DIY: new cell for NVRAM chip

[joger]

hi,
I've my falcon bugging at boot with 2 or 4 bombs sometimes 3 or all... I thought it was the NVRAM that I changed with the Dallas 12887A (Thx Maxxim for your cool way of samples! ) as you advised above, following all instructions with a socket too ...

But now... its the same...and I cant boot with floppy as mine is dead and the other from an old 286 I got I cant remember how to set it (0 or 1?) I got an 1443MB HD I formatted for my falcon and I'm on...the off!... please help!

[wongck]

hi,
I've my falcon bugging at boot with 2 or 4 bombs sometimes 3 or all... I thought it was the NVRAM that I changed with the Dallas 12887A (Thx Maxxim for your cool way of samples! ) as you advised above, following all instructions with a socket too ...

But now... its the same...and I cant boot with floppy as mine is dead and the other from an old 286 I got I cant remember how to set it (0 or 1?) I got an 1443MB HD I formatted for my falcon and I'm on...the off!... please help!

Try to reset the NVRAM by pressing the SHIFT-CTRL-ALT-UNDO keys.... may be less or more keys (it is mention here in one of the replies...)

[joger]

Try to reset the NVRAM by pressing the SHIFT-CTRL-ALT-UNDO keys.... may be less or more keys (it is mention here in one of the replies...)

Hi, I've already done this... it doesn't work neither...
Any other idea? I think about the RAM but I dnt know how to test it, I dnt have any other RAM neither...

[wongck]

I think your repair procedure did something wrong. When I had my NVRAM problem, it did not boot because it read some rubbish and does not know what to do. A reset using the keys clear it and it was able to boot.
Your falc now is booting but bombs, so it is trying to boot may be some wrong address by the repair procedure make it bomb.
You should have just put the power across the NVram, like I did, which is 2 solders only....

[joger]

You should have just put the power across the NVram, like I did, which is 2 solders only....

please join a pic or explain again how.. maybe can help

[wongck]

You should have just put the power across the NVram, like I did, which is 2 solders only....

please join a pic or explain again how.. maybe can help

No need to join a pic.
It is already on the web, just go to my web site. URL on my sig.

[joger]

You should have just put the power across the NVram, like I did, which is 2 solders only....

please join a pic or explain again how.. maybe can help

No need to join a pic.
It is already on the web, just go to my web site. URL on my sig.

Yes sorry I saw it but already sent the post... thank you I'll try and tell about...

[simbo]

Hi all

i looked at this mod
and decided to add a recharging circuit using a BD140 as a constant current source
of about 30-40ma you can use any 100ma sink current pnp transistor like bc557
youll see how the design works

when the falcon is powered the +5v rail runs the nvram and also charges the battery @up to 5v @ 45ma
via the falcons 12v rail when not powered the batteries take over
becouse the batteries will dictate the voltate due to there loading the charger
the voltage wont exede 6v {well 1.2v per cell {more like 1.37 or so when full} X4}
if the batteries fall below 6v then the recharge will allow 40 or so miliamp charge
and the cells will stay put at 6v

this is a preliminary design
and has NOT been tested yet..!and is purly posted for any view and any errors people spot
im making a pcb first for my falcon ill callibrate the R1 current limit to suite
it may turn out to be another value than 15R this is an educated guess
i need to build the circuit add the batteries and measure the current avalible
i recond 10 -45 ma @ ~4v is fine to charge four cells every so often
with nicads etc its the current that matters
as long as its limited By R1 its cool and infact this circuit will output about 6v@45ma
i think this is a better idea for the falcon than batteries that run out and cause boot up problems
and maybe adding a 6.2V clamp zener to the battery pack is a good idea

credit for the diode switching
{ive used this method many times to replace the dodge switches on psu socket on ham radios}
and the 'clear' thinking to realise how the actual ic will use its own internal battery
and resolve
http://www.geocities.com/SiliconValley/6880/phsw.html
a few posts back


