Repair of Tek TDS3014B - need a source for a DS1742W NVRAM

[MarkL]

Well, it finally happened.  The DS1742W battery died in my TDS3054.  The scope was purchased in 1999, so that's 20 years.  Not too bad.  The symptoms were that the settings and SPC constants were retained, but the clock stopped when the scope was powered off.  The time when the scope was powered off was retained.

As mentioned in the above posts from two years ago, I purchased a backup DS1742W from an HK seller knowing this would eventually happen.

After removing the dead DS1742W, placing it next to the replacement made it quite obvious it was a fake.  See photos below.  I hadn't noticed the size difference before and it's significantly larger than stated in the Maxim datasheet.  I also noticed that there was no battery bump visiable on the bottom of the original.





For the record, the ebay seller was "crystal830303".

I soldered in a 24-pin socket and plugged in the fake.  To no surprise, the battery was dead in the fake.  It behaved in the same way; settings were retained, but the clock wouldn't run when the scope was powered off.

Since I had nothing to lose, I proceeded to cut into the old DS1742W.  Placing a strong, small magnet against the case allowed me to locate the battery near the top on the right hand side.  I sliced the top off with a razor saw, dug out the old battery, and replaced it with an external one.  Photos below.

It took a couple of hours and works great now.  Maybe that will hold for the next 20 years, at which point I probably won't care anymore.

[Jwalling]

Nice work!

That seller is still selling them. Looks identical to the one you bought.
https://www.ebay.com/itm/282116689792?

87 sold!
6 positive feedbacks on it. WTF?

    by 9***s
    During past 6 months
    Beijing to SanFran: 25days slow! Bag: anti-static. Foam, Bubble: not. Items OK.
    by e***o
    During past 6 months
    Fast shipping, excellent packaging, items as described, highly recommended!
    by z***2
    During past 6 months
    Good communication, Fast shipping, Great eBayer, Thanks!
    by z***2
    During past year
    Good communication, Fast shipping, Great eBayer, Thanks!
    by i***o
    During past year
    Ok
    by l***y
    During past year
    Got the order ontime and in good condition

[MarkL]

Nice work!

Thanks!

That seller is still selling them. Looks identical to the one you bought.
https://www.ebay.com/itm/282116689792?

87 sold!
6 positive feedbacks on it. WTF?

Fake reviews?  Or maybe they work long enough for people to give it a good review?  Mine sat around in the packing tube for 2 years.  With a real DS1742 with that date code, 2 years wouldn't have mattered.

Since I wasn't ever going to use it, I broke my fake one open with a sharp chisel.  Somewhat less finesse than my work in the previous post, but definitely satisfying.

The main chip was an unmarked 24-pin DIP with two extra connections on either end (never saw that before!).  One pair was connected to the crystal and the other pair to the battery (or was at least involved somehow with the battery).

It sat on a PCB with some resistors and one 6-pin chip (U1).  U1 was marked with "10A45".  I didn't bother trying to track down the code to find out what it was.  The main chip's 24 legs were extended through the PCB to the external pins.

There was nothing that identified who really made it.  Even the battery had no manufacturer name on it (except a small "SC" under the yellow wrap).  The voltage on the battery was 0.87V.

High-res photo below of the extracted pieces, if it's of any interest/help.

[f14]

seem in my case  I think I will mod external battery in old chip and don't need wait the new chip arrived . when new chip arrived I will test this  working or not 
btw I think a fake chip use too much current make internal battery runs out of power sooner . it is very usual in cheap chinese caliber the battery dead after some months even turn off

[Jwalling]

They spend quite a bit of effort to do this. I just don't understand why they don't make a product that friggin' works!
All those TDS3K scopes are reaching the age where they are going to need those chips replaced. Tek sold a ton of them, there's real money to be made -  you could easily get $30 each for a working unit...

[MarkL]

I can think of a few other possibilities for DS1742W replacements.

