CMOS button batteries

[Clunkclick]

Been having trouble recently with starting up one of the home built Asus based desktops - keeps on halting the boot showing the mem OK light at red. Up to this date it had been running fine. When it stalled on-boot, hitting the mem-OK button on the MB enabled it,  after some time, to find a set of CMOS values it was happy to boot with. It then attempted to continue the boot and stalled (Mem-OK light off) but subsequent hitting of  the system warm-boot enabled the system to boot to Windows (10). It ran perfectly OK in Windows and was only experiencing difficulties with the BIOS.

 

Eventually, I established that the CMOS was loosing some of it values (Only the values which weren't default (Not video or main memory)) when the computer was switched-off - the reset CMOS values message was being displayed as the thing attempted to boot (But when I went into the system, the hardware detection i.e. disks, memory, video were being detected correctly and only minor items were being lost e.g. "Num lock off on start".

 

So I took out the CMOS battery, which is only two years old and  showing 80-90% full power and 2.8-2.9 volts and replaced it with another Duracell from stock (Same age and showing slightly poorer values) and fitted that. But it did not solve the problem - my understanding is that as long as the voltage is over 2.5 volts there should be no problem.

 

Nevertheless, today, I bought a pair of new button batteries  and fitted the best showing one of these (70-80% power, 2.8 volts) and the system booted without problem !!! (Remains to be seen whether the CMOS values will be retained after extended  closedown)

 

Both sets of batteries were tested on an Ansman tester (Designed specifically to test button cells and AAA to DD cells).

 

Why should there be such a different response of the machine to batteries which were clearly above the voltage threshold and which appeared to have an adequate charge ?

The pair of batteries (Same blister pack) I bought today were marked  that they were good to 2027, yet none of them showed 3 volts fresh out of the pack and one of them only showed 70% charge.

 

Further point, is it likely that the CMOS is faulty i.e. by not recognising acceptably charged and outputting batteries ?

 

Nick

[gadgetman]

Are you certain that your testers calibration is correct? 

 

A good button battery should be 3.2v.

 

I would say your tester isn't reading correctly and needs recalibration. 

[@Lee]

2 minutes ago, gadgetman said:

Are you certain that your testers calibration is correct? 

 

A good button battery should be 3.2v.

 

I would say your tester isn't reading correctly and needs recalibration. 

Was about to say the tester's ****ed but you put it more succinctly lol

[SuperbTWM]

Do yourself a favour and throw that tester in the bin

 

It does seem unlikely that the new batteries are duff but I would check them with a multimeter to make sure

Edited 10 May, 2017 by SuperbTWM

[CWARD]

Could also be corrosion on the batteries and/or terminals

[VWD]

So -question is what to test batteries on. I'd say , an average load, but since I don't have one for each type of battery I test, I use "engineering judgement( from over 50 years of electronic/electrical engineering expertise)", along with a high impedance voltmeter, and expect to see a "new" cell reading 10% above it's rated value, or else it's given the "bin" test . ( 10% , because, a high impedance volt meter will NOT draw current from the cell and show up any problems caused by internal resistance in the cell, due to old age etc). However ( with extreme care) a button cell can be given a new lease of life ,to show if replacing it is an option.

Corrosion would need to be bad ,due to the current in circuit, since volts drop is equal to product of current drawn times resistance in path. ( Simple OHMS Law).

[Clunkclick]

Yep, the Ansman (Self-calibrating) nis under reading compared with an old  Fluke 73:-

 

 Ansman now  says the old Duracell's are 2.62v and 2.94V and the new Panasonic is 3.05V (Room is warmer now).

Corresponding Fluke values are 2.9v, 3.13v and 3.23v.

 

Perhaps the PP7 on that needs changing.

 

The second new Panasonic battery is in the machine - presume that's better than 3.23v as it was higher originally.

 

Unfortunately, the new battery hasn't stopped the problem.

 

After an extended shutdown, it stlll halts with a red LED during boot and only resumes when the mem-ok switch is depressed.  After an extended shut-down followed by a restart, I'm finding that all the CMOS  values are reverting to default i.e. it hasn't preserved any changes from the last start-up.

 

As to cause, hardware conflict ? Short to case ? corruption of BIOS software ? Component failure ?

 

Taking the easiest first - hardware conflict.

