HP8590B Spectrum Analyzer Repair Attempt

[2nOrderEDO]

Hello fellow forum users,

Last week, I received an HP8590B Spectrum Analyzer (9 kHz - 1.8 GHz) that I got very cheaply on the secondhand market. The seller's description read as follows:

Spectrum analyzer 9 kHz - 1.8 GHz
It needs someone who can help it out.
The 230V power supply struggles a little to start up if left unattended off the grid for a while.
Otherwise, it works well, but it has a frequency shift which certainly comes from the mixer.
You'll find plenty on YouTube.

I can verify that after plugging the analyzer into the mains, it takes around 10 minutes until it can be powered on. For the first 5 minutes, nothing happens, but after that, a whistling sound comes from the supply, and after some minutes, it is gone. That is when turning it on becomes possible.

After checking that all voltage outputs from the power supply were OK, I decided to postpone fixing the startup problem of the power supply because there are more serious problems that directly affect the core functionality of the device.

When fed with any "low" frequency signal up to 25 MHz (the maximum frequency my R&S RTB2000 can generate), no signal peaks show up on the display. What is even more strange is a peak at around -600 MHz (yes, minus) that I can visibly see changing when varying the frequency and amplitude of the signal I'm feeding from the generator.

When using the CAL Out signal from the HP8590B front panel, I can see the fundamental at 300 MHz and all the corresponding harmonics. Funny enough, the fundamental sometimes is at 316 MHz, sometimes at 360 MHz... depending on how the analyzer feels after a power cycle. I verified with my RTB2000 (300 MHz BW) that at least the fundamental of the CAL Out signal is precisely 300 MHz. This problem with the frequency shift I was able to trace to a defective LT1013 opamp that is in charge of converting the current output of the DAC channel used for the YTO Coarse tuning. Nevertheless, this does not explain why the CAL output can be measured, but all "low frequency" signals I feed from a signal generator do not show up on the spectrum.

Fearing it could be the RF diodes of the first mixer, I disassembled the unit and checked them with a multimeter. It's hard to tell if they are OK by measuring in-circuit, but I would prefer not to disassemble these diodes (SOT23) unless I can justify my suspicion.

Are there any fellow users with experience with the HP85xx series that can offer a helping hand?

[inlinebeginner]

Hello 2nOrderEDO:
  I am repairing my HP8593A, still need some time to finish.
HP859x was built around 1990, it's been 30+ years, you may want to check:
1) NVRAM battery. Most likely it is dead. Mine measured 0.3V (should be 3V), therefore, all the correction constants data were lost. With an invalid NVRAM, the boot is not consistent. After the battery is changed to a new one, mine cannot boot. I have to discharge NVRMA, boot and connect the battery back to do calibrations. The battery TL-5104 life is 8 years at 25degree C, and 1 year at 50degree C, mines never changed since 1990. After 33 years, it finally died.
2) The EPROMs may be bad (some bit flipped). My HP8566B had that problem, after I copied that EPROM from a working unit, it was fixed. HP859x has 16 LEDs, the top 4 are for EPROMs U7, U6, U23, U24, if any one of them is bad, that LED will turn on. Just check that.
3) Power supply: some electrolyte caps ESR are too high, replace them.
4) The power supply PCB connector: a D-sub, some of the pins may not contact well. Over the time, it causes randomly power shutdown, unplug and plug back may help.
  I hope this can help.
Lin

[2nOrderEDO]

Hello Inlinebeginner, thanks for your help. I will add my comments to each of your points:

1. The NVRAM on my HP8590B was replaced back in 2020 and It correctly measured around 3V when I checked it.
2. The LEDs in the motherboard do not indicate a problem with the EPROMs, but I will anyway re-flash a newer version of the FW. I was able to find the firmware for the HP8591A, which I think it is compatible, based on the following document from HP: https://www.mikrocontroller.net/attachment/308962/Firmware_History.pdf
3. Agreed, I will perform a full recap of the supply and check if any component related to start up is damaged.
4. Thanks for the tip. Luckly, it doesn't seem to be the case for this unit.

Can someone verify spectrum analyzers from the HP85xx family are supposed to show NEGATIVE frequencies? That is what puzzles me the most.

For now my action plan is to refresh the EPROMs and order some components like the LT1013 opamps and PSU electrolytics that need replacement. I will also order the first mixer stage diodes, just in case. Next week I will borrow another SA analyzer and an EPROM UVC erased from a friend to continue the troubleshooting.

In the meantime, any suggestions, opinions, etc. are welcome. If anyone thinks of a test I can make to report back, I would gladly do so.

Regards,
2nOrderEDO

[807]

...Can someone verify spectrum analyzers from the HP85xx family are supposed to show NEGATIVE frequencies? That is what puzzles me the most...

I have the HP 8591E. There is a frequency marker at 0 which effectively acts as a mirror. So if you set the analyzer to centre frequency of 0MHz with a span of 100MHz and input a signal at 10MHz, you would see a signal at 10MHz & another at -10MHz of equal amplitude, 1 division away from the centre.

I cannot adjust my analyzer to show down to -600MHz though. What settings were you using to see that?

[ps]

I can verify that after plugging the analyzer into the mains, it takes around 10 minutes until it can be powered on. For the first 5 minutes, nothing happens, but after that, a whistling sound comes from the supply, and after some minutes, it is gone. That is when turning it on becomes possible.

I had a similar problem with an 8591A. A small capacitor in the power supply unit was leaking and the electrolyte got into the voltage reference circuit. At first the analyzer had starting problems, which disappeared after checking the capacitors and replacing the defective ones. A few days later, the analyzer started regularly but switched off after a few minutes. This was because the supply voltages increased until they triggered the overvoltage crowbar. Cleaning the circuit board, in particular the area below the leaking capacitor, solved the problem.

[inlinebeginner]

Hello Inlinebeginner, thanks for your help. I will add my comments to each of your points:

2. The LEDs in the motherboard do not indicate a problem with the EPROMs, but I will anyway re-flash a newer version of the FW. I was able to find the firmware for the HP8591A, which I think it is compatible, based on the following document from HP: https://www.mikrocontroller.net/attachment/308962/Firmware_History.pdf
2nOrderEDO

1) Write down NVRAM data first before you try a new version software.  "8590 Series Analyzers Assembly-Level Repair 08590-90316"  Page:201:
2) If the no LED turned on, may be you can keep using the current firmware for troubleshooting, unless you have to change to a new firmware. After troubleshooting, you may try the new firmware. Step by step.

Lin