pcb board size is 17 mm X 17 mm

its always better to snap off the old cell
'most' of the nvram chips
have the battery in a plasic case mounted above the actual chip
so cut it off and throw away the top bit
make neet the connections

il mod my own falcon tomorrow and post some pics all going well

if a few 'more than 10' pm me
ill make a kit

this will be pcb populated and tested
a holder or cell pack and batteries and instructions

all with plugs and short wired up to suite board
should mount on top of the nvram
especialy if you remove the old cell from it with cutters
dont leave it there even with the other members mod
it may get hot if you used say alakline batteries
and its own diodes have gone short say or leaky
it can explode showering your machine with eating alkalie
its a ticking time bomb much like the amegas batteries

the archive below has the .bmp files if you prefer these when making pcb's

views are needed please !!!

[1024MAK]

There are a number of circuits to perform this function.
If you are using a NiMh or NiCad 3.6V PCB mounting cell, these only need a resistor connected to the 5V supply line.

See this circuit:-

RTC battery back-up.GIF

The value of resister R1 depends on the capacity of the battery. The above circuit is most suitable for a PCB mounting battery that is designed for this appication. If a larger battery pack is installed then reduce the resistor value to say 47R.

Zener diode ZD1 is optional - it should not be needed for this circuit, but it may prevent a wrongly fitted battery from killing the RTC chip.

Also have you considered the self discharge current of the battery?

See http://www.maxim-ic.com/appnotes.cfm/an_pk/3816

[simbo]

the problem and its still a problem with 3.6v cells is your just on the boundry of the ram requirements
lowest voltage last time i checked nvrams run at vcc=5v so the battery needs be be more than 5 or the ram will just forget
atari falcon psu usualy gives out about 5.2-5.6 v on the vcc rail

the diode the 3.6vcell connects through will drop .6 v
so the actual rtcc supply will be 3v and this is not enought drive
if you look at the site

here is the quote
"
To implement, we will need to solder 2 diodes to the Vin of the NVRAM at pin 24. To have the a diode between the Vin and the +5V power, we need to cut the leg is pin 24. A sharp cutter is required, making sure that the cut leave sufficient space at the chip leg to solder the diodes on to.
The other end of the battery can be soldered directly onto the ground at pin 12. There is no need to cut the legs here.
I decided to put the diode assembly on a small PCB.

Unfortunately for me, putting a 3.6V battery did not work, and the Falcon refuse to boot up.

After sometime, I replaced the 3.6V battery with a pack of 4 x 1.5 AA batteries (similar to my TT030). With the 6V battery over the NVRAM, the Falcon was able to boot AND remember the datetime plus NVRAM setting.
"

i always prefer to use a small transistor to charge a battery from 12v
as 12v - 6v = 6v left over and give constant current
youll get a hot resistor as it may drop as much as 45maX4 =200ma
iv=w so .2 X 6v = 1.2watts and thats too much heat.... unless you use a 2watt resistor from a tv

as you say there are many ways to do it
using a 3v cell isnt one of them and for this reason alone
i disagree with the cell type you used as the falcon just may not boot
dispite what the maxim datasheet says
as myself and many others have found
therefore elfacto reverso nor does charging it from the 5v rail as you need 6v 45ma.... constant current
and for this you need the 12v rail i did think all about this before i posted the design above
and i also built one and tested it

6v-.6v = 5.4 for the supply needs of the rtcc
and this is acceptable given that if the original cell is still there
even though its diode coupled itself it will still have a reverse leekage to the internal cell
i recommend removing the older cell totaly from the top of the ram
as very often the chemicals its exudes in later life will eat through and kill the chips die
if you leave it there and just add a battery then this move will also make the battery leek electrolytes faster

[simbo]

i revised the above circuit to add a diode to stop any chance of the cell discharging via its charger circuit

i added this mod to my own

you can also just use a 500ma 5v regulator and three diodes on its center ground pin
this will raise its output to ~6.8v adding just one resistor is then enought
the inline load resistor to use will not be 1R maybe more like 22-33R

