?

Hi,

?

It would be easier to check the attenuator between the RF input of the 8591E and the output of the attenuator.

This way no other circuits would be involved. You will have a better idea of ??the state of the attenuator.

?

Yves

?

?

Hi,

I did the same tests as you, results in the tables.

?

I attached a picture showing the diodes of the first mixer, I did the repair of two Spectrum this year with damaged diodes, so the one with which I did the attenuator tests today.

Maybe you should do this check. Ideally, if you had another measuring instrument that would allow you to know the RF level at the J3 connector of the first converter, you could know if the first converter is OK.

?

correction at OFF

SPAN 100kHz

resBW 10kHz

vidBW 100Hz

MKR peak

REFlevel at -10dBm

?

?

Yves

?

?

Hi,

?

If you have lost your calibration constants, it will be possible to rebuild them.

For the moment there is the attenuator to repair... it is easily achievable according to the image.

?

But I fear for your mixer, there is a good chance that it is damaged.

But that too is repairable, it is only two SOT-23 components to replace (diodes).

?

Attenuator values-based of attenuation.

?

?

?

Attenuator from RF input to first converter input. At left, absolute loss vs frequency, at right, VNA normalized to 0dB attenuator trace.

?

?

Yves

?

?

Hi Marco,

?

I don't know what value your Spectrum's attenuator was set to.

Even if it wasn't connected, the IF gain will change according to that value.

So, I did a test at 0dB and 10dB, even if the Spectrum's internal attenuator wasn't connected.

I inserted a precise level at -20dBm at 300 or 1500 MHz.

?

Now looking at the results, it may seem surprising that the values ??at 300 and 1500 MHz are very similar.

In fact, this is normal in my case since, in the calibration table (Flatness) there is a gap of only 1dB between the level at 300 MHz and 1500 MHz, at 1500MHz the signal is attenuated by 1dB, and this 1dB is mainly due to the cables connecting the Spectrum input to the attenuator and from the attenuator to the first converter.

Also, the attenuator does not have a flat frequency response.

Since I connected directly to the first converter, these losses are not included in the test measurement.

?

?

?

?

Marco, I forgot to include these remarks

?

Here are two signs of a damaged mixer that I have already noticed (blue curves):

- At 0 Hz, the carrier exceeds the reference level

- PRESET: ATTENUATOR = 0 dB (whether the attenuator is installed or not does not change anything). Between 0Hz and 10 MHz, the noise increases in the lower frequencies instead of remaining relatively flat.

?

Yves

?

Hi Marco,

?

Here are your results and mine tabulated.

That the source signal is not exactly -20.0dBm, will not change the conclusion.

Just like you, I used my 300 MHz and harmonics calibrator as a signal source.

I had done the tests for the Spectrum attenuator at 0dB or 10dB, not knowing what this condition was in your previous email.

Here we have the 2 conditions, so I used them.

?

Using the fact that there is a test where there is no external attenuator, allows us to evaluate the external attenuators of 10dB and 20dB as well as their sum.

We were able to assume that the values ??of the external attenuators (10dB and 20dB) are quite close to their nominal values.

This allows us to see the linearity over a 30dB range.

Calculating the difference between, without attenuator and 10dB, I would expect -10dB difference, if the attenuator was precisely 10dB.

And whether the Spectrum is with its attenuator configuration of 0dB or 10dB, should not change this result, even for jumps of 20dB, or 30dB.

?

So, columns E and F should show values ??of -10, -20 and -30 if the external attenuators were perfect and the mixer was perfectly linear.

Column G should show the value of 0dB everywhere.

?

If you look at the results of my Spectrum 8591A, the results follow the logic of results -10, -20 and -30, then 0 in column G

?

But your Spectrum 8591E has non-negligible deviations, and considering the images, I would say that your first converter is damaged… but it is easily repairable.

I will send you the details.

?

-----------

?

For screenshots, it's a software I made for a GPIB interface, but I can certainly use the serial link if that can help you.

?

Having lost your calibration constants, it is probably normal to have this offset.

But it would be preferable for now to do the attenuator repair before doing a CAL FREQ.

?

If you manage to get a TinySA ultra, you will be able to directly measure the output of the first converter at 2.1214GHz

?

Yves

?

De?: [email protected] <[email protected]> De la part de jansen.marco90 via groups.io
贰苍惫辞测é?: 9 novembre 2024 08:57
??: [email protected]
Objet?: Re: [HP-Agilent-Keysight-equipment] Troubleshooting a 8591e with opt 130

?

Hi Yves,

?

I did the measurements all with the SA's attenuator set to 10dBm. I redid the measurements with both SA attenuator settings, see table below.

I'm still using the cal out of the SA because I don't have a more trustworthy source yet.

?

?

On the other images, I tried to reproduce them, I'm just taking pictures of the screen though. Are you using a GPIB interface to make those? Or can it also be done with the serial output?

