EEVblog #1168 - Dumpster Diving UPS

[max_torque]

What is the architecture that this unit uses?

The FETs shown are only rated to 150V, so either are used for the Dc/DC to the battery only (36v nominal) or the large transformer is used to step up/ down to the A/C voltage and all the "switching" happens at the low voltage level?

It would be a nice little donor unit for the basis of a home PV / Battery project....

[Unixon]

How about blown power traces at the bottom side of the board? It's a weird case for an UPS but...

[TruslowPJ]

Those heatsinks.
"Should we place an extruded aluminium HS?"
"Nah. Lets just place there a chunk of metal cut to size by a saw."

I'm not sure if this is over or under engineering.

Those heatsinks were designed for thermal mass, not for dissapation to ambient. they only need to sink the amount of heat generated when running at full load for the duration of the batteries before the aluminum is too hot. the inverter is only intended to run for 5-6 minutes at full load. UPS units that are intended for extended battery operation will use finned heatsinks, but units only intended to run as long as the internal batteries last are often designed with large block heatsinks instead.

[Pentium100]

I expected to see some huge capacitors in that UPS in order to maintain output when the mains goes down and relays are switching (which take some time). How does it work then ? Is that job done by huge inductor seen in the video ?

A capacitor won't really help with AC output. Anyway, line-interactive UPSs have a switching time of a few milliseconds - the capacitors in the power supplies of the load are supposed to take care of that.

The FETs shown are only rated to 150V, so either are used for the Dc/DC to the battery only (36v nominal) or the large transformer is used to step up/ down to the A/C voltage and all the "switching" happens at the low voltage level?

Yes, the FETs create a PWM sinewave (or a square/pulse wave for cheaper models) that then gets stepped up by the big transformer. When there is power coming from the wall, the big transformer steps down the voltage and the FETs rectify it to charge the batteries.

[floobydust]

Those heatsinks.
"Should we place an extruded aluminium HS?"
"Nah. Lets just place there a chunk of metal cut to size by a saw."

I'm not sure if this is over or under engineering.

Those heatsinks were designed for thermal mass, not for dissapation to ambient. they only need to sink the amount of heat generated when running at full load for the duration of the batteries before the aluminum is too hot. the inverter is only intended to run for 5-6 minutes at full load. UPS units that are intended for extended battery operation will use finned heatsinks, but units only intended to run as long as the internal batteries last are often designed with large block heatsinks instead.

Heatsinks as a brick, a thermal mass to heat up- a strategy by a particular UPS manufacturer whom we all know.
Here it's shit engineering. Forced-air cooling with the little fan doesn't remove much heat from a brick. Spend the pennies on some fins already

At 1,500W and 90% efficiency, there is 150W of heat to get rid of for say 30 minutes. She'll run pretty hot.
I don't think this UPS is much above residential grade or worth investing in new batteries.

[Towger]

There was what looked like a white rack shelf on the floor in the dumpster room beside the UPS.  They are alway nice to get and for what they are surprising expensive to buy.

And Yes, there needs to be power in the batteries to turn on most UPSs.

Sent from my GT-I8200N using Tapatalk

[Pentium100]

At 1,500W and 90% efficiency, there is 150W of heat to get rid of for say 30 minutes. She'll run pretty hot.
I don't think this UPS is much above residential grade or worth investing in new batteries.

At full load it's probably 5-10 minutes.
Also, IMO, a set of new batteries is always cheaper than a set of batteries and the rest of the UPS, so I usually take unwanted UPSs (in some cases broken ones that I then repair) and get some new batteries. Even if the UPS is really cheap, it's better for me to get it for free and buy new batteries, as I can buy higher quality batteries for the same amount of money.

I have taken a couple of broken cheap UPSs from my workplace (with permission of course), repaired them and am now using them.

[floobydust]

UPS have so much hyped specs to fool the IT guys.
Rated 1500VA/1000W for 5 minutes, 800W for 8 minutes and 200W for an hour. It's main feature is small size, only 1U of rack space. But over 50lbs!
CSB HRL 634W batteries are 6V 34W (hr?) at 1.67V/15 minute rate. Six of them spec would be 654W/15 minutes. But HP spec is 500W for 17 minutes.

The manual says the unit runs without batteries, it will show an error on the cryptic,confusing user interface.
I can't imagine trying to figure out WTF is going on with this thing during outage panic 
R1500 G3 user Guide

Guessing around $800 new and a battery set is ~$110 plus elephant shipping costs.
I keep some old UPS after recapping them and replacing the batteries.
For repair, I would suspect the small SMPS supplying the 7805 is dead. Perhaps an 8-pin IC doing that work.

[rigrunner]

This one is a hot swappable battery, so it's designed to be disconnected and still function.

I have a UPS here that is hot swap but it must have good batteries in place to power up.
Once it's powered up i can remove the batteries and it stays powered on..

[Pentium100]

Rated 1500VA/1000W for 5 minutes, 800W for 8 minutes and 200W for an hour. It's main feature is small size, only 1U of rack space. But over 50lbs!
CSB HRL 634W batteries are 6V 34W (hr?) at 1.67V/15 minute rate. Six of them spec would be 654W/15 minutes. But HP spec is 500W for 17 minutes.

The manual says the unit runs without batteries, it will show an error on the cryptic,confusing user interface.

but 20
1U size is important, weight is not as much. In some places you may pay for every rack unit you use. Or if it's your own rack in your own server room, you can place more servers in that rack because the UPS is smaller.

