Battery replacement in HP 8591A Spectrum Analyzer

[kawal]

I read in this forum about someone that has turned on a unit like this and to their surprise the constants where lost before they could back it up. So finally decided to change batteries.

Hp was nice enough to provide the battery installation date on a sticker on back of the units. Both units have a Battery install date of 1991. I did check the battery dimensions inside the unit and they are 50mm long and 15mm diameter with Axial tabs. Original battery was made by Tadiran and the only marking I could see was AA.

I did a little research for the right battery choice.  First we need to know the current requirements. The HP 8591A uses Static RAM  TC55257 which only provides max standby current range between 2uA to 30uA per chip. This is useless information.  Next step would be to ignore the current usage  since after all the battery lasted almost 30 years and still works.
These are ~2Ah batteries (assuming all the capacity is used up by now ) so we can estimate that the average current to be no more than 7.8uA . In reality this is probably a lot less.   

So you want a battery with low current discharge and long life.  The TL-4903 is rated for 40Years on shelf and  is rated for ~55.000 hours (6.3Years) at 38uA discharge  while producing 2.1Ah.   This should be a very good battery for this application.

Based on the data sheets I  ordered TADIRAN TL-4903/P batteries..
 
To get ready to replace batteries on 2 of mine HP 8591A spectrum analyzers we need to backup Correction constants that are stored in RAM.
First step is to get the manual and print out the section 9-2..9.6  "Backing up Analyzer Correction Constants"
In  worst case these could be lost if something goes wrong so worth having a hard copy (or Soft copy) to avoid expensive calibration.

Here are the calibration constants from the 2 units I have.  Maybe someone can use this info.

With this data I feel good about waiting for the battery. No more stress over lost Calibration .

[kawal]

Batteries came in today.  The packing on these was crazy .
These where made in July 2019 so now they only have 39 years 2 months of rated life left.   

[kawal]

After opening up the unit  and sliding off the cover it was time to remove the back.  Its not needed but easier to remove the GPIB card.
The old battery is best to be cut away and new battery soldered in its place.  There is a reserve capacitor that should be good for min 30min of backup but probably will last a lot longer.  I also removed the sampler card  to take some pictures but it is not needed.

[kawal]

Some more pictures. Unit powered on and after pressing the defaults button  the display showed the usual noise floor.
I connected the 300Mhz cal output to check if the calibration is still good and it is.  I saved the parameters either way so no problem if they where lost.

[kawal]

Did the battery replacement on the second unit. I tested both of old the  batteries and they seemed to OK and showing 3.6V. I attached a 63.4K resistor to each of the old batteries to get an estimate on remaining capacity . The first unit is holding stable at 3.6V after 2 days but the second started at 3.59V and dropped to 2.8V within 1 h. After 12h the battery is at 2.74V  so  its about depleted.   
Obviously the load of the resistor is  about 10x of the actual load but that is still close to losing data.  I would estimate that a few more weeks and the battery would have died and taken the calibration data with it.  I would say that I was lucky.

The second unit has more options installed but also has some interesting bodges. There is a few small boards installed on the processor board that where not present in the newer unit. 

[kawal]

The closeups of the bodge boards  (or options)

[Bicurico]

I took pictures of all calibration screens - found that to be quicker than writing down the numbers.

As a note: newer revisions will ha e a coin style battery so one shoild take a look before ordering the battery.

Regards,
Vitor

[kawal]

I took pictures of all calibration screens - found that to be quicker than writing down the numbers.

As a note: newer revisions will ha e a coin style battery so one shoild take a look before ordering the battery.

Regards,
Vitor

The calibration data depends on Firmware version .  On both of my units each frequency is displayed separately. I want to try newer firmware i found online from 1995 and 1998.  Mine are from 1990/10 and 1991/05 so the newer Firmware might change the way the calibration is displayed. 

[kawal]

Tried  Eproms from 1998 and the unit will not boot. Will try older ones

[kawal]

Tried 1994 Eproms that i found in Bin format and these did work. Now calibration data is  shown in different format.  After a short period of trying to restore the flatness data and getting 2.9Ghz in span setting (which is wrong) i decided to call it quits and reinstall original firmware.  Flatness data went in not problem and the self calibrations pass not issue.

[Rik]

Did you solder the new battery leads to the stubs of the old battery, and not lift the board?

[Rik]

I pulled the board, installed the new battery. Working fine now.