Defective HP 54504A --- NVRAM substitute?

[Wim_L]

I recently acquired an old HP scope that was going to be thrown away. Poor sampling rate (200MS/s), good bandwidth (400MHz). It's clearly defective, but may be fixable. It gives plenty of errors when booting, and fails several of the self-test procedures. However, using the appropriate magical incantations from the service manual (power cycling with a button pressed, then going through the full self-calibration procedure) its full functionality can be restored. Until the next time it's turned off, when it fails again. So it's fairly clear the NVRAM battery has failed. Now, for the fans of instrument pictures...

First a front view. Showing its age with some visible screen burn-in, and a missing knob. Standard shaft size, so it should be no problem to find a replacement for that, I'll see if I can find something with a large diameter and not too high. The metal box to the left is the power supply which I had removed there. All seems fine there, the voltages where it matters are still spot on. Electrolytic and tantalum caps all over the place in this scope, and yet no sign of obvious failures during a first inspection. I did clean some dust out, the fan pulls a lot of dirt in.

Next, after all calibration procedures have been performed. It can be used in this state, but the rather lengthy procedure makes it impractical to do that every time. It wouldn't be as much of a pain if it were used as part of an installation where it's always on, but for home use, its fan is a bit too noisy.

The insides then! After removing the back panel, the main board slides out through a slot in the rear part of the screen. It's a heavy construction, full of thick metal plates, though the scope isn't all that hard to lift. Fairly light compared to some older portable analog scopes even. An inside view of the scope front panel shows they made room for more connectors than are actually present on the front.

And the main board, top and bottom. Yep, that's gold plating all over the tracks. The whole thing is shiny yellow metal.

Final picture, the likely problem case! An NVRAM chip that, as far as I know, isn't being made anymore. DS1235YWL-120. Maxim still makes a similar chip, the DS1230 (in various versions, Digikey has a whole page full of variants). Looking at the datasheets, they look quite similar indeed, though which of the many variants would be the best replacement will require a closer look. Anyone who has done this before and knows which one will or won't work? The replacement procedure itself may also turn out to be tricky. It's not socketed (they probably figured the scope would be discarded before the battery ever ran out) and the pins are under the package so cutting them and desoldering one by one isn't going to work.

Finding probes may also be tricky, considering it only has 7pf input capacitance at 1MOhm, which is outside the compensation range of many cheap and some not so cheap passive probes. Of course, I could make some passive resistive divider probes to use on the 50 Ohm inputs, which probably is the best way to measure at the high end of the frequency range anyway. But conventional passive probes can be useful too at lower frequencies.

So, any suggestions on where to go next with this one? I do already have an 150MHz Hameg combiscope 1508-2.

[Rerouter]

top right, is that a speaker or a battery?

[Wim_L]

Top right on the mainboard? A speaker.

Having the battery separate from the RAM would be very convenient, but I'm not so lucky here... That kind of NVRAM module has the battery built into the IC package.

[Rerouter]

well then as an alternative if it really is shy of impossible to replace it, how many mA does it chew under normal operation, perhaps you could fit some external battery to maintain it, its dip so you could lift the power leg and run it that way as a worst case option,

[Wim_L]

That's possible in theory... With the disadvantage being the place to do that is buried inside the IC. It would involve drilling into the package and connecting wires to that. I'd take my chances with desoldering and trying the Maxim IC as a substitute before considering that ;-) (especially as misjudging the location of components inside the package means a risk of drilling straight into a lithium battery)

Oh yes, you can see a little bit of slightly dodgy stuff I did too... On the EPROMs, they were all exposed because the paper stickers had fallen off. So before I made the pictures, and not having good black tape left, I used scotch tape to stick black pieces of paper over the windows before taking pictures with flash.

[Rerouter]

i was getting at lifting the positive supply of the chips pin(s) and powering that with an externsal battery configuration

[amyk]

That reminds me of the Dallas clock chip used in early PCs.

