Visualizing a UPS output wave

[robdejonge]

Just bought a new UPS for my home office. The old unit has a "simulated sine wave" and the new unit advertises a "pure sine wave". So, just for fun, I would like to see those sine waves on the screen of my oscilloscope.

I have a GW-Instek GDS-1072b oscilloscope. According to the datasheet, it can handle 300Vrms. I'm in Thailand, where the grid provides ~230V so I think that should be ok.

What I know is that I must disconnect the UPS from the wall, to ensure it's output is floating and of course to make sure I get the output from the battery rather than what is passed through from the grid.

Maybe the probe needs to be set 10:1?

What else should I be thinking about?
Is it safe to just be hooking this up?

It's not exactly critical analysis, but I thought it would be fun to see!

[artag]

Maybe the probe needs to be set 10:1?

Yes, the data sheet actually specifies 300V WITH the 10:1 probe. Besides, a 1:1 probe is practically useless for all but a handful of cases. See Dave's video on the subject. Set your probes to 10:1 and put tape over the switch.

You're right to disconnect the mains source, but do connect the earth. Otherwise earth will be provided from the scope and if you connect that to one of the pins you'll still  have 110V or whatever on the UPS output just as if it were connected to power. If you don't understand how the earthing is being wired, you probably need to get somene to help you do it safely.

Worst case, you might shortt the output of the UPS at 110V to earth via the scope AND the UPS. This won't be good for either. Actually that's not the worst. Do it wrong enough and you could end up with exposed metalwork on scope or UPS at 110V. that isn't good for you.

[pqass]

If you wanted to just see the wave shape then I suppose you can just put a string of high-value resistors across line and neutral UPS outputs and connect the probe across only one of the resistors.

Eg.  3 * 100K + 10K resistors (in series) across 230VAC gives 0.75mA with 171mW total across all four.
Probe (10:1) across ONLY the 10K resistor gives only 21Vpp (7.4Vrms).
KEEP all resistors FIXED to a protoboard (which is clamped to your bench) to minimize any accidental movements.
Place 100K resistors in an insulated sleeve; only expose the 10K resistor leads.

DON'T CONNECT ANY DEVICE TO THE UPS RECEPTICLES.
KEEP UPS ON AN INSULATED SURFACE.
ONLY CONNECT PROBE WHEN UPS IS FULLY DISCONNECTED FROM MAINS (NO PROTECTIVE EARTH EITHER)!
DISCONNECT PROBES BEFORE RECONNECTING UPS TO MAINS!
KEEP ONE HAND IN YOUR POCKET!

[donlisms]

230 volts RMS is... what... 650 volts peak-to-peak?

[pqass]

230 volts RMS is... what... 650 volts peak-to-peak?

Indeed it is.

[Vovk_Z]

TS may use any step-down power transformer he has as an isolation ttansformer - e.g. 230/12, 230/24 etc..

[pqass]

TS may use any step-down power transformer he has as an isolation ttansformer - e.g. 230/12, 230/24 etc..

It would be interesting to see if the transformers output waveform differs from the original.
I would imagine a rectangular "stepped sine wave" would get smoothed out.

[radiolistener]

I have a GW-Instek GDS-1072b oscilloscope. According to the datasheet, it can handle 300Vrms. I'm in Thailand, where the grid provides ~230V so I think that should be ok.

you're needs to be careful. 230V mains means that it's peak voltage is higher you can get value by multiplying with sqrt(2) for sine wave.

Since 230V mains wall outlet allows up to 253V (and even up to 600V for a short period of time), it means that peak voltage on the mains outlet can be 253*sqrt(2) = 358 V and even more.

So be careful, if your oscilloscope rating is 300 V, it will be burned out from mains. If your oscilloscope rating is 300 Vrms it can survive, but there is possible high voltage pulses on the mains which can damage your oscilloscope.

[bdunham7]

The datasheet on this scope says it is CAT I/300V_RMS without a probe and CAT II/300V_RMS with a specified probe.  There's no worries about peak voltages and such.  A UPS not connected to mains seems to be CAT 1 to me and there isn't any likely issue with high energy transients.  Of course even though the scope is protected to 300V without a probe, you won't be able to look at the signal because you can't reduce the vertical gain enough and it will be clipped or off the screen.

 The output of a UPS in backup mode (off-mains) should be isolated from ground, but I would still be careful not to touch it or allow it to touch anything conductive during the test.  If you just want to see what it looks like, a step-down transformer will work more or less fine, you won't lose much signal integrity and there's a bit less risk of a mishap.

[Vovk_Z]

It would be interesting to see if the transformers output waveform differs from the original.
I would imagine a rectangular "stepped sine wave" would get smoothed out.

Any unloaded transformer work as voltage transformer quite fine. It will not smooth anything. It typically has a frequency range up to 10-20 kHz. That's enough if want just to look at a mains voltage waveform. 