 

I had noticed that when the add-on card for graphics (MSI, ATI R7 360) was installed part of the on-screen menu for the on-board BIOS program was being corrupted. Perhaps I'll  revert to integrated on-board graphics (ATI 4250) again remove the MSI graphics board to see what happens to the BIOS.

 

N

 

 

 

 

[gadgetman]

Sounds more like a memory or voltage issue. 

 

 

[VWD]

Q1- does battery tester test batteries under load ? If so , then that's the difference between Fluke and tester, as a Fluke will test the open circuit voltage .

[SuperbTWM]

3 minutes ago, VWD said:

Q1- does battery tester test batteries under load ? If so , then that's the difference between Fluke and tester, as a Fluke will test the open circuit voltage .

 Usually the only load on battery testers is the shunt across the needle mechanism which will be more than a multimeter, and probably nothing for say a AA cell but maybe a bit much for a button cell hence the lower values

[Clunkclick]

The battery problem appears to be solved, after multiple battery swops and, oddly taking out and re-seating all four memory DIMMS and wiping the CMOS.

 

The CMOS, after system power-up this morning, appeared to have retained all of its values after the power was switched-off overnight. I'm not counting my chickens that this will last though.

 

However,  when the system is switched-on after an overnight power down, its  still stopping during boot with a red "Mem-Ok" light. This can be resolved by depressing the adjacent "Mem-OK" button and then re-booting on the system warm- boot sub-button. When the system' s been powered-up for  some time, you don't then encounter a stalled boot and the red light if a warm boot is performed !!!!!!

 

I've also removed the ATI R7 360 gaming video card (From the PCIe x 16 slot) as it was causing video interference with part of the BIOS display - the part that related to the on-BIOS  BIOS update software ?????????

 

Never had any problems until I put Windows 10 on the system as a dual boot with W8.1. And it seems to have got worse since I removed Windows 8.1 ? Coincidence ? - As this Mboard's design is now 8+ years old, ASUS no longer support the drivers (Finished 2012/13), Microsoft now provides the Chipset support . . . . ?

 

I'm loath to think the problem is the power supply -  an 18 month old Corsair 750w bronze unit.

 

Nick

Edited 13 May, 2017 by Clunkclick

[Clunkclick]

Voltages on all the ATX rails  are reported by the BIOS as within 5% of tolerance for ATX standard 2.2 i.e. 3.26v (3v rail), 4.935v (5v rail) and 12.080v (12v rail). . .

 

. . . . but the thing still keeps stalling mid-way through the BIOS boot - most recently on USB drivers.

 

Nick

[gadgetman]

And you've updated the bios.....? 

[Clunkclick]

The BIOS is up-to-date as it can be - the last version ASUS issued, version 1801 (Dated 2013). But I will unload the old copy and refresh it if I can find a valid 1801 - at the moment the fresh 1801's I've downloaded from ASUS are being rejected by the BIOS's on-board update program - "Invalid module".

 

I think I have found the answer to the BIOS boot stall/MEM OK conundrum, but only an overnight power down will validate the theory.

 

Its a memory issue. Basically, about 18 months ago, in order to prolong the usuability of this 8 + year old system, I upgraded the on motherboard integrated Radeon 4250 graphics chip with a new MSI (Radeon R7 360) graphics board and added 2 extra 2GB Corsair DIMMs to sit in  the remaining two empty memory banks on the motherboard.

What I didn't realise was that the new ram, whilst almost  the same specification as the earlier ram (CMX4GX3M2A1600C9, 9-9-9-24, 1.65v) was  a later version, version 4.19 (As opposed to version 2.3) and was implemented with more capacious, bigger chips and implemented as SINGLE-SIDED DIMMS, whereas the original two DIMMS (Version 2.3) were implemented as DOUBLE-SIDED DIMMS.

 

This difference was  not stated on the external packaging, or on the labels attached to  the DIMMS themselves, and was  only visible on close examination of the units because Corsair DIMMS are manufactured with a plastic cover over the memory chips which is applied to both sides the DIMM whether its SINGLE-SIDED or a DOUBLE-SIDED DIMM - you can see the metal legs of the memory chips on both sides of the DIMMS on the DOUBLE-SIDED UNITS when you look closely, ditto a single side of metal legs on the SINGLE-DIMMS.