Maxim says the DS1230 should work as a replacement:
http://www.maxim-ic.com/app-notes/index.mvp/id/202#q13

And others have done what you're suggesting with those chips too, dig in and replace the battery:
http://www.mcamafia.de/mcapage0/dsrework.htm
http://classic-computers.org.nz/blog/2009-10-10-renovating-a-dallas-battery-chip.htm

Another,  more permanent option is to use nonvolatile (no battery required) compatible SRAM like a CY9C62256 or STK16C88 (the latter says it's a direct replacement for the DS1230.)

[alm]

Anyone who has done this before and knows which one will or won't work? The replacement procedure itself may also turn out to be tricky. It's not socketed (they probably figured the scope would be discarded before the battery ever ran out) and the pins are under the package so cutting them and desoldering one by one isn't going to work.

Did you search the hp_agilent Yahoo! group archives or ask there?

Finding probes may also be tricky, considering it only has 7pf input capacitance at 1MOhm, which is outside the compensation range of many cheap and some not so cheap passive probes. Of course, I could make some passive resistive divider probes to use on the 50 Ohm inputs, which probably is the best way to measure at the high end of the frequency range anyway. But conventional passive probes can be useful too at lower frequencies.

The first thing I'd do is just try any probe I had. Compensation trimmers are not known for particularly tight tolerances, and I have gotten 8-12 pF probes to compensate on a 20 pF scope. For best signal integrity, especially at the higher end of the bandwidth, you should get the original probes. Probes with bandwidths above ~200 MHz are usually matched to the scope front-end, it matches the bumps in its impedance (which is usually not terribly flat for hi-Z inputs). Unless these are much more expensive than good alternatives, I see no reason to buy something else.

[rf-loop]

I have repaired several HP54501 - 54504 oscilloscopes just for NVRAM.

But not last 2-3 year so I do not know what are exactly available.

Time when I have done it there was Maxim NVRAM socket.

I take out old NVRAM
good quality 28 pin soket to mainboard.
Nvram socket to this
low stand by current C-MOS ram to Nvram socket.
+ for this construction is. Li batt voltage can measure.


IF you do not have professional tools for desoldering original NVRAM safely out.
(Remember - this main moard is easy fragile. With poor tools it is just easy to destroy it. 
Becouse material is soft and it is very weak for overheat! And some pind have middle layer high temperature sinking so soldering need enough power but not too hot. There in this kind of works it is easy to find what is professional Metcal or equal.  Try with cheap crap and after it mainboard is just piece of waste.)

It can also do with heat air gun! But....  this later

Some 5450x models ADC and some hybrids have very special connection to main board, do not shock - handle careful)

Then there is one "rural area method" if really have nearly only tools as hammer and hands.

Take wet thick newspaper and put it on wooden or other not slippery surface.. (it is cheap shock absorber so that shock do not so much travel to whole board)

Put mainboard bottom surface tightly over this wet 1cm thick newspaper. Push it tightly.
Put small soft things  between DALLAS and near other chips. Example strong carton pieces.
Take 10-15cm hard around 1cm diameter hard-wooden stick with the straight ends.
Other hand keep it tightly on tha DALLAS long side other end. Look that stick is as horizontal as possible.
Take (small) hammer. 

Hit with a hammer a very sharp high-speed attack. By weight of the hammer must not be so high as to cause excessive inertial forces. Impact-commercial energy is absorbed mainly in a good district DALLAS legs cut off. When a high-speed impact, but does not yet light shake more motherboard. The impactor may not be sloppy so that damage to adjacent circuits.
 The first shock is probably too low and your feet just warp.
 Hit the next attack in the same place on the side close to the other end.
 Probably the part where the legs off. Maybe half. The second blow to one end of the same side. Finally, you can use pliers to break off or to persuade.

If do this exercice 10 times it takes just 30 seconds and old DALLAS is in your hand and legs are nicely cutted without any damage.

I have also tested this.