[boB]

If you want to see some funny looking waveforms (~sinewaves) check out non-inverter based small-ish home generators !

They are usually awful, and not just a square wave.

boB

[BrianHG]

TS may use any step-down power transformer he has as an isolation ttansformer - e.g. 230/12, 230/24 etc..

It would be interesting to see if the transformers output waveform differs from the original.
I would imagine a rectangular "stepped sine wave" would get smoothed out.

Small transformers, like 250ma and below, will have enough bandwidth to let a square stepped waveform through.

You need a large iron core to begin to smooth out the square wave, however, it still shows through.  Think something like a 10amp transformer.  Though, these will also buzz like mad.

[pqass]

Curiosity got the better of me so I connected a 16VAC 20VA transformer to my UPS (APC Back-UPS XS 1300 LCD).
See attached for the mains vs. the "step approximated sinewave" produced when on battery (with a 160W load).
I can definitely confirm that the sharp rectangles are coming through.  Ick!

[bdunham7]

Curiosity got the better of me so I connected a 16VAC 20VA transformer to my UPS (APC Back-UPS XS 1300 LCD).
See attached for the mains vs. the "step approximated sinewave" produced when on battery (with a 160W load).
I can definitely confirm that the sharp rectangles are coming through.  Ick!

What are you using for a load?

[pqass]

What are you using for a load?

Mainly my office PC, monitor, oscilloscope, bench meters, network switches, DSL modem, phone, LED desk lighting, etc; mostly switching power supplies.

I think it's a mangled form of the attached red trace. 
Sorry, I can't be arsed to disconnect everything from the UPS and re-test without any load.

[5U4GB]

Another thing to do is look at the output waveform with various levels of UPS battery state.  Years ago there were UPSes on the market whose output started out as a stepped square wave but got closer and closer to a pure square wave as the batteries ran down.

After seeing that waveform I'd never buy anything other than a pure sinewave output UPS.

[vk6zgo]

230 volts RMS is... what... 650 volts peak-to-peak?

Indeed it is.

Peak to peak is convenient to use when reading voltages with an analog 'scope & a few other things, but as the two peaks are not coincident, the greatest voltage across the DSO's probe will be 325v peak!

[CaptDon]

Comment about small home generators, non-inverter class. I have refurbished many of them for friends. Carbs are always gunked up from non-use and stale gasoline with no long term stabilizer added. Anyway, as for the waveform. I have not run into any that were 'horrible'. They do tend to include some even harmonic energy and you can see a kick in the sinewave during the power stroke of the engine and a closer look at the sinewave can show if you are looking at a single cylinder or twin cylinder engine and if it is a single cylinder engine you can tell if it is a 2-cycle or 4 cycle engine.

[Jason Henry]

Ensure UPS is disconnected from the wall, set the oscilloscope probe to 10:1, and safely observe the sine wave output. Confirm the voltage range is within the scope's limit (230V) for a fun and safe analysis!

[cosmicray]

Keep in mind that the behavior of the load may not be resistive. Even if your electric source is capable of putting out a pure sine wave, some load types (SMPS in particular) may have a very non-sine wave current consumption. Put a amp-clamp probe around one of the wires to the load, then watch the current waveform on your DSO.

[MrAl]

Curiosity got the better of me so I connected a 16VAC 20VA transformer to my UPS (APC Back-UPS XS 1300 LCD).
See attached for the mains vs. the "step approximated sinewave" produced when on battery (with a 160W load).
I can definitely confirm that the sharp rectangles are coming through.  Ick!

Hi,

That bottom trace does not look like a pseudo sine wave produced by these cheaper UPS systems, so I do not think you are seeing what is really there.  The wave shown has some reactive components associated with it so it's probably an artifact of the transformer.  Transformers can draw unusual currents and produced unusual outputs when they are not loaded so you might try loading that transformer with a resistive load and see if the waveform clears up.  A typical UPS system will show more of a regular rectangular waveform that is like a square wave but with narrower pulses like that shown in one of the other replies with the red colored plot.

With a pure sine converter, you may not be able to see the pulses because there may be some filtering before the output receptacle.  You might just see what looks almost like an actual sine wave, perhaps with some small bumps where the pulses start and small dips where the pulses end.  The pulses will be shorter near the zero crossings and wider near the peaks.

[pqass]

So I shed all but 16W of load (except network switches, firewall appliance, DSL modem) and the waveform looked just like the lower shot on my #12 reply (this time with 680R across xformer secondary); with the same ringing but the duty cycle appearing to be <50%.

Then I removed ALL loads except the 20VA transformer with 16VAC secondary loaded with a 680R resistor and 'scope probe across the resistor.  The ringing is now gone but there is still a low sine wave superimposed on an alternating square wave (approx 36% duty cycle).  See attached. 