The noise increase at lower frequencies doesn't seem to be there. There is an offset in the frequency though, the 0Hz peak seems to be around 580kHz

?

?

in a broader span with the cal signal connected (through 10dB external attenuator) The 0Hz peak is near the reference level, the 300MHz signal is also quite high.?

?

?

So, I’ll try to fix the attenuator and move on with another try of the self-call from there. I'm thinking of getting a TinySA ultra to be able to do some verification.?

?

Marco

Hi Marco,

?

The diode Macom MA4E1338B1 would be OK, I had also found this equivalent.

There was also TPM2840-03, a little bit more performing than the original diode, but really very close...only 5V protection however.

?

I found a picture showing the section about the first converter of the 8591E which is different from the 8591A.

?

The method I used:

• I replaced cr1 and cr2
• I close the lid and I test, I see if the signal to the monitor at 0Hz, is lower than ideally, at the first division (upper left, picture 8591 LO leakage).
• If the diodes are not matched well enough, the carrier at 0Hz will exceed the first division at the top of the left screen.
• In such a case, change CR2 only, because easier to replace and redo the test.
• It may work from the first attempt as in my case, but I had still tried a second attempt out of curiosity, and got slightly better, so I did not continue with further attempts.
• I successfully closed, tested and calibrated everything because there was no other problem.

?

I will do some tests with the screen capture program controlling the rs-232 port.

?

Yves

?

?

?

Hi Marco,

?

The A12 module acts on a range determined by the Reference Level, between 0dBm and -50dBm.

With the following configuration, I was able to verify that all the attenuation jumps are functional.

?

I connect the reference to the Spectrum input, but put an external 30dB attenuator there.

Thus, the source at 300 MHz which normally is -20dBm, will have been read as -50dBm.

?

PRESET

CENTER FREQ 300 MHz

SPAN 0

REF LEVEL 0DBM.

?

By changing the reference level from 0dBm to -50dBm, by 10dB jump with the UP/DOWN arrows, you will see if all the attenuators are functional.

And at your choice for a Reference Level between 0dBm and -40dBm, make 1dB jumps to check the calibration attenuator.

?

But having as RF level read at 300 MHz of -5dBm instead of -20dBm, add another 15dB of external attenuation and put the resBW at 100kHz.

The goal is to check the 50dB range of the A12 module. If A12 has 50dB of range then, the problem is maybe on A16.

?

But, a measurement on AUX IF could be revealing.

Without any external attenuator, enter the 300 MHz reference to the Spectrum and put the reference level at 0dBm.

On my side I read, -38.2dBm at 21.4MHz.

The AUX IF OUTPUT on the rear panel is connected to A13, the module that follows A12.

?

Yves

?

?

?

?

Hi Marco,

?

Just to be sure, is the A14W1 jumper there or not.

Because, if there is a jumper installed at that position, it must be removed according to the documentation on that option 130.

?

Yves

?

?

?

Hi Marco, I'll look into a way to test the A12 module, but here are a few notes...

?

The test you did injecting the 300 MHz reference directly to the Spectrum, the measurement on the screen was -5dBm, and at the AUX IF output being -39.2dBm means that the overall gain of the Spectrum input was normal, if the REF LEVEL was 0dBm.

This would mean that the compensation was added after the A12 stage. But it is probably during the SELFCAL that the constants received incorrect values.

?

To try to find a bad stage for the gain, it will probably be necessary to work in CAL CORRECT mode at OFF

?

It would be a good idea to measure the LOG detector by doing the following test (CAL CORRECT at OFF):

- Ref Level at 0dBm, 300 MHz, zero span, resBW 1kHz, sweep time 1sec

- Put a signal of about -70dBm, and adjusted so that the trace arrives on the -70dBm line.

- Raise the level in 10dB steps, from -70 to -10. The trace should move in 10dB steps as in the picture.

?

Filters check (resBW):

- CAL CORRECT to OFF

- REF 300 MHz connected to input 8591E

- Verify each resBW include option 130 (resBW 300Hz, 200Hz, 100Hz and 30 Hz).

- Here I have superimposed the images to show you that there will be a little difference in levels and frequency, which is corrected during SELFCAL, but no big gaps between the filters.

?

?

?

?

?

-----------------------------------------------

?

?

Here are the three images in comparison: My 8591A, your 8591E after SELFCAL and your 8591EE with default CAL.

- I also injected -70dBm, starting from this level it will be easier to evaluate the value of the excess gain.

- For my 8591A, when I change the REF LEVEL by -10dB, the trace goes up by +10dB, and this, for the 50dB of range involving the A12 module.

- For your 8591E, some jumps have a little more than 10dB but mainly the last one, for a total of +5dB with default CAL and nearly +15dB after SELFCAL.

?

?

Send me a text file captured by the rs-232 serial link with your hpgl data, I will look at the behavior with my program.

?

?