Drivers for LED displays

Abstract

The present invention provides a method, a digital driving circuit, and an LED display, wherein an image to be displayed in successive frames is converted from an n-bit digital image into an (m + 1)-bit digital image by truncating (n - m) least significant bits or most significant bits of the n-bit digital image for each pixel, where m < n, and an indication is encoded in (m + 1) bits, and if (n - m) least significant bits are truncated, a first driving current is provided to each pixel of the (m + 1)-bit digital image, and if (n - m) most significant bits are truncated, a second driving current is provided to each pixel of the (m + 1)-bit digital image, wherein the ratio of the first driving current to the second driving current is such that it results in a brightness ratio of 2^(n - m), and wherein the indication provides information regarding the driving current to be used.

Description

Technical Field

[0001] The present invention relates to the field of displays such as LED or OLED displays, as well as methods of making or operating such displays, and software for performing such methods. Background Art

[0002] The problem of realizing high dynamic range displays and light emitting devices is known in the art.

[0003] US6987787B1 describes an LED brightness control system for wide range brightness control.The brightness of a light emitting diode (used as a backlight for a liquid crystal display) must be controlled over a range of at least 20,000 to 1.

[0004] US6987787B1 describes an LED control system in which the duty cycle of the PWM signal is modulated simultaneously with the amplitude of the current pulses. Using 8 bits to encode the duty cycle and also using 8 bits to encode the amplitude of the current pulses gives a total of 65536 brightness ranges.

[0005] Modulation of both the duty cycle of the PWM signal and the amplitude of the current pulses will allow for smaller brightness steps at lower brightness levels and larger brightness steps at higher brightness levels.

[0006] US6987787B1 does not mention how to maintain the ability to control luma over a range of at least 20,000 to 1 while addressing bandwidth constraints (which would require encoding luma on less than 16 bits).

[0007] In particular, US6987787B1 does not discuss how to prevent visual artifacts that may be caused by variations in the brightness step per LSB of the brightness.

[0008] This is not a concern for the LEDs of the backlight system since they are not used as pixels and US6987787B1 does not provide a solution for driving the individual LEDs of an LED display with a reduced bit depth while maintaining the brightness range and without introducing visual artifacts.

[0009] In US Pat. No. 8,339,053, an “LED dimming device” is described, which utilizes two dimming mechanisms to control the brightness of an LED lighting device.

[0010] In the first, "lower brightness" regime, the current through the LED is pulse-width modulated using a constant current pulse amplitude modulation. In the second, "higher brightness" regime, the current through the LED is controlled in an analog fashion and is not pulsed. The current through the LED is continuous, and its amplitude is determined by a constant current circuit.

[0011] US8339053 does not provide a viable solution for driving individual LEDs of an LED display.US8339053 does not discuss the problem of visual artifacts, in particular color artifacts that inevitably exist when driving LEDs with different current amplitudes.

[0012] To further illustrate the shortcomings of the prior art, consider an LED display with a brightness range of greater than 20,000 to 1. Such a high brightness range requires at least 16 bits per LED.

[0013] The pulse width modulated current is used to drive LED or OLED, 16 -1 level of brightness corresponds to 2 16 -1 different duty cycles D. The lowest possible duty cycle would correspond to a current pulse whose period is 1 / (2 16 -1).

[0014] To prevent flickering, you can choose to split one frame time (interval) into 64 intervals.

[0015] The duty cycle increases by 1 / (2 16 -1), the period is 1 / (2 16 A sub-pulse of -1) is added to one of the 64 intervals. Each of the 64 intervals can accommodate a maximum of 1024 such sub-pulses.

[0016] If the period of one frame is 1 / 60 s, the period of one sub-pulse is 254 ns.

[0017] Currently available LED drivers typically have 16 outputs per color. An example of a current LED driver is the Texas Instruments TLC5940-EP, a 16-channel LED driver with dot correction and grayscale PWM control. The latest LED drivers can drive 16 RGB LEDs. Barco nv Belgium's X1.6 LED chip has a 120x135 LED array. This means more than 1013 drivers are required. From a printed circuit board design perspective, it is nearly impossible to place and route this high-current LED driver on the back of the LED chip. This situation worsens as the number of LEDs per chip increases.

[0018] A solution known from the prior art is banking. Banking means that the LEDs are connected to the same output of an LED driver. Common bank numbers are 8, 12, 16, 32, ... The number of current LED drivers is reduced, but a higher current is required at the output of each LED driver to achieve the same brightness level averaged over the duration of the frame. An example of banking (the number of banks is 4) is shown in Figure 2Given above.

