UPS battery replacement

[who where]

On Thu, 11 Oct 2012 21:40:22 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

On Fri, 12 Oct 2012 10:14:58 +0800, who where <noone@home.net> wrote:

Just as laptops do ...

The joy of specmanship. If I'm lucky, and can find a way to reduce
charge current or float voltage limit in software, I can usually
extend the battery life. However, that's often impossible without
inside information or reverse engineering. I see far too many laptops
eat batteries due to overheating and overcharging in laptops that have
never been run on battery power.

A prime spec for laptops is runtime on battery. Years ago I did a
commercial Li-Ion charger design. As part of the process I did a fair
number of charge/discharge tests of packs on an automated HP rig, and
obtained some interesting insight. Typically laptops (then) used 4v20
as their CLCV regime voltage limit. Reducing this to say 4v10 would
lose them roughly 20% of runtime (your link suggests a smaller
difference than we saw) but increase the cell lifetime by a very
worthwhile amount. If customers had this option I'm sure all the
thinking ones whose personal $$$ were involved would opt for the lower
voltage.

They also tend to recharge rather quickly, in order to deal with
repetitive power failures.

Yes, almost universally.

I've often wondered if there's a recharge time spec for UPS's. A few
quick Google searches didn't find anything. I can see why the UPS
manufacturers would want to recharge quickly, but they should at least
give the customer the choice between fast recharge and long battery
life.

Exactly. Most customers have a better than generic idea of what power
outage patters are in their area.

I have been called on to "service" quite a number of failed soho UPS'
of varying sizes from 150W to 4kW, mostly APC. The electronics are
almost universally fine - except for one with dead FETs in the
inverter, the failure mode in every single one was batteries. When I
bothered to check the charge regime, it was found to implement the
above rules 1 and 3 on how to kill UPS batteries.

Yep. I've had similar experiences. I don't see many big UPS's but
overcharging is epidemic on small UPS's.

Rhetorical question:
Why are there no UPS's that use Li-Ion batteries?
Spoiler:
http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

The first answer to your rhetorical question would be price and the
need for the UPS manufacturers to actually do contemporary designs
instead of churning out 80's technology, and I'm being kind there.

Note though that with Li-xx the "coulomb efficiency" on charge is
around 98% on our testing (admittedly with new cells with less than 20
cycles) compared to the <80% for say NiXX and probably <90% for SLA's.
The difference between those numbers and 100% obviously appears as
heat I could well be wrong, but it seemed during our Li-Ion cycle
testing that the heat dissipated during charge from that ~2% loss
wasn't reflected in a rise in cell casing temp. Maybe the chemical
change during charge is a tad exothermic? Either way, there isn't the
temperature issue with Li-Ion in their *electrical* role that one
experiences with other chemistries - laptops being a perfect example
of how heat from OTHER parts/devices causes the high temp battery
environment. Equally noteworthy is the absence of a float charge
current in Li-xx charge regimes.

If Li-Ion packs were:

(a) oversized - to keep DOD to less than 50% (and enforced by a higher
than 3v0/cell shutoff point);
(b) kept external to the UPS and its heat generation; and
(c) maintained by a sensible charge circuit
then they would be reasonably suitable for UPS duty. I doubt the
average SLA would match them for cycle life in that scenario.

 

[tm]

"who where" <noone@home.net> wrote in message
news:r8gh78h7drv6l6c4mn92h340akos8vj8i5@4ax.com...

On Thu, 11 Oct 2012 23:06:55 -0400, "tm" <No_one_home@white-house.gov
wrote:


"who where" <noone@home.net> wrote in message
news:hbue78914q7bbk913e02oh255t8llp83od@4ax.com...

A bit off-(thread)-topic, but I have a $20k+ communicatios test set
that uses Gates cylindrical sealed lead-acid cells in the battery
pack. The inbuilt charger design was such that these sets seem to
kill packs ina couple of years. Tiring of this recurrent cost, I
rejigged the charging circuit some 12 years ago, and haven't had a
pack die since. Sample of one isn't generally considered
statistically significant, but this is to me.


This does sound like maybe an IFR communications service monitor.

