Re: Cache Battery question

On 5/18/2015 12:09 PM, DrFranken wrote:
> I have asked people who know these things for further clarification. I
> will pass it along as it becomes available.
> 
> However, I cannot imagine IBM selling a multi-million dollar machine to
> a large enterprise such as a bank for example that requires its cache
> batteries to be swapped out every time they have a power outage. I have
> personally worked on machines with over 100 of these which would
> translate to over $10,000 per outage!
> 
> Even harder to believe is that IBM would sell that machine to them and
> **NOT** tell them this is a requirement. (Because they do not tell them
> this.)
> 
> Even more harderer still to believe is that IBM Legal would allow them
> to sell this AND not tell them.
> 
>         - Larry "DrFranken" Bolhuis
> 
> www.frankeni.com
> www.iDevCloud.com
> www.iInTheCloud.com
> 
> On 5/16/2015 4:02 AM, Holger Scherer wrote:
> 
>> ok, lets name it like this - most of the controllers *Try* to recharge
>> the battery, but most of them do not a good job. We checked this on
>> several older or newer controllers using a break out cable and a voltage
>> monitor.
>>
>> For example, the old 2740/2748 do not really charge
>> (as they do not have electronics for this), they try to keep 3.6V
>> which is not enough for 3.6V Batteries.
>>
>> Same with the newer LiIo controllers which do not do a temperature
>> measure from the battery (which does have the cable). They send some
>> millivolts to keep the voltage, but they cannot recharge a
>> battery at a 70% level... If there was an unplanned outage, exchangeing
>> the battery is a good idea.
>>
>> -h
>>
>>
>>
>>
>> Am 16.05.15 um 08:02 schrieb DrFranken:
>>> That's not correct, the controllers DO charge the batteries. Think about
>>> what would happen if they didn't, you lose power for a few hours and
>>> your server goes down. The battery holds the cache during that time and
>>> of course discharges partially in that effort. When power is restored
>>> you bring up the system no problem.  If the batteries did not get
>>> charged by the controller then the remaining battery life might not be
>>> sufficient to hold up the cache during your next outage.

Doc,
I'm a ham.  I've build, debugged, and busted my fair share of power
supply and battery charging circuits - enough to understand the
technology well enough to speak about it in a public forum.  I don't
know the specifics of any IBM cache battery charge controllers though.
In layman's terms, Holger is describing the difference between a 1 amp
trickle charger and a 100 amp boost charger that one would use for a
car.  Every computer problem needs a car analogy :-)

If IBM's charge controller is built along the lines of a trickle
charger, I fully expect that they have a spec which says that newly
installed cache batteries need to be within X% of full charge, where X
is a number within the controllers ability to source current.

  --buck