Re: [Apcupsd-users] apcupsd not showing any UPS load

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

It takes a long time for electrolytic capacitors to go bad.   Yes your 
going to always have a few edge cases but we are looking mainly at 10 
years out on this kind of thing.  It's also known that cap failure is 
not as quick on gear under continuous power because when the cap dries 
out and is under load it basically keeps working until it loses power.

The other problem is the cost.  A lot of those electrolytics in UPSes 
are large and more expensive, and there are multiple values.  It's not 
fun to tear apart a UPS and have it sitting chewing up your bench while 
you meticulously read off all the values and make up your order and 
collect all the new parts.

Then there is the labor of unsoldering and resoldering in all the old ones.

I've done re-capping of other lower cost electronics.  Sometimes it 
works sometimes it does not.  Always the time cost is significant.  This 
kind of thing is classified much more in the labor of love department, 
akin to automobile restoration. Economically it's not worth it - your 
still most well advised to just buy a new UPS in most cases.

It would be much better for the entire apcupsd community for someone to 
take the work that has been done on reverse-engineering the Microlink 
protocol with python and add it into apcupsd or at least into NUT.

Ted

On 11/30/2025 11:15 AM, Pavel Boček wrote:
> Majority of "internal problems" are just garbage crapacitors going 
> bad. Now some of the newer schneider-tainted models tend to go bad 
> somewhat more because of different reasons, but it's is never a bad 
> idea to just replace the caps, it's gonna help at least half of the 
> times. Also it seems that some SMT models tend to burn through even 
> more accumulators then the old SU/SUA units.
>
> I go with (properly) refurbished old unit any day of the week, there 
> is no problem with SU/SUA units with pretty much nothing you plug into 
> them, apcupsd runs notoriously great with them, since they all use 
> only the good ol' Smart protocol (UPSlink). No reason to support their 
> stupid product development they show to us in the last decade.
>
> -- 
> S uctivým pozdravem/best regards,
>
> Pavel Boček
> Jabber: Be...@ja...
> +420 739 190 151
> http://www.hwworld.cz (elektromontáže silnoproud/slaboproud, 
> kondenzátory, aku/baterie, mikrospínače pro myši a klávesnice 
> aj./capacitors etc.)
> https://www.hardwareinsights.com (power supply reviews and more)
> stickers-cz.com (nekonformní samolepky, placky)
>
> ---------- Původní e-mail ----------
> Od: Jeffery Small <je...@cj...>
> Komu: apc...@li...
> Datum: 29. 11. 2025 21:22:35
> Předmět: Re: [Apcupsd-users] apcupsd not showing any UPS load
>
>
>     Ted:
>
>     Once again, thanks for the detailed history. It's is all making
>     more sense
>     now and yes, I do wish I had known about this before purchasing
>     another
>     SMC model, but since my old unit was working just fine on my previous
>     workstation and only needed a new battery, I really didn't give it
>     any
>     thought.
>
>     When I get a moment, I'll look into the MODBUS code and see what's
>     up.
>
>     I may swap out the battery from the new unit into the old one and see
>     if that helps. As you said below, my past experience in installing
>     new
>     batteries in the UPS has twice been followd by some UPS internal
>     failure
>     requiring the unit to be replaced, so I also just replace the
>     entire unit
>     rather than update the battery.
>
>     Regards,
>     -- 
>     Jeff
>
>     Ted Mittelstaedt <te...@mi...> writes:
>
>     >The main product lines hit are the SMC (and SRT) series. These were
>     >always targeted at the small business market - the 10-15 person
>     company
>     >with 1 server, running Windows Server.  The protocol change happened
>     >before Schneider purchased APC.  The really small individual
>     market or
>     >1-2 person company market was buying BackUPS and didn't care
>     about the
>     >self-test stuff.
>
>     >Although APC makes a rack-mounted SMC they mostly sell the
>     standalone
>     >UPSes in the SMC model line.
>
>     >I'm pretty positive  that the way things played out is that back
>     in the
>     >"olden days" during the reign of the beige-colored UPSes
>     (SmartUPS and
>     >BackUPS) that APC had some really sharp in-house talent that
