[Nut-upsdev] Logic problem in NUT with upscode2 driver
Ted Mittelstaedt
tedm at mittelstaedt.us
Wed Feb 19 18:56:01 UTC 2014
Hi All again,
OK I found the section of code in upscode2.c I think is applicable - my
UPS does not supply battery charge current, meaning I think that it
operates using this logic:
}
/* Old method, assumes battery high/low-voltages provided by
UPS are
* applicable to battery charge, but they usually aren't */
else if (batt_volt_low > 0 && batt_volt_high > 0 && batt_volt >
0) {
if (batt_volt > batt_volt_high) {
chg=100;
}
else if (batt_volt < batt_volt_low) {
chg=0;
}
else {
chg = (batt_volt - batt_volt_low) /
(batt_volt_high - batt_volt_low);
chg*=100;
}
}
else {
return -1;
}
I believe the bug here is:
if (batt_volt > batt_volt_high) {
chg=100;
this should be:
if (batt_volt > batt_volt_nom) {
chg=100;
The problem (with my UPS) is that battery voltage is never going to
exceed the battery voltage maximum, (that would be a bad overcharge)
thus that decision is never taken
and we are always dropping through to the next formula which is
senseless when calculating state of charge of batteries that are ABOVE
their nominal value. By definition a lead acid cell above 2 volts is
fully charged.
I propose that anytime the batteries are above nominal, (48 volts), they
should be considered at 100% of charge.
I further propose that in the formula
else {
chg = (batt_volt - batt_volt_low) /
(batt_volt_high - batt_volt_low);
we change this to:
else {
chg = (batt_volt_low / batt_volt);
Thus, a 48v battery pack that has dropped below nominal value of 48
volts to 46v would be considered 86% charged.
Suggestions?
Ted
PS: Here's a patch
--- upscode2.c.original 2012-07-31 10:38:59.000000000 -0700
+++ upscode2.c 2014-02-19 10:54:27.000000000 -0800
@@ -1391,15 +1391,15 @@
/* Old method, assumes battery high/low-voltages provided by
UPS are
* applicable to battery charge, but they usually aren't */
else if (batt_volt_low > 0 && batt_volt_high > 0 && batt_volt >
0) {
- if (batt_volt > batt_volt_high) {
+ if (batt_volt > batt_volt_nom) {
chg=100;
}
else if (batt_volt < batt_volt_low) {
chg=0;
}
else {
- chg = (batt_volt - batt_volt_low) /
- (batt_volt_high - batt_volt_low);
+ chg = ( batt_volt_low /
+ batt_volt);
chg*=100;
}
}
On 2/19/2014 9:50 AM, Ted Mittelstaedt wrote:
> Hi All,
>
> I have what is probably a 12 year old (or older) Compaq R3000 UPS
> plugged into a FreeBSD 8 system running NUT version 2.6.5
>
> This UPS contains a total of 16 batteries in 2 banks of 8,
> each battery is 6 volts and the bank produces 50 volts.
>
> The UPS monitors battery voltage and the battery charger in
> the UPS works like this. For about 2 days it will apply a
> charge/float/topping voltage of 55.2 volts to the batteries.
> During that time the battery voltage measures 55.2 volts
> both with a voltmeter on the batteries and as reported by
> upsc, and the "calculated state of charge" of the UPS is 92.7%
>
> (keep in mind that UPSes that are driven by upscode2 are unable
> to report battery charge, so per the man page the upscode2 driver
> synthesizes that value based on nominal battery min/max and current
> battery voltage)
>
> Anyway, then the UPS shuts off the battery charger. The battery
> voltage then starts dropping gradually until the battery voltage hits a
> bit above 48v then the charger turns back on. (this takes about 2 days)
> For example, right now the battery voltage of each bank (as shown by
> upsc and confirmed by a voltmeter) is 50.5 volts yet the calculated
> state of charge is 64% even though the UPS is reporting
> a nominal battery voltage of 48v and the batteries are 2 volts ABOVE
> that.
>
> The problem I am having appears to be the logic in the upscode2
> driver. For starters, the calculated state of charge is quite wrong.
> 48v on a bank of 8 6v batteries is normal, it IS fully charged, it
> is NOT 64% charged. 50.5 volts on a 48v bank of batteries should be
> almost 100% charged, not 64%.
>
> Worse, however, is if there's a power failure right near the end
> of the 2-days-off cycle. That happened to me last week - it was a
> short duration 15 second loss - and the upscode2 driver decided it
> needed to issue a forced shutdown.
>
> Very likely this was because upscode2 had decided the batteries
> were dangerously low discharged. But they were NOT discharged and
> easily kept the servers up and online.
>
> The UPS reports a battery max voltage of 56.4 and minimum voltage of
> 40 and nominal voltage of 48 (this is obviously for a given bank)
>
> Now I do not know if OTHER upses that require upscode2 have battery
> chargers that sense battery voltage like this, and that shut off. If
> I disconnect the battery pack for a second and plug it back in, the
> battery charger sees the loss of battery voltage and turns back on,
> so it's obvious the charger is sensing battery voltage.
>
> I am guessing the upscode2 driver is assuming that the battery charger
> would be switched on all of the time. 55.2v is the correct float charge
> voltage for a bank of 8 6v SLA batteries, although the driver is
> calculating a state of charge of 92.7% when the charger is switched on
> and the battery bank is at 55.2v.
>
> I believe the upscode2 driver should be calculating a battery state
> of charge of 97% with a max battery voltage of 56.4 and a measured
> voltage of 55.2. It should not be reporting 92%
>
> And, I believe the upscode2 driver should NOT be calculating that state
> of charge below 97% until the measured voltage drops below the nominal
> voltage.
>
> Below the nominal voltage - 48v - is where the calculated state of
> charge should be dropping down below 90%
>
> Can anyone tell me how to correct the logic in upscode2? I can't have
> the NUT software shutting servers down when the UPS has plenty of
> runtime and capacity.
>
> Thanks!!!
> Ted
>
> _______________________________________________
> Nut-upsdev mailing list
> Nut-upsdev at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsdev
More information about the Nut-upsdev
mailing list