The unit may very well run without batteries, but need batteries to power up from cold. After it powers up, you can disconnect the batteries.

[tigrou]

A capacitor won't really help with AC output.

I was thinking about a capacitor between rectifier and inverter, dealing with DC. Just like the batteries.

Anyway, line-interactive UPSs have a switching time of a few milliseconds - the capacitors in the power supplies of the load are supposed to take care of that.

So basically what you say is that UPSs rely on the fact most loads (a SMPS in a computer) can continue working properly even if power goes down for a very short period (because of capacitors on the main side, after rectifier ? eg : the 450V one)

[aqarwaen]

dave do you have any dead tablets,what got dead battery?i would measure batteries and salvage batteries,to get free new battery for tablet computer...............

[rrinker]

 Indeed, I've gotten many UPS units free or at low cost because all they needed was a new battery. I am not far from Batterymart.com which has all manner of SLA batteries (and plenty of other stuff) so I can choose their cheapest shipping option (batteries are heavy!) and get them in a day or two, and they are a fraction of what APC or others charge for the exact same battery.
 One I got for free, a smaller unit, was brand new! The user who opened it up and hooked it up failed to notice the BIG TAG on the front that said to pull the cover off and PLUG IN THE BATTERY first. Yes, absolutely nothing wrong with it, no repair or new battery needed, just connected the battery and it was good to go.
 The cost posted above for a local source for the batteries in AUD is FAR less expensive than that UPS would cost new, entirely worthwhile if there is nothing else seriously wrong with it (well, if you actually need a UPS for something).

[thm_w]

I don't recall having ever owned a well built UPS that didn't require battery voltage to turn on. It makes sense from a safety aspect. If something happens to make the unit fail or the batteries become useless it takes itself out of service once it no longer has power to hold down the line relay. This is how I've gotten many ups's for free or close to it. 9/10 times they just need a fresh set of batteries, there's nothing actually wrong with the unit, even though they refuse to turn on.

Its a cost saving measure if that were actually the case, as they can run the circuitry off the batteries and won't need an additional off-line switching supply.
But as floobydust posted, the unit should turn on and show an error indicator. Off course it would not enable the output if there is any error, meaning it is still safe. No reason to not power up at all in the name of safety.

[floobydust]

https://youtu.be/Fj7e3WGUKO8

I can't find any schematics or service manual for the HP R1500's.  Too bad, it would help make these repairable and out of the e-waste.
My hunch is APC did the design and build for it, it has their reliability lol.

[TruslowPJ]

HP announced years ago Eaton was their partner for UPS systems. https://www.eaton.com/Eaton/OurCompany/NewsEvents/NewsReleases/CT_065786
It appears that the R1500 G3 INTL is a rebrand of the Eaton 5115. front panel and I/O matches https://www.alliedelec.com/m/d/c4ac6d3bb8731d739d1fc1c6e6cbc555.pdf

[floobydust]

I can't keep up with the acquisitions. Eaton bought Powerware in 2004. Schneider bought APC for $6.1B 2006.

It certainly looks like a Powerware 5115 1500i RM (Australia).  But no luck finding the service manual.
I looked at other schematics of Powerware UPS and they are a simple architecture. The 7805 gets powered from the batteries. The little ferrite transformer might be a CT, not a SMPS.

Dave could connect a 36VDC power supply in place of good batteries and see if the MCU then comes up.
Then the PSU gets bumped up to 42VDC "charging". Then we'll know if it needs batteries to boot up.

[thieringpeti]

Hi!

I have two of these, from a fab parts auction. I have the original Eaton 5115, but the smaller, 24V version with four batteries. The internals are the same, even the battery drawer. Only the label is different. One of them was renewed with a fresh set of Panasonic batteries, and now it's operating well. The other one was tested, and put in storage for spare, without batteries. I can confirm, it won't power up without a battery set. The only con is the bundled software. Eaton's Powerware suite software is a bit of crap, it monitors the UPS only for 15 minutes, then it says UPS is offline. I haven't tried it with XP which the UPS designed for, and also Windows 7 don't have the drivers for it, (unknown usb device), so only that downloadable crap PM software can be used from Eaton's site to monitor the state.

The two loads can be swithched separately, I tested that functionality. As You can see, at the connector, there's a fourth black wire hiding under the blue neutral on Your video. With the software suite, you can specify, that on power loss, it turns off first the least important stuff, to a save battery life before it shutting down finally, e.g. shutting down first the web server, then the sql server, or first the monitor, then the PC. You can also set the shutdown timers and treshold voltages from the software.

[ve7xen]

So basically what you say is that UPSs rely on the fact most loads (a SMPS in a computer) can continue working properly even if power goes down for a very short period (because of capacitors on the main side, after rectifier ? eg : the 450V one)

Exactly. It's typically less than a single AC mains cycle (2-4ms for this unit), so most types of power supplies should tolerate it. If your load can't tolerate the switching time, you need an 'online'/'double conversion' UPS rather than a 'line-interactive' one. This type would always run the load from the inverter, using the line to power the inverter and charge the batteries, rather than power the load directly, so the batteries can pick up the load instantly on loss of line power.

[thieringpeti]

So the may be a killer for this UPS... 3 identitical car batteries in series would be a nice solution for extended runtime... So that 7805 should be checked if bigger capacity batteries would be used...