>     designed
>     >and built microprocessor-controlled boards and very simple firmware.
>     >What I think happened was those people retired or moved on and for a
>     >while they were just making a few firmware tweaks but things
>     basically
>     >were the same through the reign of the "Black-fronted UPSes with
>     just
>     >lights"  But then the Chinese (CyberPower and Goldenmate and so on)
>     >started releasing UPSes with big digital readouts on the front
>     and APC
>     >had to respond to that - and also very likely they were also
>     facing the
>     >end of production of the older microprocessors and support chips
>     used
>     >for the UPSLink firmware.
>
>     >I'm pretty sure whoever was in charge of APC farmed out a
>     redesign of
>     >the new firmware to a 3rd party and I'm 100% positive that they
>     didn't
>     >know what the hell they were doing (the APC person not the 3rd
>     party)
>     >and the 3rd party took APC down the yellow brick road. Without
>     >understanding anything APC signed a developer agreement for a
>     >PROPRIETARY SDK that required them to not release source code or
>     other
>     >documentation for Microlink. When word of this got out the large
>     >enterprises (that's me) basically told APC we won't be buying
>     anymore
>     >UPSes from APC, we will switch over to Eaton, because your not
>     going to
>     >make us pay licensing for PowerChute Business over our dead bodies.
>
>     >That was what triggered the decision to write MODBUS support into
>     the
>     >APC firmware.  And as Klaasdc found out 6 years ago, the MODBUS
>     >implementation copies the same formatting of parameters from
>     Microlink,
>     >so clearly the MODBUS firmware was sort of a quick hack - and I'm
>     also
>     >pretty sure that the development group that screwed over APC was not
>     >particularly happy about writing it because I'm pretty sure they
>     get a
>     >portion every time APC sells a UPS with a copy of PowerChute.
>
>     >But, having 2 independent protocols inside of a UPS chews up more
>     flash
>     >than a single protocol, so the cost-reduced SmartUPS line - the
>     SMC -
>     >appears to be frozen out.
>
>     >I do hope you try out that KlassDC github python code I posted
>     the link
>     >to and report back to us - we get so little info on Microlink
>     because
>     >few people using apcupsd buy Microlink-only UPSes.
>
>     >There has, unfortunately, been some bit-rot in the apcupsd code, the
>     >primary problem at this time is that apcupsd has problems with
>     running
>     >MODBUS over USB now, whereas a few years ago, that worked.
>     However, the
>     >FDDI and ATEN and Prolific USB-to-serial adapters are so
>     ridiculously
>     >cheap on Ebay for the Chinese clones, AND they are STILL including
>     >honest-to-God serial ports on server-class hardware (HP Proliants
>     and
>     >such) that limiting MODBUS to serial port only inside of apcupsd, is
>     >really not the limit it appears on the surface.  The SMT models that
>     >include MODBUS are all targeted to Enterprises who are using them in
>     >cabinets and relay racks with server quality hardware that can
>     plug into
>     >the APC serial cable, the low-end BackUPSs that have monitoring
>     ports
>     >all support USB-UPSHID and work out of the box with apcupsd.
>
>     >It's actually more significant that NUT finally appropriated the
>     MODBUS
>     >code out of apcupsd just like they did when the UPSLink protocol was
>     >reverse-engineered.  NUT has a far more wide support for
>     different UPSes
>     >as well as they had some financial support from Eaton at one
>     time.  They
>     >also are affected by that bug with MODBUS not working on USB but I'm
>     >happy to let them try to fix it.
>
>     >It's the folks like yourself - people wanting to buy a cut above the
>     >SOHO BackUPSes, but not understanding why it's important to pay
>     an extra
>     >$200 for a SMT instead of a SMC, are the ones getting screwed
>     over.  I
>     >think the issue with the SMCs is that they only speak Microlink
>     out that
>     >USB port, not USB-UPSHID like the lower end BackUPS.  (the lower end
>     >BackUPS don't speak Microlink)  But it's really hard to know.  I
>     >personally don't have a SMC model UPS in my fleet so I can't
>     test.  It