Yep, 1200S

If so, can you give any details on your mod?

Well, it was 12 or so years back. I may have documented the change
somewhere, and I'll look through my mountain of "design notes" for it,
but I wouldn't go without food and drink waiting if I were you. I
certainly didn't put the note in the Service Manual.

Ok, no big rush as I need to save up for a round tuit before I get to work
on it.

tm

 

[who where]

On Fri, 12 Oct 2012 21:49:26 -0400, "tm" <No_one_home@white-house.gov>
wrote:

"who where" <noone@home.net> wrote in message
news:r8gh78h7drv6l6c4mn92h340akos8vj8i5@4ax.com...
On Thu, 11 Oct 2012 23:06:55 -0400, "tm" <No_one_home@white-house.gov
wrote:


"who where" <noone@home.net> wrote in message
news:hbue78914q7bbk913e02oh255t8llp83od@4ax.com...

A bit off-(thread)-topic, but I have a $20k+ communicatios test set
that uses Gates cylindrical sealed lead-acid cells in the battery
pack. The inbuilt charger design was such that these sets seem to
kill packs ina couple of years. Tiring of this recurrent cost, I
rejigged the charging circuit some 12 years ago, and haven't had a
pack die since. Sample of one isn't generally considered
statistically significant, but this is to me.


This does sound like maybe an IFR communications service monitor.

Yep, 1200S

If so, can you give any details on your mod?

Well, it was 12 or so years back. I may have documented the change
somewhere, and I'll look through my mountain of "design notes" for it,
but I wouldn't go without food and drink waiting if I were you. I
certainly didn't put the note in the Service Manual.

Ok, no big rush as I need to save up for a round tuit before I get to work
on it.

Time to find the tuit then. Just dragged out the maintenance manual.
From the input transformer secondary/bridge and via a diode the
battery charge line goes to an LM317T. They use the LM317 as a
current-limited voltage source. That's two parameters they could
potentially screw up. IMHO they did both.

While they recommend setting the voltage at 14v4 (i.e. 2v40 per cell)
many of these instruments are exposed to long period of AC supply and
this is above the Hawker/Gates recommended float setting of 2v27-2v35
per cell.

I wanted more than the customary 2-year pack life, so I adjusted R4 to
set the float voltage to around 13v8 (i.e 2v3/cell). See below (*)
regarding setting this.

Now to deal with recharge current limiting. To quote from the Hawker
Cyclon Selection Guide: "There is no need to limit the inrush current
to the battery during the initial phase of constant voltage charging.
The internal resistance of Hawker Cyclon single cells and Monoblocs
allows for large inrush current without damage." The LM317T will
internally limit current to around 1.5A, which is puny when you
realise these cells have a 5-minute constant-power discharge
capability of 45A. Give that useless 0R22 2W current limit resistor
the flick. Its only contribution is to unnecessarily slow recharge.

(*) After the voltage divider branch on the 317 output, there is a
dirty big diode in series with the battery line, to prevent unwanted
backfeed/discharge This has the effect of puuting a slope on the
charge I/V curve anyway, so the voltage setting needs to be executed
at zero drop across this i.e. measured at the battery terminal point
with the battery either fully charged or preferably disconnected
(charger O/C).

If you don't have the rather rare maintenance manual, I can send a
part schematic to a valid email addie if you wish.

Caveat: this information is provided for educational purposes only
and the consequences of applying it in all cases lie with the user.

 

[Jeff Liebermann]

On Sun, 14 Oct 2012 15:31:26 +0800, who where <noone@home.net> wrote:

If you don't have the rather rare maintenance manual, I can send a
part schematic to a valid email addie if you wish.

Scanned IFR manuals and schematics can be found at:
<http://www.repeater-builder.com/test-equipment/aeroflex/aeroflex-index.html>

The IFR-1500 charger circuit is worse:
<http://802.11junk.com/jeffl/crud/IFR-1500%20battery%20charger.jpg>
Same 14.2V float. I'm using 13.8V for gel cell float voltage.
<http://www.powerstream.com/SLA.htm>

Thanks much for the details.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

 

[Jeff Liebermann]

On Sat, 13 Oct 2012 09:48:52 +0800, who where <noone@home.net> wrote:

Exactly. Most customers have a better than generic idea of what power
outage patters are in their area.

