Technical and business challenges of display integrated with touch

Vertical alignment in-cell displays
Let’s focus on this VA type of hybrid in-cell displays with shared VCOM and TX layer, and discuss the impact on the touch controller. There are two main issues: first, the RX, and especially TX, conductors are much closer to a layer with fixed DC potential. The TX lines are close to the traces that carry video data from the DDI to each of the pixel TFTs, and the resulting TX line capacitance can be up to 10 times higher than for on-stack sensors. Therefore, the TSC needs to have sufficient current drive to drive the TX lines to its full excitation voltage. Cypress touch controllers have an on-chip charge pump that can generate up to 10V output on these TX lines, from only a 3V external supply. A higher TX voltage is beneficial to increase signal-to-noise ratio (SNR): every increase by a factor of 2 in signal amplitude increases SNR by 2x, or 6 dB. 

The higher RC time constant of both the TX and RX in-cell sensor traces also reduces the maximum frequency range for the touch excitation signals. This reduces the range of possible TX frequencies that can be used to mitigate any discrete frequency spurious noise components, such as periodic noise from the LCD or the switch frequency of the DC-DC converter that drives the backlight unit.  Normally the TSC hops to a different TX frequency when there is too much discrete noise at the current TX frequency. If the useful TX frequency range reduces because of a ‘slower’ panel, it limits the degrees of freedom for the firmware to mitigate the effect of this external noise.

Because of the shared layer, there needs to be a time-multiplexing between using the same conductors for driving TX pulses and VCOM reference potential. VCOM needs to be driven during the ‘active’ video window so only non-active, or ‘blanking’, time is available for the TSC function. This leads to less time available to do a complete touch frame scan vs regular “on-stack” or even “on-cell” panels. Scan time is inversely proportional to channel bandwidth. Therefore a shorter scan time, and thus wider receive channel bandwidth, reduces immunity against broadband noise, such as from AC chargers connected to the phone. On the other hand, blanking time cannot be extended beyond a certain maximum; otherwise LCD image quality degrades. This trade-off presents a real technical challenge for in-cell touch.

The key point is that in-cell touch, as opposed to on-stack or even on-cell, creates a dependency between the touch and display functions. Therefore at least a signal interface is needed between display driver and touchscreen controller ICs. Display driver interface (DDI) ICs are supplied by specialized vendors that develop custom ASICs for a given panel resolution, panel type (a-silicon vs. low-temperature poly-silicon vs. AMOLED) and feature set. Touch controllers, on the other hand, are platform products that do not depend on display resolution or display type. The business model is radically different. Single-chip integration of touch and DDI is not required to support in-cell touch – it can be well served with a two-chip approach. Actually, for various reasons (time to market, choice of optimal silicon process for either part, multi-sourcing ability of DDI, etc) it seems best to keep both separate. If not, a smartphone OEM adopting in-cell touch across its product portfolio with different screen sizes and resolutions would need to qualify the touch function on many different ICs. With a two-chip solution, the same touch IC can be used across the portfolio. Figure 2 shows typical display resolutions and screen sizes in use today for phones and tablets.