[0019] Another disadvantage is that this requires multiplexing of the time for which the LEDs are driven. Without the grouping in the above example, there are 2 n clock pulses within one frame. Now if there are 16 groups, then 2 n x16 clock pulses are required. This means that the clock frequency for driving the LEDs will increase by a factor of 16.

[0020] Given that the maximum clock rate of the driver is + / -30 MHz, then 2 n x16 clock pulses within one frame are no longer achievable. Taking the example of 16 groups, 60 Hz x 2 16 x16 groups = 63 MHz.

[0021] The designer of the LED driver for the display panel thus faces the following dilemma:

[0022] - Limiting the number of groups alleviates the problem of the brightness range that can be covered using a single current amplitude for the current pulses. However, this increases the number of LED drivers per panel and thus increases the cost of the panel (assuming this is possible). Limiting the number of bits to be displayed significantly alleviates the bandwidth problem. For example, going from 16 bits per pixel per color to 14 bits reduces the required bandwidth by a factor of 4. However, details are lost at low brightness levels and the color gamut may be affected depending on the behavior of the LEDs of different colors.

[0023] There is a need for improvement in the prior art. Summary of the Invention

[0025] According to one aspect of the present invention, there is provided a method of using image data to drive an active matrix display having pixels including a driving transistor and a light-emitting device, the image data representing an image to be displayed in successive frames, the method comprising:

[0026] - Converting each frame represented by an n-bit digital image into an (m + 1)-bit digital image by truncating (n - m) least significant bits or most significant bits for each pixel, where m < n, and encoding an indication in the (m + 1) bits,

[0027] - Providing a first drive current to each pixel of the (m + 1)-bit digital image if (n - m) least significant bits are truncated, and providing a second drive current to each pixel of the (m + 1)-bit digital image if (n - m) most significant bits are truncated, wherein the ratio of the first drive current to the second drive current is such that it results in a brightness ratio of 2^(n - m), and wherein the indication provides information regarding the drive current to be used.

[0028] The first and second drive currents that produce a brightness ratio of 2^(nm) can be determined by optical measurement of the output of the active matrix display.

[0029] The relationship between brightness and drive current may be at least partially non-linear.

[0030] If a single bit is used for the (m+1) bit, only two currents are encoded by this bit: 0 or 1. Based on the Barten curve / PQ curve, a test pattern needs to be displayed. Preferably, several pixels are used to determine the maximum and minimum currents required to provide a brightness ratio of 2^(nm). For example, when displaying a pattern, a tiled display or a tiled display can be used to generate the output.

[0031] Thus, optical measurements may be made using a set of pixels displaying a test pattern, such as according to Barten.

[0032] An advantage of the present invention is that the timing constraints on the hardware are reduced by a factor of 2^(nm). Instead of n bits, the data sent to the driver is (m+1) bits, including an indicator of whether high or low current is to be used. This reduces the bandwidth of the LED or OLED driver. Furthermore, when the number of bits is reduced to m, detail is no longer lost at low brightness levels, and the color gamut is unaffected. Furthermore, bandwidth is significantly improved.

[0033] In an embodiment of the present invention, the conversion may be performed using an optoelectronic transfer function such as a gamma curve or a PQ curve, or a lookup table derived from the optoelectronic transfer function.

[0034] In a preferred embodiment of the invention, the transition from said first current to said second current is performed such that it results in a brightness increase below a minimum detectable contrast evaluated according to a model of the human visual system.

[0035] Thus, because lower current is used for the lower brightness range, it behaves as if the display is an n-bit display (although implemented as a -bit display), and for the higher range, with higher current, the same behavior as an n-bit display is provided because the current change is performed in the area where no difference can be observed. The problem of not being able to be observed can be determined by following a model of the human visual system.

[0036] Preferably, the model of the human visual system is based on at least one of the following: just noticeable difference, Barten model, Weber's law, De Vries-Rose square root law, PQ curve.

[0037] Advantageously, the Barten model and the PQ curve have proven to be very efficient for this type of display and have also been implemented as a standard in this technical field.

[0038] In a preferred embodiment of the present invention, the pixels of the active matrix display are LED or OLED pixels.

[0039] In a preferred embodiment of the present invention, each frame can be divided into sub - frames.

[0040] Advantageously, one sub - frame of the frame is driven with a first drive current and another sub - frame of the frame is driven with a second drive current.