We don't have too many outages in this area. Mostly, they're short
term glitches and gaps lasting at most a few seconds. The UPS hardly
gets a workout with such short duration runtimes. Mostly, the
batteries sit there, do nothing, get warm, and die without ever
getting used. Short recharge time is totally wasted on such systems.

The others are big UPS's that are intended to only run for a few
minutes while the automatic generator starts, warms up, and kicks in.
For those, I may want 10kw, but only for maybe 2 minutes maximum.
That's quite a different style of operation, that again doesn't need
short recharge times. We occasionally do have long term outages, but
that's what generators are for.

Rhetorical question:
Why are there no UPS's that use Li-Ion batteries?
Spoiler:
http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

The first answer to your rhetorical question would be price and the
need for the UPS manufacturers to actually do contemporary designs
instead of churning out 80's technology, and I'm being kind there.

The "answer" is that LiIon batteries will self-deteriorate if left on
100% full charge and in a hot box. See above URL.
Charging to 4.10V/cell the battery holds about 10 percent
less capacity than going all the way to 4.20V. In terms
of optimal longevity, a voltage limit of 3.92V/cell works
best but the capacity would only be about half compared
to a 4.20V/cell charge (3.92V/cell is said to eliminate
all voltage-related stresses).
In other words, if the LiIon battery were charged to full capacity, it
would be dead in a few months, even if it had never experienced any
charge/discharge cycles.

However, I haven't been paying attention to new developments. Drop in
Lithium Iron Phosphate UPS replacement batteries are available:
<http://www.powerstream.com/LLLF-12v-7ah.htm>
The claim is that they will last longer and survive more
charge/discharge cycles. At the $135/ea price, we're not going to see
these in commodity UPS's in the near future, but maybe for critical
applications. Personally, I don't see the benefit. If longer life
were a problem, just lowering the float voltage on gel cells would
accomplish the same thing at much lower cost.

Note though that with Li-xx the "coulomb efficiency" on charge is
around 98% on our testing (admittedly with new cells with less than 20
cycles) compared to the <80% for say NiXX and probably <90% for SLA's.
The difference between those numbers and 100% obviously appears as
heat.

Yep. That's a standard part of solar power calculations. The 10 year
end-o-life efficiency for even a good quality (Trojan) battery is
about 70%. With a new battery, it's about 85%, but it drops linearly
until the battery becomes more of a heater than a storage device. I
use 75% for my mountain top installs.

I could well be wrong, but it seemed during our Li-Ion cycle
testing that the heat dissipated during charge from that ~2% loss
wasn't reflected in a rise in cell casing temp. Maybe the chemical
change during charge is a tad exothermic?

I've been playing with cell phone batteries most of which use a
conservative charge cycle. During charging, I haven't seen any
heating on new cells. Old cells will get warm. For fun, I charge
cycled an obviously bulging battery which finally became warm enough
for me to worry about it catching fire. I don't know much about the
chemistry involved.

Either way, there isn't the
temperature issue with Li-Ion in their *electrical* role that one
experiences with other chemistries

Agreed. New batteries by themselves do not get how when charging. Old
batteries are another story.

- laptops being a perfect example
of how heat from OTHER parts/devices causes the high temp battery
environment. Equally noteworthy is the absence of a float charge
current in Li-xx charge regimes.

Agreed. Laptops are a horrible environment for LiIon. However, that
also applies to UPS's. The heat from the UPS or the server closet
will cause the LiIon batteries to deteriorate more quickly. Keeping
the battery at 100% charge (4.1V) will also speed up its destruction.
Of course, in order to produce impressive UPS runtime numbers, the
manufacturers will set the charge point to as close to self
destruction as possible.

If Li-Ion packs were:

(a) oversized - to keep DOD to less than 50% (and enforced by a higher
than 3v0/cell shutoff point);

That would double the battery costs. Since that seems to be the most
expensive part of the puzzle, it's unlikely. Still, it may be a
requirement to prevent short battery lifetimes.

