HP 3457A NVRam Battery change.

[Labrat101]

Hi All,
I know this subject has come up before but I could not find a final solution  .

I picked up a very nice deal on Ebay for a HP3457A
That Has a Cal sticker on it 10 years ago.
I powered it up all the Test pass OK and the self Acal  work Ok .
No errors . WOW this has to be too good to be true .  Left it on the bench running for a few hours .
after about half an hour an almighty BANG .  The Mains line filter had decided to leave the room ..
the metal tin was burst open and all the sticky smelly stuff had oozed out .
Did a clean up and replaced it with a temporary socket  .
All started up fine No Damage other than the house stunk of burnt plastic .


I checked the Ram battery which is a Panasonic BR-2/3A  with a date code 2344 . which is by my reckoning  1992  3 day of week 44    27 years it still measured 3.36 volts but I know that these batteries can show good and then have a rapid drop off . instant death .
The only Battery I had available  was a CR123 Duracell (3.260 v) 
So I have a Battery operated isolated soldering Iron and I soldered 2 Patch leads to the board and connected the new battery as a piggyback  with a Schottky diode IN5819
as it has a very low voltage drop 0.26v compared to the diode on the hp board which has 0.6v drop apx .
So if the old battery does do a dirty on me . the CR123a hopefully will save the Calibration .

Has any one done some thing similar as the BR-2/3A are a commercial cell and are far better but at the moment the post is hopeless and some of the stuff from Ebay
is taking way longer . OK its lasted 27 years and I have seen these last 30 + years but the drop off can happen within a hour . And relying on Murphys Law   
Is there another alternative way of doing this ??
So any help on this would be appreciated .

Thanks

[EHT]

I did roughly the same as you, and soldered in both a temporary battery pack and then a new BR2/3A. I've left the old one in situ as i thought it probably easy to end up messing up the Cal RAM when disassembling the unit and desoldering the battery. I realise now I made a mistake of just using 1N4148s and not low reverse leakage diodes. I also later found out that there are better substitute batteries.
At some point I'd like to dump the Cal data and cut out original cell. New one is on a flying lead.

[Labrat101]

I did roughly the same as you, and soldered in both a temporary battery pack and then a new BR2/3A. I've left the old one in situ as i thought it probably easy to end up messing up the Cal RAM when disassembling the unit and desoldering the battery. I realise now I made a mistake of just using 1N4148s and not low reverse leakage diodes. I also later found out that there are better substitute batteries.
At some point I'd like to dump the Cal data and cut out original cell. New one is on a flying lead.

I did notice even using a Schottky diode that there is a small leakage of a few micro volts and the original BR2/3a has dropped from 3.39v
to 3.31v  maybe just natural decay 27 years thing .
I did also wonder about using a regular 4.2volt rechargeable and use 2 diodes in series to drop the voltage to 3.3v apox .
and have it trickle charging @ about 10ma off the 5v rail during power on.
 The NVram needs just over 2v to keep it alive.   
I have always wondered why HP did not use a different type of battery and why 3v, that seems to be the standard .
For sure the HP techs are far more smarter than I am . so there must be a good reason .
I also noticed that there is a 4 pin jumper marked as RW* , RD*, Gnd , 5v (J604) on the new model & (J504) on older model.
its marked not used . this looks like a connection to read & dump cal .. I maybe totally wrong as I have no idea on how to program.
I am very temped to but just a 4.2v battery at the junction of CR605 & C603  older model (CR505 & C503) CR605 Schottky diode .
the voltage at this point 4.89 v . so if this is held at 4.2volt when OFF .
As this rail drops to the battery voltage minus diode loss + 200   = 2.8v.
Or does the watch Dog check these voltages or something else ??

I have ordered a new BR-2/3A  battery from Ali so it may arrive in the next 1yr .

[Labrat101]

Sorry I made a Big typo on the Part Numbers  Above ..
Looking at the Wrong Diagram year     .

There were small differences 
The Battery voltage was increased to 3.4v  but the component values seem to be the same even on my 92 build.
 My model has 94 stamped on the back but serial and chip codes say 92 and so does the sticker on the PCB .

