HomeMobile EuropeDevil in the detail

    Devil in the detail


    The challenge is on for UMTS handset manufacturers to keep the IC count down, as Howard Curtis, vice president global services at Portelligent, explains

    Analysts have paid considerable attention to the financial commitments that carriers made in spectrum auctions, and to the costs of introducing UMTS-capable base stations and networks, in forecasting the near- and middle-term prospects of 3G technology. Two further key components of the overall business equation that will determine when 3G becomes economically viable are the cost and complexity of UMTS handsets.

    Recent product teardown analyses conducted by Portelligent on three UMTS handsets designed for the European market — the NEC e-606, Motorola A830, and Nokia 6650 — reveal extremely high average values on several system-level metrics that are good indicators of relative complexity, and also good predictors of overall manufacturing cost. For example, on the dimensions of IC count, total silicon die area, and total component count, the UMTS averages are:

    IC Count: 68 devices
    Silicon Die Area: 9.86 cm2
    Total Component Count: 995 devices

    Comparable values for a representative EDGE-capable 2.5G handset of 2003 vintage are dramatically lower:

    IC Count: 13 devices
    Silicon Die Area: 1.93 cm2
    Total Component Count: 381 devices

    When estimated product manufacturing costs are calculated according to Portelligent’s cost models, the average total manufacturing cost of the UMTS handsets exceeds that of the EDGE example by more than 3x, while the average cost of IC devices in the 3G products is a whopping 5.3 times the EDGE case.

    UMTS versus FOMA

    A comparison of the UMTS handsets along these system metrics with the first generation of W-CDMA “FOMA” handsets that were introduced in Japan in 2001-2002 is also instructive. The first three FOMA handsets analyzed by Portelligent had an average IC count of 32, silicon die area of 10.45 cm2 (with a range from 7.07 to 14.64 cm2), and total component count of 727.

    It is worth noting that, in contrast to UMTS phones, which must implement both the W-CDMA and GSM protocols, the first generation of 3G handsets that NTT DoCoMo introduced in the Japanese market supported only W-CDMA (and the inability to communicate with pervasive 2G infrastructure hurt FOMA badly in the marketplace for the first year after its introduction).

    The average silicon die area is lower in the first-generation UMTS phones than it was in the early FOMA handsets in Japan, while both IC count and total component count are significantly higher. This can be attributed to achievements in feature-size shrinkage in the semiconductor industry over the intervening two years.

    Range in UMTS design

    Although an analysis of the average level of complexity and average manufacturing costs of the three UMTS handsets paints a grim picture for the near-term business prospects of 3G networks, these averages do hide considerable variation.

    On many of the cost and complexity metrics we assess, the Nokia 6650 falls far below the UMTS average as outlined above. IC count, for example, is 29, versus an average value of 68 (and by comparison with a whopping 108 ICs in the NEC e-606). Silicon die area measures 4.54 cm2 in the Nokia 6650, versus a UMTS average of 9.86 cm2.

    How did Nokia achieve this efficiency in designing the 6650? The Nokia handset shows much higher levels of IC integration that are specifically designed to meet the requirements of UMTS communications than either the NEC e606 or the Motorola A830, as well as an overall system design that is more efficient and better integrates the GSM and W-CDMA requirements.

    For example, the TI digital signal processor employed in the 6650 handles the baseband processing requirements for both protocols, as well as integrating an ARM core for applications support (these functions remain separate in the NEC handset). An analog ASIC produced by ST Micro handles the analog baseband requirements for both protocols, as well as providing multiple support functions including system power management, audio I/O, and the SIM card interface.

    Challenge for the future

    While the data transfer rates and potential feature sets offered by UMTS technology may exceed what EDGE can offer, Portelligent’s analysis of these three early UMTS phones indicates that the price is still steep in increased complexity and higher cost. The challenge to designers of 3G UMTS handsets is to achieve products that provide the consumer with the benefits of full 3G communications, but at substantially reduced levels of system complexity and cost. It will be very interesting to see how much progress the second generation of UMTS handsets reveal on these dimensions.

    Previous articleA process of power
    Next articleFlexibility is the key