Notes for Procurement - Real world power use

Today I wanted to discuss the topic of power use of digital media equipment. It’s one of those areas that has become quite the focus over the last couple of years, with various manufacturers making bold claims about their efficiency and comparing positively to their peers, with stands at AV conferences and exhibitions having various eco credentials at the forefront of their design. 

Power usage of AV equipment, especially displays, has not traditionally been too much of a focus for buyers. With smaller scale networks of digital signs, the costs are relatively small compared to the overall operational overheads. As the scale of media networks has increased, however, particularly those that face outdoors through store windows, the long-term costs of powering a network of screens have grown considerably. 

First off, some basics. Displays use power mainly to create light with heat as a byproduct of an imperfect conversion. The more heat generated by a screen compared to its brightness, the less efficient it is at creating light output from a given input of electricity.  

LCD screens illuminate a white LED backlight behind the LCD panel, whereas LED displays light each pixel individually. This is inherently what makes direct view LED displays more efficient than an equivalent LCD – you are only lighting the pixels that need to be lit at the intensity required, rather than blocking the full intensity of white light to create the various shades of pixels. 

So how do manufacturers make their displays run more efficiently and what about their bold power use claims? It comes down to a few things:

• More efficient power supplies and other components create less heat and therefore use relatively less power for the same performance

• Local dimming of backlights – turning portions of the backlight of an LCD display off or down when the image processor sees that light is not needed (i.e. for dark or black areas of content)

• Ambient light sensors – allow for turning down the light output when it’s not needed, when the ambient light in the installed environment is low

Efficiency costs money in the electrical world, so you generally find that more expensive displays use less power for the same performance. In specification terms, manufacturers generally quote maximum and typical power use figures, but there’s no standard for comparisons with peers. This is where the issue lies – one manufacturer’s approach to labelling is not the same as there are, as yet, no regulations for commercial displays. This means that manufacturers can use whatever stats they like to out-punch their competitors on paper. If a manufacturer bases their figures on energy-saving out of the box settings, using ambient light sensors and local dimming, you’ll naturally see a lower ‘typical’ use figure.

In the real world, those out of the box settings are probably not going to be what you’ll want to use for a commercial digital signage network. For starters, you can’t really turn off the SOC processor out of trading hours, as this will negate your ability to apply updates and content. You’ll probably also not be so happy with the vibrancy of the default picture when the display is installed in a typically brightly lit retail store. 

This is not to say that the specs are useless – they should simply be treated as a starting point and, ideally, measured in a real world setting before committing to purchase. In our own testing, we find a definite correlation between the relative performance quoted by the big-brand manufacturers, but the exact figures are not the same as those given in the specifications, which can make a big difference to your total cost of ownership calculations over 5 years. 

A last word of caution – specifications from manufacturers you’ve not heard of in the consumer world are to be treated with care and should absolutely be verified. The big brands abide by European and global regulations, those not so well known can fly under the radar, reducing their in-costs by using lower grade components that are inherently less efficient and then quoting unverified data on their specifications.