[atarimusic[dot]net]

Just to let everyone know, you can order an 'ultra-low profile' IC socket for your new NVRAM Dallas chip that requires no cutting of the shield or ugly mod-jobs with wires and tape. I just changed 3 NVRAM chips on all my Falcon this month without any cutting or butchering using 'ulta-low profile' IC sockets. Most vendors sell just the 'regular' or 'low profile'. Try www.Digiparts.com to find the ultra low profile IC sockets. I even had 2mm of room left between the top of the chip and shield. I shiver at the thought of soldering wires and taping batteries to boost an old dead chip. It only taken an additional 1.5 - 2 hours to desolder the old chip and install a new one cleanly and professionally.

I hope that helps others in the future from cutting into their machines...

[simbo]

my perfered method is to not try to desolder such ic but
just cut it off with sidecutters all the pins
heat and remove them then suck out the solder

all from above there is no need to remove the main pcb

replacement of the nvram isnt really needed a carefull battery mod requires very little skill in soldering

so any method that makes it work is perfered
is the ground rule

[simbo]

i discovered something about the falcon
when the battery is so flat it wont boot {usual prob about 1 year after the clock fails to keep time}
monitor setting etc fail in the nvram


or if its a hardware problem and not this battery issue!!


if the machine when you turn it on the yellow floppy light will lightly flash for i think about 50ms
wait a few seconds and turn off
the yellow light for floppy access will pulse again for about 150ms
this is the battery

however if there is no tos at all loaded or processed the yellow led wont light atall
this is most likely a hardware fault like ram rom etc... failure of an ic
so no processing
just something i noticed

maybe its just the different machines
however with my own falcon if i disconnect the clock from the mpu then also the battery mod
it doesnt flash
reconnecting the mpu clock and no battery it flashed as described

so this maybe a quick check

[wongck]

i discovered something about the falcon
when the battery is so flat it wont boot {usual prob about 1 year after the clock fails to keep time}
monitor setting etc fail in the nvram


or if its a hardware problem and not this battery issue!!


if the machine when you turn it on the yellow floppy light will lightly flash for i think about 50ms
wait a few seconds and turn off
the yellow light for floppy access will pulse again for about 150ms
this is the battery

however if there is no tos at all loaded or processed the yellow led wont light atall
this is most likely a hardware fault like ram rom etc... failure of an ic
so no processing
just something i noticed

maybe its just the different machines
however with my own falcon if i disconnect the clock from the mpu then also the battery mod
it doesnt flash
reconnecting the mpu clock and no battery it flashed as described

so this maybe a quick check

yes
when the nvram battery is gone there is also a clicking
sound from the speakers
I think it retries to boot up continuouly
sorry cannot count in msec so not sure the frequesncy of it
something similar to what yuo are seeing on the flashing
floppy disk light I guess

[jfl]

I received a chip from Maxxim a few months back and I still have a non-working nvram. Is this possible that I was sent a faulty chip?

[wongck]

I received a chip from Maxxim a few months back and I still have a non-working nvram. Is this possible that I was sent a faulty chip?

Have you tried reseting the NVRAM?
IIRC on mine I had to reset it before the falc boot up correctly.

[jfl]

Have you tried reseting the NVRAM?
IIRC on mine I had to reset it before the falc boot up correctly.

Yes, I've reset it with resetnvm.prg as found in the CT60 archive. It worked for a day then back to non-working.
The Falcon boots fine, though, so this is not a big issue. I just like things working perfectly

[wongck]

It worked for a day then back to non-working.
The Falcon boots fine, though, so this is not a big issue. I just like things working perfectly

Probably as you mentioned, it's one with the battery dried out as well
Any idea when was that device made? It should last 10 years or so from the date.
May be you should just put AA batteries over the pins like what I did.

[simbo]

Have you tried reseting the NVRAM?
IIRC on mine I had to reset it before the falc boot up correctly.