(b) kept external to the UPS and its heat generation; and

Submerge the LiIon battery pack in a water jacket or may anti-freeze?
That might just work for larger cells, or maybe a drop in gel cell
replacement.

Or, maybe a mess of 18650 LiIon batteries shoved into a finned
aluminum tube. Each cell protects itself from over/under voltage with
internal circuitry making replacement easy. Hmmmm...
<http://media.photobucket.com/image/recent/Benckie/torch/100_5998.jpg>

(c) maintained by a sensible charge circuit
then they would be reasonably suitable for UPS duty.

History had demonstrated that this isn't going to happen. Nobody is
going to advertise a UPS with a battery that lasts longer. Instead,
they're going to advertise that it runs longer.

I doubt the

average SLA would match them for cycle life in that scenario.

Yep. If you want to see what it takes, read about what Tesla Motors
and A123 did with their battery packs. It has a cooling systems built
in which has become a problem. They claim that the new Nanophosphate
EXT battery will eliminate the cooling requirement, but I'm not so
sure:
<http://www.a123systems.com/lithium-ion-battery-technology.htm>
(Note the reference to telecom backup in the text). If it does work
as advertised, it would make a nifty UPS battery that wouldn't get hot
during charge.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

 

[who where]

On Sun, 14 Oct 2012 09:58:01 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

On Sat, 13 Oct 2012 09:48:52 +0800, who where <noone@home.net> wrote:

Exactly. Most customers have a better than generic idea of what power
outage patters are in their area.

We don't have too many outages in this area. Mostly, they're short
term glitches and gaps lasting at most a few seconds. The UPS hardly
gets a workout with such short duration runtimes. Mostly, the
batteries sit there, do nothing, get warm, and die without ever
getting used. Short recharge time is totally wasted on such systems.

The others are big UPS's that are intended to only run for a few
minutes while the automatic generator starts, warms up, and kicks in.
For those, I may want 10kw, but only for maybe 2 minutes maximum.
That's quite a different style of operation, that again doesn't need
short recharge times. We occasionally do have long term outages, but
that's what generators are for.

Bacvk in the mid-90's I had the end-to-end task from specifying though
purchasing/commissioning of a 900 kVA UPS system for a large 'pooting
centre (when such centres used "big iron"). This was configured as
3x300 - load was just under 600 - hence 3 units with one redundant.
Three diesel-alternator sets produced around 1500 kVA total (1x900 and
2x300). Paralleling was given about 10 seconds after the first set
came onto the generator bus. PLC was set to allow bus connect to
load with one large or any two, once that time had expired. The
diesel reliability was such that we never had an occasion where the
UPS battery banks saw a full minute of discharge, although they were
sized for about 25 minutes at full 300kVA load. The float voltage was
set rather conservatively on that too.

Rhetorical question:
Why are there no UPS's that use Li-Ion batteries?
Spoiler:
http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

The first answer to your rhetorical question would be price and the
need for the UPS manufacturers to actually do contemporary designs
instead of churning out 80's technology, and I'm being kind there.

The "answer" is that LiIon batteries will self-deteriorate if left on
100% full charge and in a hot box. See above URL.
Charging to 4.10V/cell the battery holds about 10 percent
less capacity than going all the way to 4.20V. In terms
of optimal longevity, a voltage limit of 3.92V/cell works
best but the capacity would only be about half compared
to a 4.20V/cell charge (3.92V/cell is said to eliminate
all voltage-related stresses).

I did note one line in the referenced page which states:

"The author of this essay does not depend on the manufacturer’s
specifications alone but also listens to user comments.
BatteryUniversity.com is an excellent sounding board to connect with
the public and learn about reality."

and that left me uncertain to what extent these numbers are anecdotal
rather than the result of testing - and if so, whose testing.

I have a problem with their figures. For me the 90% for 4v10 doesn't
sit well with 50% for 3v92. The difference is too wide.

As I mentioned, I'd expect - and this is consistent with our own
cycling tests on 18650's - more like low 80% range for 4v10.

In other words, if the LiIon battery were charged to full capacity, it
would be dead in a few months, even if it had never experienced any
charge/discharge cycles.

