3MHz) when displaying 0...10MHz and a resolution bandwidth of 30KHz or less is selected.

This disappears for resolution bandwidths of 100KHz or more

Changing the sweep speed does not seem to change the location of the spurious.

In both pictures, the 5MHz -20dBm signal is from my signal generator.

I am unsure if the spurious is from a fault in the bandwidth filters, or maybe the YIG oscillator, or maybe a glitch in the DAC drive to the tuning coils.
Also, the zero spur seems pretty strong. Maybe I have a blown first mixer.

Does anyone have any thoughts?

Thanks in advance, Phil

No one else has this question

Outcomes

3 Replies

tabbott Dec 7, 2010 2:40 PM
Mark Correct
Correct Answer
Hi Phil -

Even though the self calibration appears to be running and you are presumably storing that cal, there may be a goofy number in the cal table that is causing you problems. That cal routine will try to correct even when there is broken hardware. You will know this by a large correction value. Please review the attached and look over each number in the display cal data (Cal, More, More, Display Cal data…something like that), which will show typical cal values. When you change from a 100 kHz RBW to a 30 kHz RBW, the RBW filters change from an LC path to a crystal path. Maybe the crystal path is foo-bar and the filters need adjusting…or worse the hardware is broken altogether.

Interesting to that the L.O. feedthrough, basically your 0 Hz signal is way off the top of the screen. This is usually caused by a blown first mixer do to overpowering the front end.

Regards -

Relevant Products:#8590A#
Attachments
- Pages from 8590AServiceGuide[1].pdf201.2 KB
Like • Show 0 Likes0
Actions
phil_r Dec 9, 2010 3:43 PM
Mark Correct
Correct Answer
Thanks for that info. I had just got round to having a close look at the mixer (after suspecting the bandwidth filters, because the problem appeared only when the crystal filters were active). One mixer diode was BLOWN! (capitals because close inspection with a good magnifying glass revealed the whisker to be burnt off . The attenuator is also faulty - the resistors are all ok, but some relay contacts are badly intermittent. ( I think I can fix that one myself).

I replaced the mixer diodes with some matched 5082-2804 diodes which I had, but think these may not be suitable for the frequency. The result of this was that the spurious disappeared, the "close in" noise dropped to an acceptable level and the zero spike dropped to just below 0dBm. Maybe the YIG oscillator was getting upset by the loss of loading by the mixer and started putting out a spurious or two (I would have expected to see two spikes on the display, one either side of zero - and extra spikes around any valid input (none obvious), so I find this a bit confusing).

I am now thinking of replacing the mixer diodes (again) with a 1900-0264, which I saw referred to in another posting about 8591E mixers & maybe adding a second 1900-0264 as in the 8591E.

This analyser was purchased as "not going - for parts or repair", so it really doesn't matter what I do to it, though I want to fix it. At the moment it is looking very promising even though I completely stuffed the calibration by doing it with a bad attenuator (should not have pressed save . As you suggested, I will have a closer look at the cal data.

More on this later. Regards, Phil
Like • Show 0 Likes0
Actions
tabbott Dec 23, 2010 9:31 AM
Mark Correct
Correct Answer
phil_r asked...

Hi,
Thanks for your post about my problems with spurious responses on the display of an old 8590A spectrum analyser. It was very helpful.

I have managed to "trash" the software calibration due to a faulty attenuator. I think that fixing the attenuator will fix this.

Uncalibrated operation seems fine except that all signal levels are reported as being about 17dB too high (for example, apply -20dBm @ 5MHz, anaylser reports about -3dBm). Do you know how close the uncalibrated levels are supposed to be to the "correct" levels? Or does 17dB too much gain indicate a fault?

answer...

At 5 MHz the uncorrected amplitude result should not be 17 dB. I would expect 2-3 dB at the most. You may have to inject a signal at a know amplitude and follow it through the front end of the instrument to see where the power levels go south with another functioning spectrum analyzer. If you inject a signal at 50 MHz at -10 dBm, tune the suspect analyzer to 50 MHz, zero span, 0 dB of input attenuation, the output of the input attenuator should be 50 MHz at -10 dBm. You can change the input attenuation on the suspect analyzer in 10 dB steps and you should see that power level on the functioning spectrum analyzer at 50 MHz drop by 10 dB as well if the attenuator is working correctly.

Next, you can measure the output of the 1st converter. The 1st I.F. is 2050 MHz (2100 MHz 1st L.O. mixing with 50 MHz R.F.) and that power level should be -22 dBm, so there is 12 dB of conversion loss through the 1st converter. It is vital that the L.O. power is sufficient going into the 1st converter in order for it to operate correctly. If you tune the analyzer to 50 MHz, zero span the 1st L.O. should be at 2100 MHz with a power level going into the 1st converter of +12 to +18 dBm. If that is low, the conversion loss will be high. The next good measurement point is the output of the 2nd converter. That should be 321.4 MHz at ~-25 dBm. There is ~5 db of conversion loss through the 2nd converter. So, the overall conversion loss through the RF front end is ~17 dB. Measure your suspect analyzer. Due to the high L.O. feedthrough I’m guessing your 1st mixer is blown, however the input attenuator if broken can cause a strange match to the 1st L.O. and cause issues. With 3-4 simple measurements you will know exactly what is broken.

Regards -
Like • Show 0 Likes0
Actions

8590A Spurious signal on display

Outcomes

Attachments

Related Content

Recommended Content