[pqass]

I did roughly the same as you, and soldered in both a temporary battery pack and then a new BR2/3A. I've left the old one in situ as i thought it probably easy to end up messing up the Cal RAM when disassembling the unit and desoldering the battery. I realise now I made a mistake of just using 1N4148s and not low reverse leakage diodes. I also later found out that there are better substitute batteries.
At some point I'd like to dump the Cal data and cut out original cell. New one is on a flying lead.

I did notice even using a Schottky diode that there is a small leakage of a few micro volts and the original BR2/3a has dropped from 3.39v
to 3.31v  maybe just natural decay 27 years thing .
I did also wonder about using a regular 4.2volt rechargeable and use 2 diodes in series to drop the voltage to 3.3v apox .

I did a similar operation on my HP3478A.  However, I used a series resistor+1N4148 diode on the output of my alternate battery before placing it in-parallel with the existing battery.  See: https://www.eevblog.com/forum/testgear/first-bench-multimeter-fluke-8840a-vs-hp-3478a/msg3008694/#msg3008694 and https://www.eevblog.com/forum/testgear/first-bench-multimeter-fluke-8840a-vs-hp-3478a/msg3015428/#msg3015428

In my case, I paralleled my alternate battery+resistor+diode with NVVCC node and ground; ie. same node as CR636 cathode (on your diagram in your last message).  However, in my case, I don't have a trace going to a LM339.  If you do it my way, rather than directly paralleling the old battery, then there is no requirment to keep the alternate battery voltage at just below the old battery voltage so as not to risk charging it.  Besides, the LM339 appears to use the old battery as a 3V reference from which it uses to make sure the +5UNREG node gets up to at least 7V before enabling other parts; ie. some type of power-good signal. 

My alternate battery was 3AA cells between 4 and 4.5V.  It doesn't matter exactly as long as it's less then 5V and higher than 3V.  In fact, I used the higher 4V to my advantage.  If I measured ~3.5V (> old battery voltage) on the NVVCC node, then I knew that my alternate battery was connected properly and therefore could snip the old battery out before soldering its replacement.

and have it trickle charging @ about 10ma off the 5v rail during power on.
 The NVram needs just over 2v to keep it alive.   
I have always wondered why HP did not use a different type of battery and why 3v, that seems to be the standard .
For sure the HP techs are far more smarter than I am . so there must be a good reason .
I also noticed that there is a 4 pin jumper marked as RW* , RD*, Gnd , 5v (J604) on the new model & (J504) on older model.
its marked not used . this looks like a connection to read & dump cal .. I maybe totally wrong as I have no idea on how to program.
I am very temped to but just a 4.2v battery at the junction of CR605 & C603  older model (CR505 & C503) CR605 Schottky diode .
the voltage at this point 4.89 v . so if this is held at 4.2volt when OFF .
As this rail drops to the battery voltage minus diode loss + 200   = 2.8v.
Or does the watch Dog check these voltages or something else ??

[Labrat101]

pqass
In my case, I paralleled my alternate battery+resistor+diode with NVVCC node and ground; ie. same node as CR636 cathode (on your diagram in your last message).  However, in my case, I don't have a trace going to a LM339.  If you do it my way, rather than directly paralleling the old battery, then there is no requirment to keep the alternate battery voltage at just below the old battery voltage so as not to risk charging it.  Besides, the LM339 appears to use the old battery as a 3V reference from which it uses to make sure the +5UNREG node gets up to at least 7V before enabling other parts; ie. some type of power-good signal.

My alternate battery was 3AA cells between 4 and 4.5V.  It doesn't matter exactly as long as it's less then 5V and higher than 3V.  In fact, I used the higher 4V to my advantage.  If I measured ~3.5V (> old battery voltage) on the NVVCC node, then I knew that my alternate battery was connected properly and therefore could snip the old battery out before soldering its replacement.

Thanks that makes perfect sense  .
 That was the point I was referring to but mess up on the numbers ..
 I had all ready scratched a check point at that location on the NVVCC line . But I was not sure about the 1k going to the LM339 and what this was actually doing .
 I measured the power off voltage on pin 28 on my NVRam  (U608) and it sit @ 4.858volts and drops to the old battery voltage which is now 3.2846v
so the leakage on the paralleled battery had drain it a bit .. not good.
 So I am changing it to your way .. Diode + 200 then the Pin 28 vdd on the NVram will be kept @ 4v . and I hope when I power down the LM399 wont give
and error . But the Cal will be safe . 
 The HP3457A has a lot of differences to 3458. 
I did check the data sheet for minimum data retention 2v is the safe . but can go lower . Not going to test it.  TC5564APL-15 .