The first is to use the DS1744W.  The DS1744W has more SRAM than is needed and is in a 28-pin DIP package.  An adapter board could be designed which grounds pulls high the unused address pins and, in the case of the TDS3000 series, moves the package out of the way of the chassis.

The DS1746W and DS1747W are also possible, but are even bigger with 32 pins.  But it looks like they would still fit in the space available with an adapter board.

The DS1744WP could also be used similarly.  The package is a module which is meant to to be surface mounted, and then the battery and crystal are snapped on top after reflow.  It's more expensive than the DIP version, but maybe the battery cap could be removed and replaced after it was depleted.  I don't know for sure since I've never actually seen one in person.

Alternatively, the chips are exposed on the DS1744WP module and they could be removed and remounted onto a smaller board that would have the same dimensions as the original DS1742W, with of course a crystal and battery.

Unfortunately, I wasn't able to find any matches for the RTC chip just by itself that has the same register layout as the DS174x series.  That would be nice.  I'm guessing the RTC chip has an SRAM controller on it which handles the power down and addressing, so probably any SRAM would do.  Maybe ordering and examining a DS1744WP would reveal which RTC chip was being used.

Lastly, the access is slow enough for the -150 and -120 parts that maybe some really fast microcontroller could emulate the RTC and SRAM.  But that would take a lot more work, and probably get ripped off instantly by the counterfeiters.


The DS1744W and later parts are currently available, at least for the time being, according to the EOL/NRND spreadsheet at Maxim:

  https://www.maximintegrated.com/en/products/nla/index.mvp

Any of these solutions would probably end up being in the $40-$50 range (at least).

EDIT: Fixed unused pins description.

[MarkL]

I said:

...
The first is to use the DS1744W.  The DS1744W has more SRAM than is needed and is in a 28-pin DIP package.  An adapter board could be designed which grounds the unused address pins and, in the case of the TDS3000 series, moves the package out of the way of the chassis.
...

No one caught this, but since the clock registers are at the high addresses in memory, the unused (highest order) address pins need to be pulled high, not grounded.  Otherwise the clock registers would not be accessible.

Not sure anyone is interested in this option, but I didn't want to leave bad information out there.

[steve_01]

Hi,

Annoying as it IS but quite expected. I did order and recieved a blacktopped DS1742W-120!
It was a really bad blacktopped device, seems to be original part - but grinded away original marking!

Tryed to choose someone serious with low neg. feedback, but there is none I guess.

I have grinded off the CR1225 as MarkL did show on the original part. Thanks, havnt finalized it yet - but there is + and - solderable bits!

As suggested, there must be tons of TDS3xxx - out there with bad lithium batterys ; witch is a pretty good scope even today. I have the 3032-300MHz one!

Really isnt there a working replacement part yet!? Like MarkL is suggest - it should not be difficult to make a emulator using a fast microprocessor with DMA
and low RTC current. its very possible to make, but more difficult to get it *CE/FCC* aproved. Well, I guess Dallas got some big $ for disable all replacement
parts from Techtronics and many other manufactories. The registers are not the same on any other DS-xxxx device i have looked on, how annoying - and expected!

/Steve

[MarkL]

You might be interested in this series of posts:

  https://www.eevblog.com/forum/repair/tds3014-adventures-(seeking-75-75mhz-oscillator)/msg2733286/#msg2733286

Instead of just musing on what could be done, user james_s went to the trouble of actually making an adapter board for a DS1744WP module to replace the DS1742W.  The board files are up on github.  The adapter board is a much more eloquent route if you don't mind spending a little extra.

[steve_01]

Hi Mark!

Thanks for the heads up. Actually I did finallize the DS1742W-150 modified original with a CR2032. Should last for the rest of my life!
And you should have the credit for that rev-eng. digging. The dude on the other thread has access to a X-Ray .. that would be a nice tool!
Having astma - and many times did use nitric acid fumes to hack into such things.. 

And for the other thread and using a overprices Black Box Time Keeper, is not anything I would ever do. Dallas/Maxims weird prices on those is outrages! 

