Please help: HP8593A has no display after a dead battery changed

[inlinebeginner]

HP8593A, a spectra analyzer, 9k-22GHz, I repaired its power supply several years ago and it has been working well until recently the battery (see attached) died and lost all the NVRAM data. Before I change the battery, it can boot (see attached).  After I replaced the dead battery with a 3.7V battery, at power on boot, it only displayed one time, after that power boot or Preset, the display flashes randomly.
I removed the new battery and discharged the 0.22F cap at the RAM power pin, it can have a normal  display ( but not always) at boot as before, scanning very slowly, this display is only at power on boot.  If I use  “Preset”, the display will disappear and become screen flashing randomly.
I tried to use GPIB commands such as “CF-2001 Hz”, but its GPIB address can not be found in NI MAX scan device. Since “GPIB address 0”, is displayed at power on boot, I force writing to GPIB address 0, it has no response. I guess its GPIB module was not initialized due to the RAM data loss.
The firmware information: EPROM: 27C512  64kbytes
Label:08592-80076 U7  @90-01-03
Label:08592-80075 U6  @90-01-03
Label:08592-80077 U23  @90-01-03
.bin  Label:08592-80078 U24  @90-01-03
Since kazubek was looking for this firmware, I uploaded to here: https://www.eevblog.com/forum/repair/hp8593a-spectrum-analyzer-firmware-roms-needed/

Looks like both the local display/keypad and remote GPIB are not working, what should I do to bring it back and  calibrate the correction constants? Any ideas?
Thanks for the help and any comments.

[m k]

From "it can boot" to "display flashes randomly" you and took the machine to pieces, changed a battery and put pieces back together.

Changing a battery and "display flashes randomly" doesn't add up.
Maybe your assembly went wrong somehow or a connector pin cracked its joint.

[Hamelec]

"I removed the new battery and discharged the 0.22F cap at the RAM power pin"

now you lost all your calibration data if you havent saved it before.

Good luck..

[inlinebeginner]

From "it can boot" to "display flashes randomly" you and took the machine to pieces, changed a battery and put pieces back together.

Changing a battery and "display flashes randomly" doesn't add up.
Maybe your assembly went wrong somehow or a connector pin cracked its joint.

Thanks for the reply M K.

I think the firmware has bug to initialize the display. On the service manual  "8590 Series Analyzers Assembly-Level Repair 08590-90316"  Page:204:
"Firmware startup sequence
The firmware installed in the analyzer recognizes when the analyzer
non-volatile memory is lost by comparing the contents of two RAM
locations with known values. If there is a discrepancy, the startup
routine is initiated.
The analyzer startup routine does the following:
• User memory is erased
• DLP editor memory is initialized
• Power-on state is set to PRESET
• Windows are initialized
• Video constants are initialized " <=== firmware might have bug in this code, or load a wrong parameter for blank NVRAM. It could have several options, depends upon model # or SN#, since all lost, the firmware does not know which one should be loaded.

I have seen someone else had similar booting problem after change new battery.

To be short:  I already solved this problem with a small trick.
1) Power off the SA, and connect the battery with a connector (instead of solder it on PCB. Unplug the battery).
2) Discharge the 0.22F cap with a 100 ohm resistor (about 2 minutes). it unit is "F" not "uF", this cap can hold data more than 12 hours if charged to 3V, it is used for battery change. for my case, I have to discharge it on purpose (the new battery I put in charged it to 3.7V) otherwise, there is no way to have display.
3) Power on SA you will see the normal display, and the SA response to keypad. If it does not have display or does not response keypad, just keep trying, until it reaches this step.
4) Plug in the battery connector, wait for several minutes, let the 0.22F cap charge to above 2V. Between the battery and cap, there are 1 k resistor and diode, that is why you have to wait several minutes. You can watch the cap charging with a multi-meter.
5) Use "IDSET 8593;" set the model number. Then reload the default configuration under cal. and save the configuration to NVRAM.

Above 5 steps will solve the problem. After that SA can boot and has normal display. GPIB default address 18 is also recovered. Further calibration is required. With auto amplitude and frequency cal, the SA functions are recovered, just needs more calibration to make it accurate. The timebase, the flatness correction constant, the step attenuator correction are the three major work. I am still working on.

Thanks
Lin

[inlinebeginner]

"I removed the new battery and discharged the 0.22F cap at the RAM power pin"

now you lost all your calibration data if you havent saved it before.

Good luck..

Thanks for reply Hamelec.
The calibration data in NVRAM was lost before I changed to a new battery, it doesn't matter to loss the data one more time by discharging the 0.22F cap. I did not realize that the data should be backup until it stopped working.
My previous reply already described how I solved this problem.
Regard
Lin

[m k]

So the actual flashing frequency was much slower than what I had in mind.

Maybe you should change thread subject part "please help" to "solved" or something like that, for wandering souls in the future.

[inlinebeginner]

So the actual flashing frequency was much slower than what I had in mind.

Maybe you should change thread subject part "please help" to "solved" or something like that, for wandering souls in the future.

@m k, thanks for your suggestion.
I calibrated flatness up to 19.3GHz, it works accurately. Then I calibrate the step attenuator 1,2,4,8 dB, it is also very accurate, but after power off and power on, the "correct on/off" under CAL is set to "off" and can not be changed to "on", that means all my calibrations are disabled. Don't know why. Anyone has similar experience? Maybe it requires doing the step attenuator error calibration first, then do the flatness?
I will load default config again, input step attenuator error first, to see what it happens.

[inlinebeginner]

update:
  After I tried many combinations, finally, my HP8593A calibration is done.

Timebase: 300,000,004.1Hz measured at 300MHz CAL OUT with a 300MHz low pass filter.

Band 0: 0-2.9GHz, dBm error <0.1dB (reference is HP8484 power head)
Band 1 to Band 3 (2.75-19.3GHz), dBm error <0.3dB
Band 4: 19.1-20.5GHz, dBm error <0.3dB （estimated）
Band 4: 20.5-22GHz，did not calibrate, my RF signal source is only good up to 20.5GHz. Also, RF power head may not be accurate for frequencies higher than 20.5GHz. With a HP8485 power head,  it should be easily calibrated to 22GHz.

Step attenuator error: <0.05dB （I could hardly see error).

YTF cal, -37Hz Freq Cal solve the problem. Looks like after any YTF cal, you need to redo the flatness cal again, otherwise, it could have about +/-1dB error.
For 0-2.9GHz, calibration is a little bit easy, but for higher frequency, I have to repeat several times to make it accurate. A lot of details, very time consuming.

Lin