Advertisment

The UPS Tests

author-image
PCQ Bureau
New Update

alt=" We used our HP 54603B digitizing storage oscilloscope to capture the waveforms. A trigger was fed to it through an isolator circuit (also designed at IDDC, IIT), and the waveform was later transferred to a PC using BenchLink for Windows"

align="right" hspace="5" vspace="5">Our test suite has evolved over

the years, based on changing user requirements and expectations. We also consider reader

and vendor feedback when designing the tests. This year, the test suite has remained more

or less the same as in the year before, with some modifications. As usual, we used the

three-axis model of price, performance, and features to judge the UPSs. To assign weights

to the parameters, we used a statistical method called the Brown-Gibson model.

Advertisment

The qualifying round

We had specified the following as

qualifying entry criterion for the UPS shootout: the UPS should be a currently-available

model rated at 400 VA or more, and priced at or below Rs 10,000.

Thus, before performing the evaluation, we

subjected all UPSs to a qualifying test round to knock out any that didn’t

live up to the 400 VA spec.

Advertisment

height="174"

alt="The test circuits, designed by IIT Delhi’s IDDC labs, measure voltage during switchover and backup time; the load circuit simulates a typical computer SMPS load but can be varied to the required load equivalent"

align="left" hspace="5" vspace="5">We connected a load of 248W at a power factor of 0.62

(equivalent to 400 VA), and checked how many UPSs were able to sustain the load when

switched to battery. We used the same simulated computer-SMPS load with every UPS. To

capture the waveforms, an HP 54603B digital oscilloscope running in single shot mode was

used.

Similarly, we subjected the UPSs to

anonymous price checks in the market, marking and rechecking suspect ones, and eliminating

those that did not qualify as below Rs 10,000 (street or list price, whichever was lower).

After the qualifying round, the following

parameters were measured for each UPS:

Advertisment
align="left" width="50%"> Droop Voltage Points Awarded

COLOR="#000000">More than 220 V (very good)

2

Below 220 V (poor)

0

Switchover

characteristic:
The droop voltage, the minimum DC voltage maintained at the electrical

load during switchover to or from battery, should be at least 200VDC
    Output voltage

variation
Points Awarded



COLOR="#000000">Less than 10%

2

10-15% (average)

1

Over 15% (poor)

0

Output

stability:
A good UPS should be able to keep its output voltage stable to within 10%

of 220 V, however much the input mains voltage varies
    STays on mains even

when mains voltage varies by
Points Awarded

COLOR="#000000">Less than 20% (best)

2

10-20% (good)

1

Over 10%  of 220V

(average)

0

Input

voltage tolerance:
A good UPS should be able to stay on mains without switching to

battery even when the mains varies by as much as 20% of 220 V

Performance

Switchover characteristic: This decides

how quickly and cleanly the UPS can switch over to battery, on a mains fault. If it

doesn’t do so fast enough, your PC could reboot or lose data.
This decides

how quickly and cleanly the UPS can switch over to battery, on a mains fault. If it

doesn’t do so fast enough, your PC could reboot or lose data.

Advertisment

While we have the equipment to easily

measure this switchover time in milliseconds, the results are not always representative of

reality with real-life PCs. We have thus arrived at another measure, which more accurately

represents the switchover characteristic of a UPS under a PC’s SMPS load: the

"droop voltage".

This is the minimum rectified DC voltage

that the UPS maintains at the load while switching from mains to battery or vice versa.

It is essential that this value remains above 220 V DC for the safe operation of a PC.

UPSs were awarded points based on whether they sustained this voltage during and after the

switchover (see table: Switchover characteristic).

To measure this "droop voltage",

we used a test circuit developed by IDDC Labs at IIT Delhi. The circuit cuts off the line

voltage supplied to the UPS, and a negative peak detector detects the minimum rectified

output voltage. The value is captured in the digital storage oscilloscope, and the

waveform recorded on our PC over a serial link.

Advertisment

Output voltage stability:

This test

measured and documented the output voltage stability of each UPS–the minimum and

maximum voltages the UPS gave when the input mains voltage varied widely. A well-designed

UPS should be able to keep its output voltage range within 10% of the line voltage of 220

V. This test

measured and documented the output voltage stability of each UPS–the minimum and

maximum voltages the UPS gave when the input mains voltage varied widely. A well-designed

UPS should be able to keep its output voltage range within 10% of the line voltage of 220

V.

Voltage tolerance:

This is the

maximum and minimum line voltage at which a UPS can supply power to your PC without

switching to battery. A good UPS should be able to stay on mains power without switching

over to battery, even when the line voltage varies by 20% of 220V. (See table: Voltage

tolerance.) This is the

maximum and minimum line voltage at which a UPS can supply power to your PC without

switching to battery. A good UPS should be able to stay on mains power without switching

over to battery, even when the line voltage varies by 20% of 220V. (See table: Voltage

tolerance.)

Backup time:

To perform this test,

we first measured the actual power consumed by two Pentium II machines with 15" color

monitors. This value came to an average of 185 W, which is approximately equal to 300 VA,

at a power factor of 0.62. A 400 VA UPS should easily be able to handle this load, and

give a backup for several minutes, at least. We thus set our test load to exactly this

value, throughout the backup timing tests for all UPSs. The greater the backup time for

this load, the more the points. To perform this test,

we first measured the actual power consumed by two Pentium II machines with 15" color

monitors. This value came to an average of 185 W, which is approximately equal to 300 VA,

at a power factor of 0.62. A 400 VA UPS should easily be able to handle this load, and

give a backup for several minutes, at least. We thus set our test load to exactly this

value, throughout the backup timing tests for all UPSs. The greater the backup time for

this load, the more the points.

Advertisment

This represents a change from last year,

where we set our load to 80 percent of the vendor’s claimed rating for each UPS.

