Dumping and restoring the CAL RAM of an HP3457A

[0xfede]

By the way, you can find the schematic for the newer revision board here:
http://xrms.free.fr/divers/multimetre_HP_3457A_service_manual_reassembled.pdf

Ἄκουε τοῦ τὰ τέσσαρα ὦτα ἔχοντος

[Nuno_pt]

Hi 0xfede, can I use this to read the RAM < https://github.com/Twilight-Logic/AR488/blob/master/Contributed/artag/README.md > < https://www.eevblog.com/forum/projects/ar488-arduino-based-gpib-adapter/msg2718346/#msg2718346 > or do I need a know good GPIB.

Thanks for all the hard work

[Tryer]

I had that version already. But many thanks.
At the end, it is a big advantage being able to read.
Looking more carefully, i found in the right end corner of this pdf(p.228)
Rev.A
8-68A/8-68B

I hope the last line 8-68B means Rev. 8-68B

I will write a small routine (Borland C)
like:

for all calram bytes
{
  do
  {
    POKE 48,00 // may be 2-3 times that line
    poke adrs,byte1
    poke adrs+1,byte2
  } while peek(adrs) not as just written
  adrs = adrs+2
}

The switch must have been activated previously.
Of course this will be tested in the RAM area first.
Or shall i wait till the battery quits its service?
It itches anyhow.
 
Ἄκουε τοῦ τὰ τέσσαρα ὦτα και δεκατέσσαρα ὄμματα ἔχοντος **
** today (due to inflation? :-)) ) one says:"έχε τα μάτια σου δεκατέσσερα"
meaning something different namely "be carefull"

[Tryer]

@Nuno_pt
Any GPIB adapter (functional) will do.
It must be only capable to communicate with the 3457.
Can you communicate with any GPIB device using that interface?
I am very tempted to build this device. Have you built it?

[Nuno_pt]

@Hi Tryer, I'm still waiting on the boards.

I've some GPIB cables here and a few megas, so I can give a shot at disassembling a GPIB cable and plug it to a mega also.

[wiesl]

Hi, I tried to make a dump on my 3457A that uses the firmware and the board that I attach in photos, I use the PROLOGIX interface and HP3457A Memory Dumper.zip (Thank to 0xfede), I often have difficulty doing it with win10-64b where it goes wrong when saving the file , with win10-32b it makes the rescue regularly. The multimeter has the firmware 03457-88810 Rev2 (If someone interested I have the .bin of the eprom).
I noticed that if I do more 'dump and compare the content is different, only some parts match, do you have any advice to give me?

Hello,

Got my 3457A today, looks like it is in very good shape. Includes calibration certification until 10/2020. 10V voltage reference around -39uV off (9.99961V). Agilent 34410A shows exactly 10.00000V calibration valid until 10/2017. Which one would you trust?

Ad Cal Dumping: I experienced similar problems dumping the 3457A with HP3457A Memory Dumper.zip (Thank to 0xfede). I use the PROLOGIX interface under Win10-64Bit. Didn't try Win10-32Bit yet (no machine).

Serial number SN#: 2703A06XXX
Battery voltage of Saft Lithium LX1634 (looks original): 3.0328V
EPROM: 03457-88810, D/C 8825
Main Controller: 03457-66511 Rev A 88809F
Input Signal Conditioning, A/D Converter, Inguard Controller: 03457-66502 Rev C 88809F
AC Converter: 03457-66503 Rev B 88809F
Reference voltage board: 03456-66525 Rev B 88809F
Keyboard board: 03457-66505 8809F Rev A

Guess EPROM D/C 8825 means datecode year 1988/week 25, right (according to the chips datecodes and serialnumber/productrevision that looks ok)?

Cal Dump #1 has 451 bytes, Cal Dump #2 has 453 bytes.

Any ideas?
Something special to configure in the Prologix Setup?

BTW: Is it normal that the 4 screws of the Main Controller are missing (so it is not screwed)?

@0xfede: Any chance to release the source code of the dumper?
@ipqobk: I'm interested in the firmware 03457-88810 Rev2 dump. Whats the latest one for Rev. A boards?

Thnx.

Wiesl

[Tryer]

Hi wiesl,
There is a metal cover over the main controller with 4 screws, which fix both the cover and the controller.
Are you missing that cover?
As for the length of the CALRAM it is exactly 448 bytes (0x1FF - 0x40 + 1)

[wiesl]

Hi wiesl,
There is a metal cover over the main controller with 4 screws, which fix both the cover and the controller.
Are you missing that cover?
As for the length of the CALRAM it is exactly 448 bytes (0x1FF - 0x40 + 1)

No, cover and screws are there. Was too late yesterday, didn't think about that 

Yes, expecting exactly 448 but get more.

Wiesl

[wiesl]

Hi wiesl,
There is a metal cover over the main controller with 4 screws, which fix both the cover and the controller.
Are you missing that cover?
As for the length of the CALRAM it is exactly 448 bytes (0x1FF - 0x40 + 1)

Could read the calram (hopefully). Will post details later.

There is a forth interpreter in the HP3457A. Protection mechanism is discussed in the patent, too. Maybe that helps for a software solution.
See: https://patents.google.com/patent/US4747048
Forth interpreter discussed also at: https://www.eevblog.com/forum/repair/hp-3478a-how-to-readwrite-cal-sram/msg1185473/#msg1185473

Wiesl

[Tryer]

Attached a small script to read the CALRAM of the 3457.
It works for me with win7 64HE and the AR488 SW on an arduino Mega256 board.
Tested with Octave 5.2

[wiesl]

Could read the calram (hopefully). Will post details later.

Could read it hopefully.