Yes, I've reset it with resetnvm.prg as found in the CT60 archive. It worked for a day then back to non-working.
The Falcon boots fine, though, so this is not a big issue. I just like things working perfectly

dont turn it on !!!
instead take a VERY carefull look around all packages
pins !!

TAKE NOTHING FOR GRANTED HERE WITH SHORTS !!!
and packages like the shd {super high density } combel
mega hard too change and even find for less than $45 + post chip
you must examine every area of them to ensure you have no 'solder shard's'
flakes of solder on the combel or under the sockets of the GAL/PAL or mpu pins
its MOST important this or your bird will die
replacement of the nv ram isnt needed just the duff battery bypassed is best
this avoids problems like this
i use several eye loops X8 - X80 you get them on ebay and also in sets
so they are worth having to inspect pins
tip is to not put them in your eye but to magnify the pins with them
its suffice to say that a good X20 or more illuminated mag glass is all thats needed
+ steady hands and a very thin pin with a small hook on the end by bending the very tip on a hard planar

use 4X 1.5v = 6v dc via a diode = 5.4v supply to the correct pin as above {use the diodes}
once its working stable you can concider a recharge mod also ...

you must use 6v dc via two diodes or
i feel maybe the ram has failed

this does happen as the acids / alkalie from the batteries eats throught the die
and kills the chip
even bad volts can
removing the spent battery is the very best thing
DONT leave it there
cut it off with side cutters
it eats the new cell's
the diode and cells mod will replace the cell electricaly
all falcons use two types of nvram and cells
4 makers of chips
two are 3.6 v backup and have internal circuits
all can be run from a diode couples 6v battery supply
here is a good reset program
http://doyeuxyvan.free.fr/keeptime/v1.0/

[simbo]

if i remember
due to the nafarious nature of nvram and tos

you must place the reset .prg what ever one

in the auto folder of a boot floppy
and using a sm124 125 or hires to vga cable setup


let it auto boot like a game and set it then
but you must have battery fixes in place first
or youll compound your problem ...


afaik???

[Jorstad]

Hi!

I've got some free samples of the Dallas 12887A from maxim, but I noticed there are fewer pins on these?
Made me a bit insecure =).. Also, I've lended a solderstation from work to do the switch, because my own
15w iron does'nt get warm enough it seems.
What temperature should I have?

Regards
/Daniel

*edit* installed one of the maxim chips, 350celsius on the iron, worked at first boot =) Hope the battery will last a while.

[Linuxbuilders]

Functional Differences–DS12885/87
Pin # MC146818 DS12885 DS12887
1 NC If the Motorola bus timing is used, pin 1 needs to be tied to VCC. If the Intel bus timing is used, then pin 1 can be either tied to ground or left floating. The MC146818 had internal circuitry to differentiate between the two timing sets.
2 OSC1 OSC1 The DS12887 does not require an external crystal, so any components on the board designed to connect a crystal to this pin may be removed. The MC146818 would accept input frequencies of 32.768 kHz, 1.048576 MHz, or 4.194304 MHz. The DS12887 includes a 32.768 kHz crystal in the module.
3 OSC2
4 AD0 N/A N/A
5 AD1 N/A N/A
6 AD2 N/A N/A
7 AD3 N/A N/A
8 AD4 N/A N/A
9 AD5 N/A N/A
10 AD6 When accessing the clock memory, AD6 on the DS12885/87 needs to be low during the address strobe. It was a "don't care" on the MC146818.
11 AD7 N/A N/A
12 VSS N/A N/A
13 /CE\ N/A N/A
14 AS N/A N/A
15 R/ /W\ N/A N/A
16 NC The negative side of the backup battery needs to be connected to pin 16 of the DS12885. The DS12885 handles the switching to battery power internally and will disable reads and writes while in back backup mode. N/A
17 DS N/A N/A
18 /RESET\ N/A N/A
19 /IRQ\ N/A N/A
20 CKFS Pin 20 needs to be connected to the positive side of the backup battery. No Connect
21 CKOUT The user RAM will be cleared if this pin is taken low. There is an internal pull-up resistor, so this pin should be left floating and not tied high. This pin is a no connect on the DS12887. On the DS12887A, the RCLR pin is brought out and should be treated the same as on the DS12885.
22 PS The MC146818 used this pin to sense the battery voltage and set the VRT bit. This function is handled internally in the DS12885/87, and the pin is a no connect.
23 SQW N/A N/A
24 VDD Any on board backup battery that was connected to pin 24 should be removed. The DS12885 is designed to have the backup battery connected to pins 16 and 20. The DS12887 has the backup battery included in the module.

