Fluke 6060B replacement NVRAM battery

[caiser01]

Just picked up a very nice Fluke 6060B Synthesized RF Signal Generator on ebay a couple of weeks ago and I'm really liking it already. Apart from some scuffs on the top, it's in excellent cosmetic condition. Date codes on the ICs and power transformer indicate this unit was built in 1991.



Did some quick tests using a scope, some radio receivers, and an RTL-SDR and it seems to be working very well across it's entire operating range from 10kHz to 1050MHz. This unit includes the optional OCXO as well, which is quite nice, and the GPIB interface.

The only flaws I can find are:

• One of the rubber shock mounts for the exhaust fan is broken. The other two look a bit fatigued so probably worth replacing as well.
  I've already located a couple of replacement candidates at McMaster Carr so that repair shouldn't be a problem:
  https://www.mcmaster.com/93945K31/
  https://www.mcmaster.com/9378K23/
• The NVRAM battery is dead; not surprising after 30+ years. I get an error code 000 000 002 000 at power up which the manual says is the "Non-volatile memory test".

To confirm #2 above, I removed the bottom of the case and the bottom shield inside. This was a bit painful not just because of the number of screws involved but also because it meant breaking the factory seal sticker.

For the record, the factory seal sticker on the top is still intact so both seals were present when I received this unit; no one's been in here fiddling about in the last 30 years.

I'm the first person to see under the cover since 1991!

The suspect battery is on the "N-V Memory PCB", a Saft LS3 3.5V Lithium cell. Measuring across this battery shows a reading of 0.038V so definitely no good anymore.

As far as I can tell from the manual (https://assets.fluke.com/manuals/6060B___imeng0000.pdf) (see page 293 of the PDF for the NVRAM board), the purpose of the NVRAM is to allow you store preset configurations of front panel settings (nothing to do with calibration which is thankfully stored in EPROM). For my use case, this isn't a critical feature but with this being such a good condition unit it would be nice to get everything working 100%.

I did quick look around online for batteries. Looks like there are some similar cells available but I wanted to get some opinions here first.

Can anyone recommend an appropriate replacement battery for the NVRAM?

Are there any significant risks to replacing this battery? Again, this unit otherwise seems to work perfectly.

Is there anything else I should address while I'm digging in? Power supply caps are Nippon Chemi-Con. Should they stay or go?

Any advice appreciate.

Thanks!

[wn1fju]

Virtually any battery will work.  There is already a series diode, so you don't have to worry about the power supply attempting to charge it under normal operation.  Heck, I've replaced these things with two ordinary AA-batteries in a sealed plastic holder (to prevent damage when they eventually leak).  The CMOS RAM only needs about 2V for data retention so as long as you are comfortably above that (for example, 3V), it should be OK. 

There is no inherent risk to replacing these batteries unless you don't know how to desolder and solder.  No need to keep the 6060B turned on during the swap since you've already lost the memory contents. 

As far as the electrolytic caps go, I would leave them alone unless there are obvious signs of damage like bulging, wet spots on the board, dry white crusty material, etc.  You might, however, check the AC ripple on all of the power rails just to make sure. 

[caiser01]

@wn1fju

Thanks for the advice.

I ended up ordering an LS14250 from Newark: https://www.newark.com/saft/ls14250-ax/lithium-battery-3-6v-1-2aa/dp/40C1847  This looks like an equivalent battery also made by Saft who made the original. This one is a 1/2AA just like the original and has the specs for longevity so hopefully I won't have to dig back into this thing for at least another decade. It's a tight squeeze in the little compartment for this NVRAM board once you get the shield on so I think sticking with the 1/2AA size is best, though I realize functionally it's not critical.

Yeah the electrolytics all look fine; nothing bulging or crusty. Given that everything seems to be working well, I suspect the filter caps are alright for now but I will check the supply rails for ripple as you suggest. Wouldn't want to lose something critical over a spotty cap.

Got some new rubber mounts for the fan on the way from McMaster-Carr as well. After everything arrives and I get it installed, I'll report back with pictures of the process.

Gotta say, I really appreciate the way this Fluke 6060B is designed. Lots of standard TTL parts and discretes. Even those rubber mounts for the fan are still obtainable. Looks like this unit should be serviceable for quite a while yet.

[wn1fju]

Yes, the 6060B is designed very well.  Everything is readily accessible, although if you have to pull a board out, it can be a chore.  I thought half of my unit was missing when I first opened it up, because there are two layers of boards, back-to-back, and they are only a few inches tall inside the much larger outer case.  There was some sort of connection between Fluke and Gigatronics with these signal generators - the equivalent machine is the Gigatronics 6061A.  Mine had Fluke markings on the boards and Gigatronics no-tamper stickers on the cover plates.