 

I only become aware of this after using the CPU-ID utility program which correctly reports memory DIMMS as either DOUBLE-SIDED or SINGLE-DIMMS. When cued into this difference I then consulted the Qualified Vendors List for DIMM Memory in the Motherboard manual and found that the Mboard only supported the DOUBLE-SIDED DIMMS of CMX4GX3M2A1600C9 type. Bugger.

 

Nevertheless, it will still boot to OS on SINGLE-SIDED DIMM, in fact I am typing this post with only one SINGLE-SIDED DIMM installed in BANK 1 and no other memory installed running the graphics from the Mboard chip and only using the Adata SSD and a wired network - all other plugs disconnected. But it stalls during the BIOS Boot and requires the "Mem-OK" button to pushed followed by the system warm boot before W10 loads. Further, during power down it trashes the stored CMOS values referred to above in this thread returning them to default and does the same to the fan "Cool & Quiet" setting - does it create a CMOS current leakage path during powerdown  ? By this mechanism it might also have been draining the CMOS battery and screwing the video output of the MSI board. Boom, Boom!

 

Anyway, just to verify, I tried the one of the original DOUBLE-SIDED DIMMS in BANK 1 and booted to OS no problem. Right as rain as were the CMOS values which were preserved.

 

This couldn't be any more cosmic in coincidence terms if it tried. This is exactly the same sort of problem I had with the Fabia battery and the leakage path created by a failed temperature flap motor while the car was engine off at rest - now runs with the Diagnostics fuse and one other removed (Except at MOT and servicing time) as I haven't been fit enough to remove the faulty flap motor and the main dealer and Indy don't want to know !

 

I think this is the answer, but only a couple of  overnight powerdowns in succession will prove it.

 

I think I'll have to start hunting around shortly for 2 x used DOUBLE-SIDED  Corsair DIMMS of the same type (I don't think the low density DOUBLE-SIDERS will still be available new) and flog the two SINGLE-SIDERS (Easier).

 

Postscript

 

I've just realised, the reason the this issue didn't get recognised until recently was that I had been keeping the system unit power on (Standby) all the time. Now, courtesy of  some of those radio-controlled power plugs to switch off everything at one switch mains power gets chopped at night.

 

What a bugger !

 

Postscript 2

This is probably old hat to most on this forum, circa 2010 (But new to me now, I may have read it back in when I purchased the MBoard back in 2008/9 and promptly forgotten it) - On researching  the differences between SINGLE-SIDED AND DOUBLE-SIDED DIMMS I've found that its even more nuanced than how many sides of the DIMM are populated with memory chips, its the way the memory controller read/writes to them. With DOUBLE-SIDED DIMMS the memory controller addresses each "Side" separately and switches between sides to do so, with SINGLE-SIDED it addresses all the memory chips as one entity. Consequently, DIMMS with chips on both Physical sides of the DIMM can be SINGLE-SIDED and DIMMs with chips only on one side can be DOUBLE-SIDED.

 

So, in this case, why nothing on the DIMM labelling/packaging to indicate this difference ?

 

Nick

 

Edited 13 May, 2017 by Clunkclick

[Clunkclick]

So, this morning, after an overnight powerdown, with the first slot (A1- Blue) populated with a SINGLE-SIDED DIMM (Which isn't compatible according to the Mboard manual), I push the "Go" button and the frigging things starts first time, without stalling or prolonged illumination of the "MEM-OK" light and boots straight into W10.

WTF - it shouldn't work ! Try this several more times, does it again and again. All CMOS settings are preserved !

 

So, let's try a second, the other SINGLE-SIDED DIMM, in the next memory bank slot to the right (A2 -black). Again, no issues ! When A2 was similarly populated last night the system was stalling during boot with the "MEM-OK" light on. I repeated the boot several times, both warm and cold, started every time without issue.

 

You guessed it, I thought  how about installing the DOUBLE-SIDED DIMMS in the remaining two empty memory bank slots (B1-Blue & B2 -Black). Expectations were confounded. Again, absolutely no problem booting.

 

Perhaps its significant, that one of the last things I did before turning the system off last night was to disable the SpeedStep function in BIOS. This function (The default is always on auto) applies in this case specifically to the operation of the Integrated Graphics Chip (Odd, I thought, that an Intel function like Speedstep, is included in an Award BIOS in an AMD MBoard-I suppose they had to if ATI had built in this functionality into the Integrated Radeon 4250 graphics chip). So the system started this morning with both the Graphics Speedstep and the fan "Cool & quiet" function disabled. I think Speedstep works by directing processor read/write  operations from the permanent storage (Whether SSD or HDD) to memory. That's got to be a candidate for causing the memory problems my system has been experiencing.