Several readers asked us why we should vary our load based on the manufacturer spec of the

UPS, when users compare different UPSs for a fixed load, rather than changing their load

based on the UPS they buy! This is a very logical point of view, and we adapted our tests

to work with a fixed and consistent load across all UPSs.

A frequent reader question: why use a

simulated load rather than an actual PC? Several reasons, the top two being that a PC load

varies remarkably, depending on disk access, multimedia, display content, etc, while we

can freeze our simulated load at a constant value throughout. Also, we can adjust our

simulated load to a precisely calibrated value. However, our simulated load represents all

the characteristics of a real PC SMPS load, which is very different from those of table

lamps or other loads!

Cold start:

This is the ability of a

UPS to start up a load directly from the battery even when there is no line voltage

present. It comes in useful if you’re without mains power, but have to start up your

PC. Even when your UPS is on battery, this is far stressful for the UPS than a simple

switchover from mains to battery. Our test was done with the same load setting as used in

the backup test. This is the ability of a

UPS to start up a load directly from the battery even when there is no line voltage

present. It comes in useful if you’re without mains power, but have to start up your

PC. Even when your UPS is on battery, this is far stressful for the UPS than a simple

switchover from mains to battery. Our test was done with the same load setting as used in

the backup test.

Advertisment

Some UPSs had a separate switch that

provided the initial boot to cold-start the load.

Float voltage:

This represents the

charging voltage and is measured across a fully-charged battery with the UPS and charging

circuit "alive and active". This is usually the highest voltage you will find

across the battery. If the float voltage is too low, the battery will not charge

efficiently. If it is too high, batteries can overheat and this will reduce their life.

Lead-acid batteries are made of multiple 2-volt cells, so a 12-volt battery has six cells.

The float voltage should be about 2.17 to 2.30 volts per cell. This represents the

charging voltage and is measured across a fully-charged battery with the UPS and charging

circuit "alive and active". This is usually the highest voltage you will find

across the battery. If the float voltage is too low, the battery will not charge

efficiently. If it is too high, batteries can overheat and this will reduce their life.

Lead-acid batteries are made of multiple 2-volt cells, so a 12-volt battery has six cells.

The float voltage should be about 2.17 to 2.30 volts per cell.

This year we have taken the lower limit as

2.17 VDC because at this point the battery will be fully charged and this voltage would

not adversely affect battery life. This is also based on vendor and reader feedback.

bgcolor="#D2D2FF"> The Weights

FACE="Arial" SIZE="2" COLOR="#000000">Performance

Switchover

characteristic

21

Maximum output

18.5

Minimum output

18.5

Maximum switchover

voltage

13

Minimum switchover

voltage

13

Backup time

8

Cold start

5

Float voltage

3



Features  

Documentation

33

Indicators

27

Control software

20

Compactness

13

Others

7 Pricing  

Cost per VA/

backup-minute

70

Battery warranty

30

Price

This year, we set an upper price limit of

Rs 10,000 and invited the vendors to submit their best above-400 VA models in this range.

To calculate the cost-effectiveness of each UPS, we used two factors: cost per VA per

minute of backup, and battery warranty.

Cost per VA/minute of backup: The ratio

of the cost per VA to the backup in minutes was calculated. The list price of the UPS and

the VA rating given by the vendor was taken into account. However, we cross-checked the

prices of the UPSs in the market. We took the backup time which we got using our test load

for the calculation. UPSs with lower cost per VA/backup got better scores.
The ratio

of the cost per VA to the backup in minutes was calculated. The list price of the UPS and

the VA rating given by the vendor was taken into account. However, we cross-checked the

prices of the UPSs in the market. We took the backup time which we got using our test load

for the calculation. UPSs with lower cost per VA/backup got better scores.

Battery warranty:

Some UPSs come

with different warranties for the batteries and the UPS. We considered the lower of the

two. The higher the battery warranty, the higher the score; the UPSs were rated relative

to the highest scorer. Some UPSs come

with different warranties for the batteries and the UPS. We considered the lower of the

two. The higher the battery warranty, the higher the score; the UPSs were rated relative

to the highest scorer.

Features

Documentation: A good UPS manual should

at least contain information about the UPS, how to set it up, and a troubleshooting

section. If a UPS manual contained all these, it got a 1, otherwise it got a zero.
A good UPS manual should

at least contain information about the UPS, how to set it up, and a troubleshooting

section. If a UPS manual contained all these, it got a 1, otherwise it got a zero.

Indicators:

We rated each UPS with

respect to the information and controls available on the front panel. The ones that

provided more information and had better controls like battery and load conditions were

given 2. The rest of them were given 1. We rated each UPS with

respect to the information and controls available on the front panel. The ones that

provided more information and had better controls like battery and load conditions were

given 2. The rest of them were given 1.

Control software:

If the UPS came

with optional software it got a 1, else a zero. For those that came with software that was

part of the package, an additional point was given. If the UPS came

with optional software it got a 1, else a zero. For those that came with software that was

part of the package, an additional point was given.

Compactness:

Here we rated each UPS

on a relative basis. The most compact ones were given 2 points whereas the bulkier ones

were given 0 points. Mid-sized UPSs got 1 point. Here we rated each UPS

on a relative basis. The most compact ones were given 2 points whereas the bulkier ones

were given 0 points. Mid-sized UPSs got 1 point.

Others:

In this category, we

considered all the extra features that came bundled with the UPS. These included features

like provision for extended backup, a telephone/fax protection port, etc. If it had extra

features, it got a 1, else a 0. In this category, we

considered all the extra features that came bundled with the UPS. These included features

like provision for extended backup, a telephone/fax protection port, etc. If it had extra

features, it got a 1, else a 0.

Advertisment