Looked on a suitable microcontroller with DMA and sleep-mode for RTC and 8K SRAM - with 10 microamps current in sleep mode, and a 110 mA lithium
should last more that 20 years. This processor is priced 1.5 USD (BGA) and I could do this very good on a simple board - and be even smaller than the black boxes for USD 25!?

I spend a lot of time doing stuff just like that above, just for fun and liking rev-eng. stuff. Done my whole life! Stupd - but fun!

We will se if I finalize such a board, i put the project in a pile of other to-do-maybe projects ..

[james_s]

I'm certainly not opposed to having alternatives to the Dallas parts, although the PowerCap versions of their products are not potted, they're exposed PCBs with a replaceable battery. It's true that they're not cheap as ICs go, but with these scopes regularly selling for ~$1k+ spending <$100 on a long term repair seems cheap to me. I see it as a permanent fix, buy it once and it should last the life of the scope, 10 years from now you can pop the cap off and solder in a new coin cell. I only have one TDS3000 scope so the adapter was by far the cheapest solution even if I only value my time at minimum wage.

[steve_01]

I'm certainly not opposed for you making those adapter boards. But like you said, its not sure they fit in any other devices using the DS1742.

For the supply of those expensive Time Keepers, Maxim have maybe desided to keep this device active or not!?

A bit confusing, stating the DS1744 is NLA - No longer active on this link (searching for DS1744) :

https://www.maximintegrated.com/en/products/nla/index.mvp

But active in this link:

https://www.maximintegrated.com/en/products/analog/real-time-clocks/DS1744.html
 
However, I did some quick timing test (on CE/OE) - just to see how fast my Techtronics TDS3032 was accessing the NVRAM.
And its about 250 nanoseconds, some strange timing makes it harder to emulate the DS174X - so I most certainly dont proceed with this project.

And when i bouht my TDS in the 90's it was prices around 5K USD , so its like you say $100 is not much money.

[Dave Wise]

To add information, it looks to me like the ST M48T35AV is compatible with DS1744W, and it may be cheaper in your area.

[MarkL]

To add information, it looks to me like the ST M48T35AV is compatible with DS1744W, and it may be cheaper in your area.

Close, but it doesn't appear to be a direct replacement.

Some of the registers are a little bit different, such as 0x7ff8.  In the M48T35AV fields 5:0 are for calibration and in the DS1744W they are the century value.

Also, in the M48T35AV some of the fields "must be set to 0" where in the DS1744W they are "don't care", so it's going to depend on what the firmware is doing to those bits (and if the M48T35AV really cares that it's 0).

[Dave Wise]

That's a pity.  Thanks for giving the spec a close reading.

[Dave Wise]

MarkL, since you're active today, when you replaced the battery in your DS1742W, did you do anything to preserve the data?  I'm facing the same problem with my TDS3054B.  Do we have a consensus as to what is stored?  (Please forgive, I read all the threads but can't remember!)  I believe it's ethernet MAC address, running time, and power cycle count.

[MarkL]

MarkL, since you're active today, when you replaced the battery in your DS1742W, did you do anything to preserve the data?  I'm facing the same problem with my TDS3054B.  Do we have a consensus as to what is stored?  (Please forgive, I read all the threads but can't remember!)  I believe it's ethernet MAC address, running time, and power cycle count.

I only have a TDS3054 (plain - not a B or C), so I don't know about the MAC address.  The most important thing to me was that I didn't lose the calibration data.  That was discovered to be stored elsewhere:

  https://www.eevblog.com/forum/testgear/tek-tds3000-series-and-dallas-nvram/

I'm pretty sure I read that the run information is in the NVRAM, but I didn't care if I lost that so I didn't really notice.

The progression of death for these time keepers is that the clock fails before data retention.  So, you might be able to put your old one in a reader and still get the data out of it.