[bdunham7]

Can you safely measure it directly without a transformer?  A 10X probe properly used should suffice.  Perhaps transformers (or some transformers) aren't as suitable for this purpose as I'd assumed.   Perhaps the output filters or other circuits of the UPS are interacting with it.

[pqass]

Can you safely measure it directly without a transformer?  A 10X probe properly used should suffice.  Perhaps transformers (or some transformers) aren't as suitable for this purpose as I'd assumed.   Perhaps the output filters or other circuits of the UPS are interacting with it.

Although the label next to my scope input says 150Vrms, I opted instead to probe (10X) across the 10K resistor in this divider arrangment: L-100K-10K-100K-N.   I'd like to keep using it.   

No other devices (other than the resistor divider) were connected.  See attached waveform.
The result is not much different from the transformer output; other than the amplitude (10K/210K*120Vrms=~6Vrms), and flattening of the mid-point between alternating square waves.

[bdunham7]

Can you switch your scope to DC coupling?  It shouldn't matter at 60Hz but maybe something weird is going on here.

[pqass]

Can you switch your scope to DC coupling?  It shouldn't matter at 60Hz but maybe something weird is going on here.

Bingo!  It was the AC coupling that introduced the wavy weirdness.  See attached.

1st shot: no load, DC coupling, 10X probe across 10K resistor in the L-100K-10K-100K-N divider,
2nd shot: same as above but with 120W load,
3rd shot: 120W load, DC coupling, 10X probe across transformer open secondary.

Note:
(a) alternating square wave seen as expected; see red trace in attachment from my reply #14,
(b) ringing when under load,
(c) duty cycle increases under load,
(d) slightly rounded tops in 2nd but may not be noticeable in 3rd, and
(e) transformer doesn't appear to mangle the 60Hz waveform.

Okay then. I think I've beat this topic to death enough. 

[5U4GB]

Okay then. I think I've beat this topic to death enough.

I think you've created the definitive answer to any question of "should I pay extra for a proper sinewave UPS", thanks for putting in the effort.

Having said that, it'd be interesting to know which UPS produced that output, not for name-and-shame but just to know that product X from vendor Y behaves like this.  In particular it's to counter the obvious response "well it must have come from a cheap no-name UPS, no true Scotsman... ah, UPS would do that".

[Vovk_Z]

Perhaps transformers (or some transformers) aren't as suitable for this purpose as I'd assumed. 

Most transformers are typically suitable (at least those <= 100..200 VA, unloaded or lightly loaded with a pure resistive load) . 

[pqass]

Having said that, it'd be interesting to know which UPS produced that output, not for name-and-shame but just to know that product X from vendor Y behaves like this.  In particular it's to counter the obvious response "well it must have come from a cheap no-name UPS, no true Scotsman... ah, UPS would do that".

The product is from APC: Back-UPS XS 1300 LCD
Back-UPS is their economy line and is probably no different than other brands in this category.
The square wave output is not unexpected and has been perfectly fine these last 16 years.
Especially since I've replaced the internal 7Ah batteries with an external 18Ah pack, it should be cheaper since I won't be replacing batteries as often.

[MrAl]

Hi,

The waves being shown now look much more typical.  The duty cycle should vary a little depending on load, but they should look more rectangular than anything else, and they appear to be looking like that now.

This is very typical with the pseudo sine converter which seem to go by other names also such as "modified" sine converters.
These seem to work for some things, but if the device being powered uses a power factor correction circuit it may not work very well.

I have a little converter that puts out that kind of rectangular wave.  I tried it with an old style wall wart and it worked, but I did not have it running for long.

Rectifier circuit draw an unusually short current pulse from the line anyway so if the device being powered just has a full wave rectifier inside it may work ok.  With an automatic power factor correction circuit though it would expect a real sine wave to power it.

[5U4GB]

Especially since I've replaced the internal 7Ah batteries with an external 18Ah pack, it should be cheaper since I won't be replacing batteries as often.

If the UPS isn't set up for higher-capacity external batteries then you may want to be careful with that, a lot of them aren't designed to run at full load for extended periods of time, if you extend the runtime from (quick Google) the rated 5, to 15 minutes, you may end up cooking the electronics.  The ability to add external batteries on the more expensive models isn't purely a value-add, it's because the electronics aren't up to handling longer runtimes.

[pqass]

If the UPS isn't set up for higher-capacity external batteries ...

I'm well aware of that.  The time until shutdown hasn't changed. 
The larger external pack was used because the normal 7Ahr pack gets massacred with >10A draw for just a 170W load (>1C).
It's like they're designed to fail every 3 years.  Hmmmm.
With the 18Ahr pack, the same load will be just 0.55C which should translate into far fewer replacements and <$$$.