3478A FRAM modification

[wolfalex]

Hello,

Because I removed the battery inside my 3478A for internal shipping purposes I have already lost its calibration constants and this gave me the opportunity to implement an idea I had already
for a longer time. I really disliked the idea of buying a new lithium battery for my meter and to install it, especially because I know that there are much better options nowadays. This is when I decided to
modify my 3478A meter to use FRAM (ferreromagnetic RAM) instead of the original battery-backed up SRAM solution. I have never used FRAM before, and I thought this is the perfect occasion to learn and try out something new. I thought I add some information about this project here if somebody is interested

I do not know if somebody tried this before (or bothered to do so) or if my 3478A is now unique in this perspective

I added a page to my website showing more information about this modification. But I will also add some information here under this topic.

 http://wolfalex.bplaced.net/te/hp3478a/fram/hp3478a_fram.htm

Please keep the following in mind: Removing the battery, shorting the RAM supply voltage only for a split second or doing anything else shown on this page will result in lost RAM calibration constants. This will turn a still valueable and useful meter into a less valueable and useful one! Also handle the lithium battery contained in the 3478A with care. Never short the battery and be aware of the fire hazard. For disposal of batteries always follow safety guidlines and the industry/government rules/guidlines in your state.

I would be very happy if some of this information is useful or interesting for somebody. Just let me know what you think.

[wolfalex]

Attachted is the glue-logic circuitry I needed to interface the used FRAM device FM18W08 to the 3478A internal databus. Also attached is the TL7705 reset generation circuitry which is added, because the original 3478A reset topoglogy stops working when the battery is removed.

I also added two pictures of the initial test which still uses the breadboard glue-logic circuitry

[wolfalex]

Attached are two scope pictures showing what the glue-logic is all about. The first picture shows the original unmodified SRAM multiple read cycle timing diagram. The second picture shows the same scenario, but this time at the FRAM pins. The FRAM needs a falling edge for every read/write cycle to internally latch the adress pin information, and therefore this falling edge has to be generated by the additional logic gates.

[David Hess]

The issue with CS needing to be toggled was discussed over on TekScopes@yahoogroups.com but the Tektronix guys lucked out; so far the oscilloscopes where the NVSRAM was replaced directly with FRAM worked fine because CS is toggled between accesses anyway.

[David Hess]

I was thinking about doing this as well, but after seeing all of the extra circuitry needed, maybe not.

I wonder if my 3458A can use non-volatile RAM chips [of some sort like FRAM or MRAM or ...] without all of the extra logic to toggle the `CS line...  It would be great to never have to worry about batteries again...

It will be able to use the EEPROM based NVSRAMs made by Cypress.  They just look like a normal asynchronous SRAM.

[wolfalex]

Sorry for my late response. I was following your responses on my phone more often, but I never found time to sit down on my computer to write a response. My initial plan was to just replace the SRAM with the FRAM device, and I thought there should be no problem with the /CS signal. Unfortunetly this was not the case, and so I added the additional logic. I just wanted to get it done and I had the logic parts laying around in my parts box, so it was just additional work

The last thing I have to do is to install the PCB somewhere in the meter. I hope I find a way to mount it without looking awful at the end and spoiling a otherwise nice looking meter.

@David: I was not aware that Cypress is building NVSRAM replacement parts based on EEPROM technology. Do you maybe have a part number? It would be interesting to look at it.

@DiligentMinds.com: I wanted to note that I have not spent any effort to optimize the glue-logic circuitry... it could be that someone finds a better circuitry to do the same For example from the XOR device I only use one out of the 4 gates. The only fully used device is the OR gate.

[David Hess]

@David: I was not aware that Cypress is building NVSRAM replacement parts based on EEPROM technology. Do you maybe have a part number? It would be interesting to look at it.

Simtek was the original manufacturer but Cypress bought them some time ago.  The parts I used are marked SIMTEK STK16C88-WF45 (32k x 8b) but Cypress discontinued production of most of the through hole parts some time ago.  Some of the parts require a small external holdup capacitor.  At various times this technology has been refered to as SONOS, AutoStore, and most recently PowerStore.

Unfortunately the horrid Cypress web site prevents directly linking to most of its content so this is the best I can do.  The product selector link toward the top right will list the available parts:

http://www.cypress.com/products/nvsram-parallel

[wolfalex]

Hello David,

 thank you very much for the Link. I looked at the datasheet of a part, very interesting. I was not even aware that this kind of device exists.

 Alexander

[David Hess]

Have you actually used this in a 3458A to replace the Cal-RAM?

No, I used these to replace the Dallas/Maxim DS1230AB NVSRAMs in my 2440.