You might also find this thread useful:

  https://www.eevblog.com/forum/repair/reverse-engineer-dallas-ds1742w/

[gek]

Jonathan at "the glitchworks" has created quite a few replacements for the Dallas IC's with removable batteries. 

https://www.tindie.com/stores/glitchwrks/items/

His products are well made, and sometimes an off-the-shelf solution is easier than grinding into an old Dallas module.

- Gary

[Dave Wise]

Thanks, MarkL, I have those threads bookmarked

In late 2021, what's the best-loved inexpensive 3.3V programmer that supports the DS1742W?  (I have a Needhams EMP300 but I don't think it does.)

Gek, glitchworks has indeed done a bunch of nvram adaptors, but not the DS1742W that we care about in this thread.

[sicco]

In https://www.eevblog.com/forum/repair/tds3014-adventures-(seeking-75-75mhz-oscillator)/msg3630182/#msg3630182 and beyond, it’s explained how you can edit the DS1742W in-situ. So while it still sits on the TDS3000 scope motherboard. Read/write without even opening the enclosure.

You need a FT2232 or FT4232 USB module. And a PC. Maybe the Molex 100 pin connector. But 1.6 mm edge connector also works. Or else open the box and solder ~7 wires on the motherboard 100 pin expansion connector.

Alternatively: I did a TDS3000 plugin board with the DS1744W on it, edge connector, plus the FTDI chip. If you want to go all the way then a PC program (it’s in that post also) lets you edit the TDS3000 flash chips so that the NVRAM RTC is shifted from 0x02800000 to 0x05000000 - with that the old DS1742W /CE pin changes role with a CE pin on the expansion connector - and that’s enabling the new DS1744W for you.

Likely less expensive as sourcing a 24 DIP 3V3 programmer 

[Dave Wise]

Hmm, the idea that a standalone programmer might be the more expensive approach never occurred to me.

1. 100-pin option board with comm and NVRAM, and remove or work around the old NVRAM
2. Remove NVRAM, install socket and new NVRAM

Your trick of modifying the firmware to change the address is inspired, but it creates a pitfall for the next owner if he installs new firmware.
I lean away from that, so for me both options lead to opening the box and removing the old NVRAM.  Socketing a new one feels like the path of least resistance.  I'll shop for a programmer and see if that makes me change my tune.  Or just buy that module kellyjo_13 makes.  (I'd buy from cuebus but I couldn't find his offering.)

[james_s]

I have hand programmed the DS1742 using toggle switches for the address and data and a pushbutton for the chip enable, it's slow but works fine for writing a few bytes. I eventually wired up an arduino mega and used that to program a different Dallas NVRAM, never got around to properly building the firmware though.

[Jarek]

Hi
My TDS3032 (not B and C) had a damaged DS1742W.
The oscilloscope did not boot.
I bought the M48T02-120 (it's readily available and not expensive).
Unfortunately, it is 5V, but it is easy to program with the TL866IIplus programmer.
the batch file is here on the forum (for DS1742w)
You only need to connect + 5V to 24pin.
The photos show how I did it.
The oscilloscope has started and the date and time can be set.
For now I don't know what is not working. I will test a few days and see if everything is OK.
greetings

[Dave Wise]

Jarek, although this is a simple procedure, I suggest that it is a bad idea.  Every time the scope's microprocessor reads the M48T02, on the one hand the board drives "high" address lines with 3V which is poor noise margin.  And the M48T02 will drive the data bus with 5V which exceeds the expected 3V.  This may eventually damage silicon in the scope.

[MarkL]

The M48T02 has a slightly different register layout than the original DS1742W.  Register 0x7F8 is the calibration register on the M48T02 but is the century on the DS1742W.  If the scope writes to this register, it can affect the accuracy up to +4.068  to -2.034 ppm (+5.5 to -2.75 minutes per month).

Perhaps it's not enough to worry about, or maybe the scope doesn't write to that register at all, but just be aware it's not quite a drop-in replacement.

I would also echo Dave's concern about driving a 3.3V bus directly with a 5V part.  The output driver might be pretty weak looking at the datasheet, but generally it's not a good combination without a level translator.