Good luck with yours.  This reminds me that maybe I should take a look at the battery in mine.

[caiser01]

Well, I received all the parts I ordered for my Fluke 6060B a few days ago and got everything installed this evening.

First up were the new rubber mounts for the fan. Both versions I found at McMaster-Carr are compatible with the 6060B but the machine screw part is a bit too long. It must be cut shorter or else it will interfere with the fan blade. Not a big deal though and a few minutes with the hacksaw got the new mounts to the correct length. The first picture shows a comparison of an original blue rubber mount with the new black rubber mount both cut and uncut. Second picture show the fan and bracket with the new mount threaded in. Note that while they thread in fully they do not protrude through the nuts. The fan spins freely and there's no vibration noise, just a bit of white noise from the air moving.



Next I tackled the battery replacement which turned out to be fairly simple. Lay the unit on its top, remove the bottom half of the enclosure, and the bottom inner shield (lots of screws!) as I showed in the photos in my first post in this thread. The NVRAM board is inside its own walled pocket in one corner. There are five screws holding the NVRAM board in place. Remove those and you are now ready to remove the board. On the side of the board opposite the battery is a 2x16 row of pin headers. Near this is a plastic loop. Place a finger through the plastic loop and *gently* but forcefully pull straight upwards. You may have to wiggle and rock the board a little but it will come loose. After the board is out, you'll be able to see the two single row connectors from the controller board protruding through a port in the metal shield.


Let's take a look at the back side of the NVRAM board:


Note the battery sits in a cut-out in the board. Also note the two rows of connector pins and make sure they were not bent or damaged when the board was removed.

To remove the battery, first desolder its two leads using your favorite method. I used my Hakko FR-301 de-soldering gun and it took less than a minute to desolder the battery leads. The black strap holding the battery in its cut-out in the board is a bit stretchy. I *gently* pushed one end of the battery out of its cut-out to get the corresponding lead out of the through-hole pad as shown. Once out, bend the lead straight. Repeat for the other end of the battery. When both leads are free, the battery can slid out from black plastic strap.


Place the old and new batteries side-by-side and trim the leads of the new battery to match the length of the old.


Reverse the procedure described above to slide the battery into place and solder the leads. Make sure the polarity markings on the battery match those on the board! The new battery I used had a light longer button on the positive side but still fit just fine.


Before reinstalling the NVRAM board, I checked voltages. Between the negative side of the battery and the pin 24 of the SRAM chip, there was 3.444V and across the battery itself was 3.678V.

I then reinstalled the NVRAM board making sure to carefully align the pins on that board with sockets on the controller board before applying gentle pressure to reseat it. Then I secured the NVRAM board with its five screws, reinstalled the bottom shield with its many(!) screws, and reattached the bottom of the enclosure.

Powering on for the first time with the new battery, I still got the error code for the NVRAM. This is normal for the first power up because the RAM is still empty. On subsequent power cycles, there were no error codes, and the unit now retains the last front panel settings while powered off. The new battery is doing its job!

All in all, it was a pretty straightforward repair procedure and took far less time to do than it did to write up for this post!

While I was inside my now fully-functional 6060B, I checked the power supply rails for DC voltage and ripple. Everything was within spec according to the power supply troubleshooting section in the manual so I'm going to leave the original caps in place for now. Since it's easy to do, I'll probably just inspect the caps and re-check the supply voltages annually to head off any future trouble there.

I also tested the GPIB interface with the USB->GPIB adapters I built courtesy of the most excellent AR488 project (https://www.eevblog.com/forum/projects/ar488-arduino-based-gpib-adapter/msg3810788/#msg3810788). If you're not familiar with this project and you have some GPIB equipment, you should check it out!
https://github.com/Twilight-Logic/AR488

Anyway, I discovered that my Fluke 6060B runs V5.0 firmware, whatever that's worth. I also discovered there is a GPIB command that is supposed to return the number of hours the unit has been powered on since it was manufactured. However when I tried it, the response I got was "000000000.5" (half an hour?) so it would seem the power on hours count is stored in NVRAM and is lost when the battery dies or is removed. Judging by the light amount of dust on the inside surfaces, I suspect my unit may be low usage.

Hopefully, this documentation of my NVRAM battery replacement adventure will be of help to someone else in the future. This very nice Fluke 6060B is now ready for full-time service in my home lab!