 

Pro temp, I leave the system, as currently set-up, for a few days just to see if its stable over several powerdown cycles.

Then, I might have a play to see if this set-up can tolerate engaging Speedstep.

 

As far as I can establish ASUS haven't updated the Memory Qualified Vendors List to sanction Corsair SINGLE-SIDED DIMMS in this MBoard (Although the QVL is content to recommend Crucial 4GB DIMMS (8-8-8-24, 1.8V) and Kingston 4GB (8-8-8-24, 1.65v) in SINGLE-SIDED Form) so, on that basis, I've ordered a pair of new (Yes New !)Corsair DOUBLE-SIDED DIMMS with exactly the same specification as the original two, including the version no . . 2.3. And, I will take it from there.

 

Very weird.

 

Postscript

 

Even weirder -the donateware CPU-ID software, running in Windows 10, is now reporting the following:-

 

Slot 1 -SINGLE_SIDED DIMM (Presume this is the Blue memory bank slot (A-1), far LHS of the bank)

Slot 2 -DOUBLE-SIDED DIMM (Blue memory bank slot (B-1), third to the right)

Slot 3 -SINGLE-SIDED DIMM (Black memory bank slot (A-2), 2nd to the right)

Slot 4 - DOUBLE-SIDED DIMM ((Black Memory bank slot (B-2), 4th to the right)

 

So the order of installation being reported by Windows is different to the Left to Right order of the actual physical installation.

 

Nick

Edited 14 May, 2017 by Clunkclick

[Clunkclick]

System still stalling on boot after an overnight powerdown, but this time no red LED "MEM-OK" light - had to warm-boot it again to get it started.

 

Still awaiting the arrival of the new DOUBLE-SIDED DIMMS, but I still suspect that a less than full capacity Li-ion button battery may have some role in this continuing problem, that's despite its capability of delivering a momentary 250mah when connected to the Flukes 10 amp discharge.  And it doesn't matter where you shop for new ones, either on-line or locally, new ones arrive partially discharged. Anybody know of a reliable source for good quality CR2032s ?

Also, I see from Amazon and E-Bay that there are Re-chargeable CR 2032s - denominated LIR 2032.. And whilst some of  these are advertised as suitable for MBoards, I note that the rated voltage is 3.6v and capacity 60mah (Compared with the CR2032s 3v and 250mah). Can the LIR's be used in an Mboard  and if so has anybody had experience of using them ?

 

Nick

Edited 15 May, 2017 by Clunkclick

[Clunkclick]

Ahh, very interesting . . . .

 

http://farnell.com/datasheets/1496885.pdf

 

This states this particular brand CR 2032 should be supplied new with a voltage of 3 - 3.4v, a "Terminal" voltage of 2v and an expected life delivering 3v or near 3v of at least 200 hrs and at least 1000hrs of life before it hits 2v.

 

Gosh, it could almost be the standard for healthcare delivery in the UK.

 

Postscript

 

Ah, The full banana:-

 

http://www.alldatasheet.com/view.jsp?Searchword=CR2032

 

What beats me is why the retail display stands and storage locations couldn't incorporate miniature pulsed chargers to keep these, and all batteries, as fresh as the day they left the factory. I'm fed-up paying out good pension money for half-discharged carp - and that includes car batteries.

 

Nick

Edited 15 May, 2017 by Clunkclick

[io1901]

Perhaps your problems are down to dodgy RAM which you only see in your original configuration and nothing to do with the CMOS battery. I've seen similar things in the past.

[gadgetman]

6 hours ago, Clunkclick said:

 

This states this particular brand CR 2032 should be supplied new with a voltage of 3 - 3.4v, a "Terminal" voltage of 2v and an expected life delivering 3v or near 3v of at least 200 hrs and at least 1000hrs of life before it hits 2v.

Which it will be. It's your tester which is faulty. 

[Clunkclick]

Quote

 

I have to disagree - see the readings I've just taken from three different meters on the three best batteries (No Load):-

 

Battery 1(Duracell from stock about 2 years old) - 2.88v (Ansmann), 3.054v(Fluke 73), 3.1v (Fluke 325).

