Texas Instruments should deliver on its promise to enter the mobile TV chip market next year with its Hollywood chip, just in time for volume devices to start to ship for DVB-H services in the US.
Elsewhere mobile TV will be appearing in all sorts of formats from T-DMB in Korea and possibly parts of Europe, Qualcomm’s MediaFLO in the US, raw Digital Audio Broadcast, and DAB Enhanced Packet Mode (EPM) and already Japan’s ISDB-T is being delivered.
Texas Instruments seems to be focused on US, Europe and in the Far East just Japan, and its Hollywood chip is being built with both DVB-H and ISDB-T in mind.
Already Texas has confirmed that it will be getting the Lion’s share of the world’s first big mobile TV market, Korea through Samsung built mobile TV phones, but that’s only in terms of the host processor. In the actual radio technology Samsung is focused on T-DMB for its home market and DVB-H elsewhere and said in August that it would build both DVB-H and T-DMB into a single chip set.
But other early market suppliers such as DibCom, Frontier and Siano have targeting dual DVB-H and T-DMB chips in a single device, and have a substantial time to market lead over Texas Instruments.
And this week Nokia launched the worlds first DVB-H phone on the N92, but this couldn’t have been based on the Texas Hollywood chip, because that’s not even in samples yet and this is an announced product for delivery.
We caught up with Bill Krenik, Texas Instument’s Advanced Architectures Manager last week and asked if the Texas chip giant was in danger of missing the boat on mobile TV chips.
‘We announced Hollywood a year ago and began talking to broadcasters and operators. The Texas Instruments approach is to try to build an entire ecosystem, and whenever you are dealing with the mobile handset, these people need a solution that has good talk time, low price and a small board area, which in turn leads to a small form factor and it requires low power consumption. And in a mobile TV they are additionally looking for a decent amount of viewing time,’ Krenik summarizes.
‘Some of our competitors in this market have badly abused statements about viewing time,’ he warns, but he declines to give specific names.
‘They tend to make statements that don’t compare like with like. Their systems may only use 25 mWatts of power, but often statements we see don’t take into account the amount of power needed for the handset backlight display, and the power used when the host processor is decoding the signal? And if Multipath Forward Error Correction is used, they often don’t include the power needed for that. We do, so, as I say, we are often not comparing like with like.’
Most companies Faultline has spoken to claim two things, that they only use 25 mWatts and that they are the only company that can do this.
Krenik said, ‘We think that most players are headed for 25 mWatts, but that’s just to power their tuners. Hollywood will use up 30 mWatts, but this is all the power that you will need right across the device board.
‘We’re pretty sure that the Dibcom chip, which comes without a tuner, uses up more power than this when a tuner is added,’ he said.
‘We’re not worried about being later to market because Hollywood is going to be ideal for handset markets where the phone system already uses chips from our OMAP architecture which it is already capable of handling video decoding and display management. Everything else they need for mobile TV is on Hollywood.
‘OMAP already handles H.264 compression, the user interface and things like screen rotations, which some devices need for Mobile TV.
But is Texas Instruments too late to get in on the current round of what is common cited as 40 plus global mobile TV trials, with more being announced every day? The Hollywood chip is not expected until some way into next year.
‘We already have some prototype capability, and we are shipping samples of Hollywood at the end of 2005, so I don’t think our chips will arrive too late. Other companies have chosen to rush out an early prototype, but these are less sophisticated than the final chips will be.
‘Crown Castle is one of the most important US trials, and we know and work with them. They are focused on this market, and have spectrum in place.
‘When you look at a complete ecosystem you will find that our power consumption and board area used is competitive, especially given that we are building our parts in 90 nanometer CMOS,’ said Krenik.
‘Revenues and shipments will begin to be meaningful around Christmas 2006, and in 2007 shipments will be substantial. It usually takes about a year to go from samples to finished handsets,’ he concluded.
What will DVB-H cost to deploy? ‘That’s a challenging question, and all I can say is that cellular is dramatically greater. There is no need for backhaul, no uplink channel, just the broadcasting network and a satellite distribution system to get content to it.
‘Each cellular site will have a satellite receiver and will re-encode the signal into DVB-H. This is not an area that Texas Instruments offers products, however I am aware that our DSPs are often used by other companies within their products at the broadcasting end of the market.
‘Initially the early phones with DVB-H in will be at the high end of the market, but this will go to volume. Hollywood can enable an increase in Bill of Materials costs of just $10 to add cellular broadcast TV to a phone.
‘This will result in services accessible on phones that are midrange and upwards, with costs between $10 and $25 a month. Which should mean that a large percentage of the population will own phones that these services can be viewed on.’
Given those percentages Krenik seems to be saying this is set to become anything between a 200 million and a 350 million device per year market, once every market frees up some spectrum and gets to volume, which he expects in the 2008/9 time frame. And mobile TV could easily end up in 1 billion devices four or five years out, perhaps by 2011.
Will cellular TV in the US be a straight fight between Qualcomm’s MediaFLO and Crown Castle’s DVB-H? ‘We don’t think so. There will be other players, because there is additional spectrum coming onto the market and there is also existing spectrum that’s already been allocated. At the moment a single 5 or 6 MHz spectrum slice is not much. What this market will need is 30 or 40 MHz of spectrum in total so that a lot more TV channels are available.
‘There is always Sprint’s 2.5 MHz spectrum and there are others,’ he suggested.
‘In fact in total there is 200 MHz of spectrum available for these services in major markets, and once we get effective cognitive radios, this will make more spectrum available overnight.’
But is this a market that is stuck with QVGA screens and Standard definition TV in miniature. ‘I can see a time when we offer offpeak, overnight HDTV delivery. The battery is the only limiting factor right now.
‘You could have middle of the night network downloads which drops content into your phone, just like the morning newspaper.’
And finally will broadcast TV blow away all cellular delivered video services, once it is up and running. ‘In cellular networks using CDMA or other 3G protocols there will always be operator specific content but I believe that the Lion’s share of this market will go to broadcast, even when the 3GPP Multimedia Broadcast and Multicast Service (MBMS) arrives.’
MBMS is part of the 3GPP/3GPP2 set of standards for both W-CDMA and CDMA2000 and should be available from 2007. It allows a form of multicasting on CDMA which means that if lots of people are watching the same thing, only one copy of the video signal needs to be sent per basetation. Some people have suggested that this could undermine broadcasting as it doesn’t need additional spectrum or licensing as it sits on the conventional 3G network.