>     >MIGHT be that the USB port merely starts out Microlink but
>     receipt of a
>     >magic packet switches it to USB-UPSHID.
>
>     >You also have to keep in mind that once an Enterprise gets large
>     enough,
>     >UPS monitoring isn't as important as you would think.  The reason
>     I use
>     >UPS monitoring in the Enterprise is because that way my
>     department gets
>     >notified when a battery in a UPS is toasted, instead of having to
>     depend
>     >on the notoriously unreliable users we have, letting us know
>     "there's
>     >this strange beeping coming from the closet"  But not much of what's
>     >plugged into most of my UPSes in the Enterprise are setup to shut
>     down
>     >on power loss - because in the locations where I have servers that
>     >really matter - we have generators.  And in the less critical
>     locations
>     >it's all network gear and it makes no difference if the site
>     loses power
>     >for the network to stay up because people can't get power for
>     >workstations and monitors and such.
>
>     >The truth is that the entrance of Cloud Computing is a much greater
>     >threat to the UPS market than any of this arguing over
>     protocols.  APC
>     >for sure makes far more money off the sale of BackUPSes than large
>     >SmartUPSes, to SOHO.  But, when the 2-10 employee business goes
>     100% to
>     >the cloud - who needs a UPS?  And when the home users go to
>     laptops -
>     >which just about all of them have save the hardcore gamers - who
>     needs a
>     >UPS?
>
>     >This is why I think that APC's board allowed APC to be acquired
>     in 2007
>     >by Schenider.  And it's not lost on anyone that Schneider
>     squashed UPSes
>     >into datacenter cooling systems.  They saw the on-premise server
>     market
>     >shrinking as SOHO exited it, and the datacenter market growing as
>     SOHO
>     >entered it, and they figured they could disguise the revenue drop
>     off in
>     >the UPS division by pairing it with revenue increase from the
>     datacenter
>     >cooling products.
>
>     >Nowadays when you buy a BackUPS you are basically buying
>     something made
>     >on the same assembly lines in China that are making CyberPower UPSs
>     >CyberPower has little presence in the US Enterprise market -
>     which is a
>     >very conservative purchasing market - and likely never will.  But
>     they
>     >use the same protcol that BackUPSes use - and that GoldenMate Pro
>     uses
>     >and all of the other no-name Chinese UPSes use - USB-UPSHID.
>
>     >I'm probably an anomaly in Enterprise purchasing anyway since I
>     know a
>     >lot of Enterprise IT directors just spec SMT for -everything-
>     including
>     >stuff (like switches) that can't even plug into a UPS monitoring
>     port,
>     >whereas I'll use BackUPSes for the non-server stuff.  And in the
>     real
>     >blue chip companies they don't even bother buying replacement
>     >-batteries- they just throw the entire UPS in the garbage 3 years
>     after
>     >buying it and buy a new one.  I learned a lot about how these
>     companies
>     >operate in a prior life with judicious dumpster diving....
>
>     >Ted
>
>     >On 11/28/2025 1:31 PM, Jeffery Small wrote:
>     >> Ted:
>     >>
>     >> Thanks you for all of the very useful information. Since you have
>     >> explained the change in APC protocol, I can stop banging my
>     head against
>     >> this problem. It would be nice if apcupsd was ever modified to
>     deal with
>     >> this, but I can expect it would be a difficult problem.
>     >>
>     >> I'm sad to see another great company like APC taken over by
>     Schneider and
>     >> made worse for the end user, similar to Oracle purchasing Sun
>     Microsystems.
>     >>
>     >>
>     >> Time marches on -- not always for the better.
>     >>
>     >> Regards,
>     >>
>     >>
>     >> _______________________________________________
>     >> Apcupsd-users mailing list
>     >> Apc...@li...
>     >> https://lists.sourceforge.net/lists/listinfo/apcupsd-users
>
>     _______________________________________________
>     Apcupsd-users mailing list
>     Apc...@li...
>     https://lists.sourceforge.net/lists/listinfo/apcupsd-users
>
>
>
> _______________________________________________
> Apcupsd-users mailing list
> Apc...@li...
> https://lists.sourceforge.net/lists/listinfo/apcupsd-users