I have a problem with that too. Very often I see posts in forums that
say Li-Ion last a maximum of three years. What garbage. I have here
a BTP-T31 pack (Sony Corp, made in JP, from an AcerNote Lite 370
lappie) date-coded 9637. New it gave a tad over 3 hours. I give it a
cycle about every year - when I remember - and it now gives about two
hours. I then recharge it (to I presume 4v20 - I haven't measured -
and put it back into storage. Again, a sample of one ....

(snip)

- laptops being a perfect example
of how heat from OTHER parts/devices causes the high temp battery
environment. Equally noteworthy is the absence of a float charge
current in Li-xx charge regimes.

Agreed. Laptops are a horrible environment for LiIon. However, that
also applies to UPS's. The heat from the UPS or the server closet
will cause the LiIon batteries to deteriorate more quickly. Keeping
the battery at 100% charge (4.1V) will also speed up its destruction.
Of course, in order to produce impressive UPS runtime numbers, the
manufacturers will set the charge point to as close to self
destruction as possible.
aka 4v20 for that geberation of cells ...

If Li-Ion packs were:

(a) oversized - to keep DOD to less than 50% (and enforced by a higher
than 3v0/cell shutoff point);

That would double the battery costs. Since that seems to be the most
expensive part of the puzzle, it's unlikely. Still, it may be a
requirement to prevent short battery lifetimes.

Indeed. Even with present-day soho units, the cost of replacing the
SLA's is usually not materially different from replacing the UPS.

I recall in one forum a poster was bitching about the cost of new
SLA's for his pre-owned UPS. I explained that IMHO there are two rules
for soho UPS use - (1) expect to replace the batteries at max 2-year
intervals if you want the thing to provide a reliable backup, and (2)
if you can't afford to do that, you can't afford to own a UPS.

The cost factor drives a lot of battery decisions. On one small
AC-supplied communications site which was critical for an annual event
lasting a few days, we used a maintenance-free automotive battery.
New one installed each year just before the event. That was cheaper
than funding a proper industrial system.

(b) kept external to the UPS and its heat generation; and

Submerge the LiIon battery pack in a water jacket or may anti-freeze?
That might just work for larger cells, or maybe a drop in gel cell
replacement.

nothing that serious, just kept away from heat sources.

Or, maybe a mess of 18650 LiIon batteries shoved into a finned
aluminum tube. Each cell protects itself from over/under voltage with
internal circuitry making replacement easy. Hmmmm...
http://media.photobucket.com/image/recent/Benckie/torch/100_5998.jpg

You'd need that individual protection, as you'd need a pretty large
stack of 18650's to do much.

(c) maintained by a sensible charge circuit
then they would be reasonably suitable for UPS duty.

History had demonstrated that this isn't going to happen. Nobody is
going to advertise a UPS with a battery that lasts longer. Instead,
they're going to advertise that it runs longer.

The grim truth.

 

[who where]

On Sun, 14 Oct 2012 09:06:22 -0700, Jeff Liebermann <jeffl@cruzio.com>
wrote:

On Sun, 14 Oct 2012 15:31:26 +0800, who where <noone@home.net> wrote:

If you don't have the rather rare maintenance manual, I can send a
part schematic to a valid email addie if you wish.

Scanned IFR manuals and schematics can be found at:
http://www.repeater-builder.com/test-equipment/aeroflex/aeroflex-index.html

I'm good with a full original, but that will certainly help others.

The IFR-1500 charger circuit is worse:
http://802.11junk.com/jeffl/crud/IFR-1500%20battery%20charger.jpg
Same 14.2V float.

?? my IFR1200S/A manual shows 14v4 ...

Also, not sure if it is worse. ALthough I am a 317 fan, at least the
divided output is compared to a 5v1 zener, potentially (ouch!) a
better reference than the 317's internal. Only downside is that 5v1
is the zener value nearest zero tempco, while most SLA manufacturers
actually cite a required charge tempco.

I'm using 13.8V for gel cell float voltage.
http://www.powerstream.com/SLA.htm

Thanks much for the details.