I guess I found out how we can write to CALRAM in software only:

First of all an overview of the main controller board revisions:
- A1 Board: 03457-66501 Rev A
  - RAM: 0x4800 to 0x4FFF
  - NVRAM: 0x5000 to 0x57FF
  - Battery 3.0V
  - two address decoders are wired to A11~A15
- A11 Board: 03457-66511 Rev A (ERC<2850)
  - I own that board
  - CALRAM 0x40 to 0x1FF
  - Battery 3.0V, R644=17.4k, R645=13k
    - works well even with 3.6V battery with this resistor values
  - two address decoders are wired to A1~A5
- A11 Board: 03457-66511 Rev A (ERC>=2850)
  - CALRAM 0x40 to 0x1FF
  - Battery 3.4V, R644=13k, R645=12.7k
  - two address decoders are wired to A1~A5

Ad ERC:
- Does anyone know what ERC stands for (Equipment Revision Code)?
- Is the ERC sticker the one with "2749" or the one with "09-2839 Made in USA"?

How can we write to CAL-RAM (untested) in software only:
Signal OTL#: I/O Port OUT2 (0x3C) Bit7:
- OTL# positive Edge: resets counter U631 via U633A and U633B
- OTL#=1: sets U635B to 0 and therefore resets U635A and pulls NVE# low (as long as OTL#=1!!)

Wiesl

[Wim13]

I have a 03457-66501 the first model, and here are my results.

From the service manual section 8-40:

the CAL-RAM U511 is on this model on addres 5000 - 57FF for reading and is 2 Kbyte
The 2 Kbyte is also used for other data that has to be saved, so not only cal data,
cal data is only 512 bytes. U511 is in my unit a Toshiba TC5517APL

The unprotected write adresses are different for U511
in this case the manual says : write on 1000 - 17FF and read on 5000 - 57FF.
The other RAM U506 has same adress for read and write: 4800 - 4FFF.

The secure cal data should be on the high part of the U511 RAM
and are in the last 512 bytes: 5600 - 57FF (Read)
and write on adress 3E00 - 3FFF ( 512 bytes)

To check changes in U511 i did a dump 3 times, from the 2 Kbyte U511.
the second dump a day later and the third after a autocal.

I noticed, see attachment for the data, changes in the lower part of the RAM,
but not in the high 512 bytes location.

The last byte on U511 looks like the cal number, 35.

With the PEEK command you get 2 bytes in return, so for 2 Kbyte
you get 1024 lines with 2 bytes of info.

The Battery change: i took the A1 board out, connected 2 penlites (=3V)
by a diode to R557 for the moment, and changed easy the lithium backup batt.

[Wim13]

To make the dump complete, from the older device: HP 3457A, type: 03457-66501 Rev A.

here are all  binary dumps from:

the work RAM 2Kbyte, U506
the Cal Ram 2 kbyte, U511

The ROM 8 Kbyte, (TMM2365p) , U503
The EPROM 32 Kbyte (AM27256), U502, version 03457-88803

If some one has a copy of version 88804 or 88805, pls leave a dump.

Note, in the work ram you can see some of the peek command that were used.
 

 

[wiesl]

Which software version does your meter have?

[Labrat101]

Hi All
I have also just acquired a fully functional HP3457A
And I find this really interesting . I also got a dyeing Battery going to replace it soon.
Installed a Temporary patch battery 4.5v connected to the VMvcc line ( + diode & Resistor )
Holding the Cal Safe .

Great work

[Labrat101]

Hi All ,
I have reserved my dying battery that is date code 1992 . and is now holding around 3.26 volts.
with a 4.6v backup battery via diode & resistor connected to the  NVvcc  Node. so it does not interfere
with the Old 3v battery which is still good enough to keep the 3v ref to the Watch Dog.

My NVram is marked as Rev .2  U-602.  This is also displayed from the front panel command REV? 6,0
Being Not Programming minded at any level  .. Hardware I am OK on
Can one of you smart guys tell me can I do the cal dump on this NVram as per photo .
using the code that 0xfede very kindly put up.  .
and If Yes .Does the code have to be changed or altered in some way so it can be put back into the NVram ?
Sorry for seeming stupid but Programming and software is not my thing . Hardware I stick to Old School.

so cut a long story short I really want to backup my Cal as this hp3457A has a cal sticker 09/08/10.
All the reading are smack on compared to a 5 digit Instek and I am not worried if its a ppm out.
But to re Cal these 3457A costs $400 plus and the local cal lab said they can only cal to 6 digits ONLY
So my 3457A maybe better than there's  .

So any advice here I would really appreciate . I do have a plan "B"  and that's just to fit a 4.2v lithium rechargeable on NVvcc node and have it trickle charging while powered on and have a LED low voltage warning LED on the front panel and just keep the watch Dog happy with a coin 2032 battery. which is also probably enough to keep the NVram good for another 5 years or more if there is no other Hardware failures. Which will last me to My past expiry Date 

I also noticed that J614  looks like a Programming Plug as its marked RD & WR , 5v , Grd.
Marked in red .

Thanks .   

Update :: Adding the 4v plus but keeping below ideal 4.8v (5v) was recommend by another EEVblogger .
 Thanks.  Many thanks to "pqass"   

[Labrat101]

Hi Again,
After banging my head on the wall a few times I found a solution .
First thanks to  @ 0xfede for all the fantastic work and with the switch .
but I got bogged down with the software part .
I found an other EEVblogger  @ IanJ   has published a beautiful piece of work and a very nice
Program that works out the box even a non software guy understood it .

https://www.eevblog.com/forum/metrology/3458a-logging-via-windows-app-revisited/?all

it also has a Cal Read which I tried and it dumped the 448 bytes of cal memory with no problem
with my HP82357B . There is all the info on the link above  .

I have add the Dump bin  which looks very similar to some of the other cal dumps .