Heat gun (these cheap  temperature adjustable airguns what can buy nearly every store in every place)
main board partially over board end bottom side up.
Fix also some kind of pliers to Dallas chip so you can pull it after it is hot.
Protect main board (bottom) with aluminium paper or make mask using example tin can sheet

Then heating. carefully! It takes time!  (not too hot - do not ry more fast with overtemp)
Patiently wait... hot...wait... until you see marks that solder is melting. (in this temp also main board this area is very elastic)
Wait, you can littlebit try drag Dallas down... after temp is right through main board  it come out easy, maybe it drops itself with pliers weight.

After DALLAS is out, and always care ESD!

Also these hybrids are fragile and specially some models hybrids golden connection pins)

Input attenuator moule. Do not open without high reason.
(oh yes also sometimes they can service but, even these special relays can open..... but it need really know what to and more important: what to NOT do!)

If need clean something. IPA is just ok. Or better 75% IPA where water component is also distilled. (distilled water is always better than ion changed)
(but do not go inside input attenuator before exactly know what are doing and why.)

One attenuator take many hours work just after some "clever" people have give lot of CRC "magick" contact aerosol to input BNC's. I ask why... he did not know why. (later he know, yes probe was broken)
After then he tell... I did not know it can not use... oh if you not know, why you do this if you do not sure know it can use.  But money is good teacher.
After I tell how much I want for repais he tell... oh now I know.

[mikeselectricstuff]

Drilling to get to the battery connections may be a viable option - chances are the -ve is grounded, so you only need to find the positive, which is usually the can outer.
Probably best to find another unit to investigate before attacking the real one - maybe take it to a local vet or dentist to X-ray.

[rf-loop]

Final picture, the likely problem case! An NVRAM chip that, as far as I know, isn't being made anymore. DS1235YWL-120. Maxim still makes a similar chip, the DS1230 (in various versions, Digikey has a whole page full of variants). Looking at the datasheets, they look quite similar indeed, though which of the many variants would be the best replacement will require a closer look. Anyone who has done this before and knows which one will or won't work?

http://www.maxim-ic.com/datasheet/index.mvp/id/2648

https://www.elfa.se/elfa3~eu_en/elfa/init.do?item=73-766-05

Maxim DS1230Y-120+
It is available and status is also in production.
Battery is activated after chip is first time powered Vcc > 4.25V.
So, also littlebit "old stock" can use if chip is never used.

Last time I have talked with one people, he ask my opinion, and after small inspection I confirm that this can use.
He did it himself with this component and all was ok.

Here is picture for some other solution (but pic is not good for introduce how to do it) This model was 54501A
There is socket, nvram socket, C-MOS ram and this was not with lowest possible stand-by current.  This total height is just so that mainboard can assemble and disassemble trough bacside hole.

[wkb]

I have hacked similar battery-backed RAMs that are found in older Sun computers.  MK48T02 or somesuch? [memory is dim here].

In any case: the Li cells are piggy backed onto standard DIL cases, the battery wires are run on the outside (short side) of the package.  Some careful surgery allows you to remove the resin that covers the battery terminals, remove the whole battery 'backpack' and solder a new 3V lithium cell onto the now freed RAM DIL package.  Some double-sided adhesive foam tape held the whole thing together.  I did like 10 of those, and they never failed.  You won't win any industrial design prizes with it, but it does the job. 

[Wim_L]

Thanks for the suggestions. A search of the hp_agilent Yahoo group did reveal a couple of similar situations, and similar fixes as suggested here.

I think my best course of action will be to do a replacement with a compatible DS1230Y module (thanks for the substitution rules page, by the way). I've considered battery-backup SRAM sockets might have been a better option for future maintenance, but the Maxim Smartsocket is out of production, doesn't have quite the same voltage specs as the Y-series parts, and Maxim recommends replacing it with the modules). Socketing an entire NVRAM module may not be feasible due to the clearance when moving the mainboard into the case anyway.

As the old one was only guaranteed to last for 5 years on an instrument that must be around 20 years old now, I figure it will probably last for the rest of its useful life.