Functional Differences–DS1685/87
Pin # MC146818 DS1685 DS1687
1 NC The /PWR\ pin on the DS1685/87 is an output designed to support system power switching as a result of a date/time alarm (wake up) or a key closure (kickstart).
2 OSC1 The MC146818 required external components along with crystals of 32.768 kHz, 1.048576 MHz, or 4.194304 MHz. The DS1685 is designed to work with a 32.768 kHz crystal connected directly to pins 2 and 3. The DS1687 does not require an external crystal, so any components on the board designed to connect a crystal to this pin may be removed. The MC146818 would accept input frequencies of 32.768 kHz, 1.048576 MHz, or 4.194304 MHz. The DS1687 includes a 32.768 kHz crystal in the module.
3 OSC2
4 AD0 N/A N/A
5 AD1 N/A N/A
6 AD2 N/A N/A
7 AD3 N/A N/A
8 AD4 N/A N/A
9 AD5 N/A N/A
10 AD6 When accessing the clock memory, AD6 on the DS1685/87 needs to be low during the address strobe. It was a "don't care" on the MC146818.
11 AD7 N/A N/A
12 VSS N/A N/A
13 /CS\ N/A N/A
14 AS N/A N/A
15 R/ /W\ N/A N/A
16 NC The negative side of the backup battery needs to be connected to pin 16 of the DS1685. The DS1685 handles the switching to battery power internally and will disable reads and writes while in back backup mode. N/A
17 DS N/A N/A
18 /RESET\ N/A N/A
19 /IRQ\ N/A N/A
20 CKFS This pin needs to be connected to the positive side of the backup battery. No Connect
21 CKOUT The user RAM will be cleared if this pin is taken low. There is an internal pull-up resistor, so this pin should be left floating and not tied high.
22 PS The MC146818 used this pin to sense the battery voltage and set the VRT bit. This function is handled internally in the DS1685/87. This pin is the positive side for an auxiliary battery required for the kickstart and wake-up features.
23 SQW N/A N/A
24 VDD Any on board backup battery that was connected to VDD should be removed. The DS1685 is designed to have the backup battery connected to pins 16 and 20 in the DIP and pins 20 and 24 in the PLCC. The DS1687 has the backup battery included in the module.

IMG_1393.JPG

IMG_1394.JPG

simple steps:
1. use dremel and cut off d top section of plastic cover, clean of silicone (good luck), install new battery. Cut off just the top, like the first 1mm of it.
2. use dremel and cut off legs on both sides, then use blade and cut around the bottom part of the plastic cover, so the chip is separated from the plastic shell, then cut off the top with dremel. Then take out the chip of the plastic "box" and then clean. It is messy job but can be done.

[Shredder11]

Hey my free NVRAM samples arrived from Maxim yesterday and I too used the name Dalltech I got them as spares for my two Falcon 030 just in case and I ordered two different types; a DS12887+ and DS12887A. I just need to obtain a few sockets for them now, wherever I get those from.

[wongck]

Hey my free NVRAM samples arrived from Maxim yesterday and I too used the name Dalltech I got them as spares for my two Falcon 030 just in case and I ordered two different types; a DS12887+ and DS12887A. I just need to obtain a few sockets for them now, wherever I get those from.

Dalltech..... ought to be DULL-tech
Didn't they check it out on google??
must be a pretty BIG watch if they use a Dallas chip
http://www.dalltech.com/