Links in the video description above to a 3458A NVRam/FRAM PDF I made up, WinGPIB where you can download it and the Gerbers/Schematics for all 3 of my custom Pcb's. Notes: There are some alternatives to FRAM / Dallas NVRam such as EEProm and MRAM, however, parallel EEProms are too slow on write and MRAM IC's are not available in 5Vdc versions from what I can see. Looks like the date codes on my Dallas rams are 2014, so 10years old already.
I am not sure if you put in significant effort adapting the pinout, but since it is a lone IC, you can mix up the data and address lines any way that makes the layout more convenient. The IC itself doesn't care what bits are stored where. As long as the DIP pins are mapped to like pins on the SOIC, external devices such as the programmer or meter itself can't tell the difference.
Tip = I no longer use manufactured board sockets... I use a pair of header pin socket strips, cut to length... This has advantages - I always have the correct socket on hand, because I cut whatever number of pins that I need. I don't have to inventory sockets. The strips leave the ends of the DIY socket open. Manufactured sockets are closed on the ends. With open ends, you can use tools underneath the chip, to get it out, or route wires through, etc... HTH.
I do replace the Dallas chips in some industrial machines, when they fail but most nearer 25 years, i guess it largely depends on how they are powered up. But yes a neat option.
Hello Ian, well done, and thank you for your new ideas about replacement of the DALLAS memory by FRAMs. That reminds me, that I also have to replace the DALLAS memory in my HP3458A from 2001, which has the same µP PCB version as yours. I did that already after a few years that I purchased my instrument, but that might already be over 10 years ago. No pressure, though, because I made a backup already, and I would always be able to fully calibrate by my LTZ1000's and resistor references, and I own a MJTC from NIST for the RF calibration. Anyhow, @10:15, you better would have stuck your screw driver through the slit in the case right to the memory U122, right into the slit on the small side of U122 and its its socket, so to lever it out by turning the screw driver. Then, U123 is accessible from the next upper case slit.
Yep, screwdrivers are not the best for IC removal but when they are this tight sometimes we have to. Ultra careful though, no marks on the Pcb. Actually, I was keen to test WinGPIB 3458A ram data extraction as this would be the first time I have actually proven it works by burning the data to U132, U121 and U122 and booting a 3458A.
The ones on my bench I showed I bought new from Digikey 5 years ago (kept as spares). I can't say for certain, but maybe they are still being sold DALLAS branded! Update: The datasheet shows both Maxim & Dallas names/logos.
For anyone else wondering why keeping A11 & A12 low will not mess with the address bus: They are not routed to the calibration memory U132 pins in the first place 🙂
Great video Ian - Thanks for the work on this. I had my 3458A in for a service from Keysight - I've never taken the cover off since they also did a calibration on it but they replaced the Dallas RAM as part of that service - Does anyone know if Keysight would have gone to the trouble of putting them in sockets or are they still making/servicing 3458A's with these soldered to the board?
Repair on Component Level is long gone @ HPAK. I guess, Keysight already replaced the whole A5 Controller Board. Nowadays, they would plug in the latest version, with the Snaphat batteries. Don't be shy to open your 3458A, that will not harm anything, apart from the 'Guarantee Void' seal.
@@DrFrank-xj9bc Thanks Dr. Frank - I really should take a look, I didn't open it up as I have the unit under a service contract at the time - It's out of that contract so I should take a look before I send it back to Keysight for cal and put it back under contract.
What is the original build date for your 3458A? It seems strange to me that you have an old through-hole style controller PCB but with a more modern Keysight branded front panel. My 2005 build Agilent branded 3458A already has an SMT controller PCB and I would have assumed that the later Keysight branded models would also use the same.
Mine was built in 2000, but I bought a new replacement front panel from Keysight UK not long after I got it. I still have the original HP branded front panel.
FRAM uses a ferroelectric layer to store data, relying on the polarization of a ferroelectric material. This mechanism is not directly influenced by magnetic fields..........they say.
Why FRAM? Why can't you use EEPROMs to replace the DALLAS ICs? The 3458A only reads the calibration data at power up, doesn't it? not constantly reading/writing from/to the DALLAS ICs like DRAM.
I haven't looked up more modern parallel EEproms, but for instance the MC28C family have a nice 150nS access time, however the write cycle time is on the order 5mS. So, whilst great for reading and treating them like ram they are probably not a drop in replacement due to the write timing. Might be worth spending more time looking up more modern EEproms though.
Thanksgiving SRAM! Correct 10:39 for others. The Number below NON; ie 1448A. So 48th week of 2014. I was looking at the date code and I noted you already posted it was 2014; Regards
Turned pin sockets with those headers are not the best, while they will grip well, due to the large pins, they are now ruined for regular DIP insertion, so best to never change those again. Did see either one of the pins on the cal socket was a little corroded, there by the bypass capacitor, have you checked that little capacitor is not leaking, and causing a future issue on the board. you can see at 13:51, pin 21 looking a little darker than the other pins. C154 might be slowly venting sulphuric acid.