They operate exactly like JEDEC standard asynchronous SRAMs because that is essentially what they are.  They are not a drop in replacement for U512 which is a 512 x 4 SRAM but they will not require extra logic.

[wolfalex]

I finally installed all components on a veroboard and connected everything again with the meter. I did not want to spend time drawing and ordering a nice PCB for the modification of only two meters. With the downside that my solution will never look professional and looks more like a big mess.

The last question was how to install the new FRAM PCB inside the meter. Unfortunetly with the size of the PCB (using veroboard and throughhole components) and the space restrictions in the meter I was not able to find a nicer solution. The PCB was put into a box to prevent shorts and then "installed" in the empty space inside the meter. Because I know I will never sell my meter this solution is good enough for me right now Functionality is my number one priority.

[bingo600]

Nice solution.

But those leads are looonnnggg , could cause trouble on some systems. Where a DSxxxx is also used as "real Ram"

/Bingo

[wolfalex]

Thanks. It seems until now I got away with the long leads, and I have not yet seen problems with it. I always get a Self Test OK message, and also the meter seems to work fine. But you are right, the possibility is there to cause problems. The original installed RAM device was called u5101L I believe.

[Tektron]

wolfalex, thank you for helpful idea  !
3 weeks FRAM works in my HP3457A. No additional parts was used for memory (only wires and cutter  ), just TL431+3 resistors for watchdog.

[bingo600]

wolfalex, thank you for helpful idea  !
3 weeks FRAM works in my HP3457A. No additional parts was used for memory (only wires and cutter  ), just TL431+3 resistors for watchdog.

Schematic & writeup please 

/Bingo

[TiN]

In the works ETA This week, together with other interesting article  .

[rastro]

I finally installed all components on a veroboard and connected everything again with the meter. I did not want to spend time drawing and ordering a nice PCB for the modification of only two meters. With the downside that my solution will never look professional and looks more like a big mess.

The last question was how to install the new FRAM PCB inside the meter. Unfortunetly with the size of the PCB (using veroboard and throughhole components) and the space restrictions in the meter I was not able to find a nicer solution. The PCB was put into a box to prevent shorts and then "installed" in the empty space inside the meter. Because I know I will never sell my meter this solution is good enough for me right now Functionality is my number one priority.

wolfalex,

I'm thinking about duplicating your design but I just can't source a box of Bio-Schoko-Beeren.

[Tektron]

rastro, if say seriously, you can remove 74HC14 and use still unused parts of 74HC86 instead inverters.

[wolfalex]

Hi,

   @Tektron: You are very welcome I can't wait to see your schematic and solution.

   @rastro: If everything fails I can supply you with a box of Bio-Schoko-Beeren When I have finished my second 3478A I will have the hard choice again which box to use 
   Maybe this time I will also try to keep the wires a little bit shorter.

   

[Tektron]

Schematic & writeup please 
/Bingo

I can't wait to see your schematic and solution.

Google translated. For professional translate we need wait, until this week will ended 

[Miti]

I know some people hate necro posting but this is relevant and why start a new post when I want to talk exactly about this project?
I just found this project and I thought it is interesting so I made a quick schematic and board.
Is there any interest in this still? I still have one 3478A and I love it. I installed an easy to replace CR2030 battery but I want to try it nevertheless, particularly because I have the FRAM. I installed a socket today in preparation for the adapter board. The board may need some polishing and then I can post the Kicad project.

Cheers,
Miti

[Miti]

And here it is, no more battery junk in my HP3478A meters

[Miti]

And here are the final JLCPCB fab files.

[Miti]

I found an issue with the supervisor chip trip voltage. Somewhere between the old version of 3478A with aluminum frame and the new version with steel frame, HP changed the diode CR500 from Schottky to silicon therefore the forward voltage drop is higher. My first install was on an old frame and it works well but on the new frame it holds the reset line indefinitely due to lower supply voltage on U512. I recommend you select an MCP101 with a lower trip voltage, use MCP101-450 or 315.

[wakawaka]

Hi Miti,

nice board! I've come across a unit with 3478A without even an SRAM chip and thought i'd take the opportunity to get rid of its battery as well. (no wonder it's uncalibrated)

Was wondering if you could make the kicad files available, it would help with modifying footprints for availability as JLCPB inhouse parts.

Best,

Waka

[Dan N]

Nicely done mod.  I have a simple change for new builds that will increase endurance 8x.  Might be overkill, but the meter does constantly read the SRAM.

Because the FRAM is organized into 8 byte rows, a read of any byte causes a destructive read/write of all 8 bytes in the row.  Reading 8 consecutive bytes causes 8 destructive read/writes of the same row.

Putting only 1 value on each row eliminates this issue.  Simply use address lines A[10:3] and connect A[2:0] to gnd.