BTW There is a miss print on the Diagram there are 2 R648's  The one that is 1K should read R646 and the other R648 10k on the Base
of U637C .
  Always have a backup Meter to repair a meter 

I just will have to wait for my new Battery to turn up . but as long as I have the cal safe I will be happy the local company wanted Starting price for CAL $400
 That's twice what I paid for the unit . ..
 I would be interested in your project with gpib but I am really no good with programming  . I even messed up on Plink 

Thanks for all

[pqass]

pqass
In my case, I paralleled my alternate battery+resistor+diode with NVVCC node and ground; ie. same node as CR636 cathode (on your diagram in your last message).  However, in my case, I don't have a trace going to a LM339.  If you do it my way, rather than directly paralleling the old battery, then there is no requirment to keep the alternate battery voltage at just below the old battery voltage so as not to risk charging it.  Besides, the LM339 appears to use the old battery as a 3V reference from which it uses to make sure the +5UNREG node gets up to at least 7V before enabling other parts; ie. some type of power-good signal.

My alternate battery was 3AA cells between 4 and 4.5V.  It doesn't matter exactly as long as it's less then 5V and higher than 3V.  In fact, I used the higher 4V to my advantage.  If I measured ~3.5V (> old battery voltage) on the NVVCC node, then I knew that my alternate battery was connected properly and therefore could snip the old battery out before soldering its replacement.

Thanks that makes perfect sense  .
 That was the point I was referring to but mess up on the numbers ..
 I had all ready scratched a check point at that location on the NVVCC line . But I was not sure about the 1k going to the LM339 and what this was actually doing .
 I measured the power off voltage on pin 28 on my NVRam  (U608) and it sit @ 4.858volts and drops to the old battery voltage which is now 3.2846v
so the leakage on the paralleled battery had drain it a bit .. not good.
 So I am changing it to your way .. Diode + 200 then the Pin 28 vdd on the NVram will be kept @ 4v . and I hope when I power down the LM399 wont give
and error . But the Cal will be safe . 

If you first add the alternate battery+diode+resistor to the NVVCC line, I think the worst of what would happen if your original battery falls below 3V is that upon boot, the watchdog circuit won't give the positive power good signal and the 3457A would fail to boot.  But your cal data would be safe since the alternate battery would keep the RAM above 3V.

 The HP3457A has a lot of differences to 3458. 

I wish!  I have a 3478A; not a '58A.   

I did check the data sheet for minimum data retention 2v is the safe . but can go lower . Not going to test it.  TC5564APL-15 .

BTW There is a miss print on the Diagram there are 2 R648's  The one that is 1K should read R646 and the other R648 10k on the Base
of U637C .

I see that now. 

FYI: that whole CR634 thru U637D circuit is just so the NVVCC line gets pulled up as close to VCC as possible to make up for the drop across CR636.  Talk about gilding the lily!   My 3478A doesn't have that either.

  Always have a backup Meter to repair a meter 

I just will have to wait for my new Battery to turn up . but as long as I have the cal safe I will be happy the local company wanted Starting price for CAL $400
 That's twice what I paid for the unit . ..
 I would be interested in your project with gpib but I am really no good with programming  . I even messed up on Plink 

Thanks for all

Not my project; See the AR488 project here: https://www.eevblog.com/forum/projects/ar488-arduino-based-gpib-adapter/ (follow the GitHub link within)     All you need is an Arduino UNO and the right GPIB connector.  WaveyDipole and friends has done a bang-up job.

My 3478A has the magic read/write cal data command but isn't documented in the service manual.  It was discovered by MarkL and friends in this thread: https://www.eevblog.com/forum/repair/hp-3478a-how-to-readwrite-cal-sram/
I'm not sure it would work on your 3457A or if anyone else discovered a similar way to extract cal data.

Best of luck.

[Labrat101]

Quick update I did it as per  pqass suggested used 3 AAA Energizer batteries I only had that size holder .
I used an ST4148 diode & 200 Might make it 470
In OFF state the NVVCC  Rail sits @ 4.6volts and only draws a few Micro Amps which is less than the natural decay of these Batteries .
And the old battery is holding now stable .

Thanks all seem to be good .