Battery 2(Duracell from stock, about 2 years old) - 2.88v(Ansmann), 2.998v(Fluke 73), 3.0V(Fluke 325).

Battery 3(Panasonic, purchased this week) - 2.88v(Ansmann), 3.142v(Fluke 73), 3.1v (Fluke 325).

 

The average of these readings is 3.003v and the maximum range of the meter readings is 0.262v

 

whereas the difference (range) between the spec voltage of these batteries (3.4v) and the average of the meter readings is 0.397v - 1.51 greater.

 

Nick

 

 

[gadgetman]

But none of your meters are calibrated. 

 

There's a reason professionals get their meters calibrated, and you've proved it. 

 

Seriously, stop wasting time on this. The battery is fine. 

[StevesTruck]

RTFM and make sure your memory is installed correctly (check the pairs don't need to be in 1+3, 2+4 rather than 1+2, 3+4).

 

Download and boot off a Linux live boot dvd/memory stick (hirens is old, but good for this), when the boot loader comes up, you'll get an option to do a memory check. 

[RainbowFire]

3 hours ago, gadgetman said:

But none of your meters are calibrated. 

 

There's a reason professionals get their meters calibrated, and you've proved it. 

 

Seriously, stop wasting time on this. The battery is fine. 

 

By now you must have realised that Mr Roberts has far too much time, an amateur hobbyist knowledge, too much money (despite whinges to the contrary) too many old, broken and inept toys, coupled with the *need* to prove everyone else is wrong, and do go to great lengths to not do what people have been suggesting, all the time, every time. The fact that 3, uncalibrated, meters don't agree therefore means it *must* be the battery, surely gives you a clue as to how much of a wind up this realy is.

 

However, I am genuinely surprised that there's been no conspiracy theory as to it's all down to Microsoft or ASUS to force our poor, bewildered, Nick into having to buy the latest piece of kit from some faceless corporation.

 

I've got some popcorn, and donuts, here is you want some, it's been quite entertaining so far..........

 

 

[Clunkclick]

13 hours ago, Clunkclick said:

I have to disagree - see the readings I've just taken from three different meters on the three best batteries (No Load):-

 

Battery 1(Duracell from stock about 2 years old) - 2.88v (Ansmann), 3.054v(Fluke 73), 3.1v (Fluke 325).

Battery 2(Duracell from stock, about 2 years old) - 2.88v(Ansmann), 2.998v(Fluke 73), 3.0V(Fluke 325).

Battery 3(Panasonic, purchased this week) - 2.88v(Ansmann), 3.142v(Fluke 73), 3.1v (Fluke 325).

 

The average of these readings is 3.003v and the maximum range of the meter readings is 0.262v

 

whereas the difference (range) between the spec voltage of these batteries (3.4v) and the average of the meter readings is 0.397v - 1.51 greater.

 

Nick

 

 

Well, the Fluke 325, is only a year old, has only been used half a dozen times since then, and I would therefore hazard a guess that its still "On station" as far as agreeing with the factory calibration standards - i.e. good enough to be used as an index for home use . . . . blue touch paper lit.

 

Nick

[Clunkclick]

9 minutes ago, RainbowFire said:

 

By now you must have realised that Mr Roberts has far too much time, an amateur hobbyist knowledge, too much money (despite whinges to the contrary) too many old, broken and inept toys, coupled with the *need* to prove everyone else is wrong, and do go to great lengths to not do what people have been suggesting, all the time, every time. The fact that 3, uncalibrated, meters don't agree therefore means it *must* be the battery, surely gives you a clue as to how much of a wind up this realy is.

 

However, I am genuinely surprised that there's been no conspiracy theory as to it's all down to Microsoft or ASUS to force our poor, bewildered, Nick into having to buy the latest piece of kit from some faceless corporation.

 

I've got some popcorn, and donuts, here is you want some, it's been quite entertaining so far..........

 

 

*******s !

If consumer electronics, especially home computing, is so reliable and infallible, why are their pages and pages of reports on the internet on people not being able to get systems to start, systems failing-over etc - a good 50% of which don't get resolved. Is it possible, that white male middle class professionals aren't the oracles they make out ?

 

N

Edited 16 May, 2017 by Clunkclick