No corrosion visible. I think it's just a trick of the light with that brand of IC socket. They have quite uneven/irregular holes, especially pin 21. C154 looks ok.
@@fredflintstone1 Unless the puller slips off, then you have lots of bent pins! In spite of using high quality pullers and taking care, this has happened to me more than once.
From my experience with DALLAS NV-RAMs , they never exhausted quickly as in datasheets so even after decades of backup power , they are still functional ... So I don't recommend this upgrade at all for DALLAS products ...
@@sdgelectronics No dear Steve ... It is not early data loss with DALLAS NV-RAM , unlike other brands or RTCs , RTCs may stop clocking for the duration time stated in datasheet but NV-RAMs are not.. I still have many of these DALLAS and still functional for almost 30 years since I got them by early and mid 90's ...
So now the tech exists it makes perfect sense to take out the Dallas devices, store their contents and use a brand new device. Peace of mind ! but you are saying don't do it so you're in fact saying wait for them to fail which makes no sense at all. One day you will switch your instrument on and get 'Ram Error' then see what sort of day you have.
Links in the video description above to a 3458A NVRam/FRAM PDF I made up, WinGPIB where you can download it and the Gerbers/Schematics for all 3 of my custom Pcb's.
Notes:
There are some alternatives to FRAM / Dallas NVRam such as EEProm and MRAM, however, parallel EEProms are too slow on write and MRAM IC's are not available in 5Vdc versions from what I can see.
Looks like the date codes on my Dallas rams are 2014, so 10years old already.
I was waiting for them to flip over and leave 2.54mm pitch holes in your thumb.
I wasn't expecting quality turned pin sockets......am so used to seeing cheap Ebay stuff!
Handy tip on replacing the old Dallas chips - cheers!
I am not sure if you put in significant effort adapting the pinout, but since it is a lone IC, you can mix up the data and address lines any way that makes the layout more convenient. The IC itself doesn't care what bits are stored where. As long as the DIP pins are mapped to like pins on the SOIC, external devices such as the programmer or meter itself can't tell the difference.
I adapted the address and data bus properly, main reason is that the FRAM ic’s need to have the option to be externally read/programmed.
A handy adaptation...got to keep these wonderful instruments in service...great work
It would be great to see you tear down (take apart with a dremel) one of the old DALLAS memory modules to see the internal construction.
It's not much to it, just SRAM chip and small 3V coin cell lithium battery. You can find some xray photos on internets.
Tip = I no longer use manufactured board sockets... I use a pair of header pin socket strips, cut to length... This has advantages - I always have the correct socket on hand, because I cut whatever number of pins that I need. I don't have to inventory sockets. The strips leave the ends of the DIY socket open. Manufactured sockets are closed on the ends. With open ends, you can use tools underneath the chip, to get it out, or route wires through, etc... HTH.
I enjoyed the video! You reminded me of several DS1230's and DS1213's I've had for decades! I just freed up a drawer in my parts cabinet!
I do replace the Dallas chips in some industrial machines, when they fail but most nearer 25 years, i guess it largely depends on how they are powered up.
But yes a neat option.
i've been meaning to do this for ages. thanks for the video!
Hello Ian, well done, and thank you for your new ideas about replacement of the DALLAS memory by FRAMs.
That reminds me, that I also have to replace the DALLAS memory in my HP3458A from 2001, which has the same µP PCB version as yours. I did that already after a few years that I purchased my instrument, but that might already be over 10 years ago. No pressure, though, because I made a backup already, and I would always be able to fully calibrate by my LTZ1000's and resistor references, and I own a MJTC from NIST for the RF calibration.
Anyhow, @10:15, you better would have stuck your screw driver through the slit in the case right to the memory U122, right into the slit on the small side of U122 and its its socket, so to lever it out by turning the screw driver. Then, U123 is accessible from the next upper case slit.
Yep, screwdrivers are not the best for IC removal but when they are this tight sometimes we have to. Ultra careful though, no marks on the Pcb.
Actually, I was keen to test WinGPIB 3458A ram data extraction as this would be the first time I have actually proven it works by burning the data to U132, U121 and U122 and booting a 3458A.
Added to the fact that Dallas became Maxim years ago, so any that are Dallas marked are at least a quarter century old.
The ones on my bench I showed I bought new from Digikey 5 years ago (kept as spares). I can't say for certain, but maybe they are still being sold DALLAS branded!
Update: The datasheet shows both Maxim & Dallas names/logos.
For anyone else wondering why keeping A11 & A12 low will not mess with the address bus: They are not routed to the calibration memory U132 pins in the first place 🙂
Oh yes, that would definitely cause problems…..😂
Great video Ian - Thanks for the work on this.
I had my 3458A in for a service from Keysight - I've never taken the cover off since they also did a calibration on it but they replaced the Dallas RAM as part of that service - Does anyone know if Keysight would have gone to the trouble of putting them in sockets or are they still making/servicing 3458A's with these soldered to the board?
Only one way to find out I reckon. I’d guess a socket was fitted……..please have a look!
Repair on Component Level is long gone @ HPAK. I guess, Keysight already replaced the whole A5 Controller Board. Nowadays, they would plug in the latest version, with the Snaphat batteries. Don't be shy to open your 3458A, that will not harm anything, apart from the 'Guarantee Void' seal.
@@DrFrank-xj9bc Thanks Dr. Frank - I really should take a look, I didn't open it up as I have the unit under a service contract at the time - It's out of that contract so I should take a look before I send it back to Keysight for cal and put it back under contract.
Wonderful job Ian =D
Nice chips swap with no data corruption ! A piece of chance. 😉
What is the original build date for your 3458A?
It seems strange to me that you have an old through-hole style controller PCB but with a more modern Keysight branded front panel.
My 2005 build Agilent branded 3458A already has an SMT controller PCB and I would have assumed that the later Keysight branded models would also use the same.
Mine was built in 2000, but I bought a new replacement front panel from Keysight UK not long after I got it. I still have the original HP branded front panel.
Nice upgrade, just keep magnets away. but are FRAM's susceptible to magnetic fields never used one to find out.
FRAM uses a ferroelectric layer to store data, relying on the polarization of a ferroelectric material. This mechanism is not directly influenced by magnetic fields..........they say.
@@IanScottJohnston I was just expecting the ferro bit to mean iron. 🙂 but it could also be close to ferrite.
Nice video... Great programing.
Front panel is in good nick too..
It’s a yr 2000 HP branded 3458A, I bought a new front panel 8 yrs ago.
Did you note What and When you modded the instrument in the instrument
Not in the instrument......I always forget those things.....:-)
consider using GPIB once again to verify
Good idea.
It's service time (again)! :D Well ... the 3458A is way oput of my reach. For now!
Why FRAM? Why can't you use EEPROMs to replace the DALLAS ICs? The 3458A only reads the calibration data at power up, doesn't it? not constantly reading/writing from/to the DALLAS ICs like DRAM.
I haven't looked up more modern parallel EEproms, but for instance the MC28C family have a nice 150nS access time, however the write cycle time is on the order 5mS. So, whilst great for reading and treating them like ram they are probably not a drop in replacement due to the write timing.
Might be worth spending more time looking up more modern EEproms though.
Update: I had a look for an MRAM product, but nothing available for 5V usage. Pity!
The date code of the memories pulled out seems to be from second half of 2014.
Thanksgiving SRAM! Correct 10:39 for others. The Number below NON; ie 1448A. So 48th week of 2014.
I was looking at the date code and I noted you already posted it was 2014; Regards
Nice Tipp.
You saying that CALRAM is fitted with a socket isn't actually correct. Original A5 has all three NVRAMs soldered permanently on the board.
My mistake…..oops!
Turned pin sockets with those headers are not the best, while they will grip well, due to the large pins, they are now ruined for regular DIP insertion, so best to never change those again. Did see either one of the pins on the cal socket was a little corroded, there by the bypass capacitor, have you checked that little capacitor is not leaking, and causing a future issue on the board. you can see at 13:51, pin 21 looking a little darker than the other pins. C154 might be slowly venting sulphuric acid.
No corrosion visible. I think it's just a trick of the light with that brand of IC socket. They have quite uneven/irregular holes, especially pin 21. C154 looks ok.
I am a bit suprised you don't have a chip puller but a nice video
I have always just pulled DIP chips manually, just feel I have finer control.
@@IanScottJohnston Yes but the puller pulls them out straight and square, but fine I have also levered chips out on occasions
@@fredflintstone1I’ll see if I can pick up a good one……always willing to try something new….:)
@@fredflintstone1 Unless the puller slips off, then you have lots of bent pins! In spite of using high quality pullers and taking care, this has happened to me more than once.
From my experience with DALLAS NV-RAMs , they never exhausted quickly as in datasheets so even after decades of backup power , they are still functional ... So I don't recommend this upgrade at all for DALLAS products ...
Well worth replacing early to avoid loss of data
@@sdgelectronics No dear Steve ... It is not early data loss with DALLAS NV-RAM , unlike other brands or RTCs , RTCs may stop clocking for the duration time stated in datasheet but NV-RAMs are not.. I still have many of these DALLAS and still functional for almost 30 years since I got them by early and mid 90's ...
Yes, I can confirm that.. they at least last for 15 years, and others have working memories in their 3458As after > 20 years.
So now the tech exists it makes perfect sense to take out the Dallas devices, store their contents and use a brand new device. Peace of mind ! but you are saying don't do it so you're in fact saying wait for them to fail which makes no sense at all. One day you will switch your instrument on and get 'Ram Error' then see what sort of day you have.
@@andymouse Exactly, everything works OK until it doesn't!
Watching the ICs getting pulled made me cringe, seen so many bent legs in my time. Get a good DIP IC extractor, worth the investment.