[0041] According to another aspect of the present invention, there is provided a digital drive circuit for driving an active matrix display having pixels, each of the pixels may include a drive transistor, a light - emitting device, and a digital - to - analog converter, and the digital drive circuit further includes:

[0042] means for converting a sequence of frames represented by the n - bit digital image into an (m + 1) - bit digital image by truncating (n - m) least - significant bits or most - significant bits for each pixel, where m < n, and means for encoding an indication of which current is to be used in the (m + 1) bits,

[0043] where the digital - to - analog converter is configured to provide a first drive current to each pixel of the (m + 1) - bit digital image if (n - m) least - significant bits are truncated, and to provide a second drive current to each pixel of the (m + 1) - bit digital image if (n - m) most - significant bits are truncated, where the ratio of the first drive current to the second drive current is such that it results in a brightness ratio of 2^(n - m), and where the indication provides information related to the drive current to be used.

[0044] The first and second drive currents that result in a brightness ratio of 2^(n - m) can be determined in advance (e.g., during factory calibration) by optical measurement of the output of the active matrix display.

[0045] If a single indicator bit is used for the (m + 1) bits, only two currents are encoded by this bit, encoded by 0 or 1. <​​​​

[0048] Preferably, the model of the human visual system is based on at least one of the following: just noticeable difference, Barten model, Weber's law, De Vries-Rose square root law, PQ curve.

[0049] In a preferred embodiment of the invention, the displays of the active matrix display are LED or OLED pixels.

[0050] Preferably, each frame can be divided into sub-frames.

[0051] Advantageously, one subframe of a frame is driven with a first drive current and another subframe of said frame is driven using a second drive current.

[0052] According to yet another aspect of the present invention, there is provided an LED or OLED display comprising the above-mentioned digital driving circuit.

[0053] In any embodiment of the present invention, a low-pass filter can be used to smooth the current output. However, remember that brightness is the same as luminance, but is perceived by the human eye, which acts as a low-pass filter that averages the sequence of light pulses. So, for PWM drive pulses, the brightness or average brightness during one PWM cycle can be filtered by the eye's LPF (low-pass filter). Therefore, generally speaking, there is no need to provide a low-pass filter with the display.

[0054] The technical effects and advantages according to the embodiments of the invention relating to displays correspond mutatis mutandis to those of the corresponding embodiments of the method according to the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] These and other technical aspects and advantages of embodiments of the present invention will now be described in more detail with reference to the accompanying drawings, in which:

[0056] Figure 1 A schematic diagram of an active matrix is ​​illustrated.

[0057] Figure 2 An example of grouping is explained.

[0058] Figure 3 The luminance of a sinusoidal test pattern is shown as a function of position (typically a row or column of pixels or sub-pixels) in a given direction across the display surface.

[0059] Figure 4 The brightness of a square wave test pattern is shown as a function of position (usually a row or column of pixels or sub-pixels) in a given direction across the display surface.

[0060] Figure 5Shown is a test pattern displayed on a display surface where the brightness varies with a square wave superimposed to the average brightness.

[0061] Figure 6 Shows the principle schematic diagram of the passive matrix.

[0062] Figure 7 Shown is a 12-bit Rec1886 gamma curve with peak luminances of 100, 1000, and 10,000 cd / m². 2 .

[0063] Figure 8 An example of a rectangular pulse wave is shown.

[0064] Figure 9 A pulse over a period T is shown, which is divided into four partial pulses SP1 , SP2 , SP3 and SP4 distributed over one period.

[0065] Figure 10A Shown with a duty cycle of T cl / T pulse width modulation signal.

[0066] Figure 10B A pulse P is shown divided into four sub-pulses, each sub-pulse occurring in one of the following intervals: in which the period T is divided as in Figure 10A The above explains the division of land.

[0067] Figure 11A shows the pulse amplitude I max .

[0068] Figure 11B shows the pulse amplitude I min .

[0069] Figure 12 The steps taken to truncate BD1 (eg, 16) bits according to an embodiment of the present invention are shown.

[0070] Figure 13 The truncation of the four most significant bits and the addition of the CPA bits according to an embodiment of the present invention are shown.

[0071] Figure 14 The truncation of the four least significant bits and the addition of the CPA bits according to an embodiment of the present invention are shown.

[0072] Figure 15 Barten curves are shown for ramp and square wave patterns, and illustrate how ΔL / L varies as a function of L when using a linear LED or OLED driver limited to 12 and 16 bit depths. For the 12-bit bit depth case, two curves are given: the first curve has a low current, and the second curve has a high current, according to an embodiment of the present invention.

[0073] Figure 16a A transition from 12 bits with low current to 12 bits with high current is shown according to an embodiment of the present invention.

[0074] Figure 16b Curves ΔL / L as a function of L (luminance) when using linear LED or OLED drivers with 12 and 16 bit depths according to an embodiment of the present invention are shown, as well as Barten curves for ramp and square wave patterns.

[0075] Figure 17A A current source according to an embodiment of the present invention is shown.

[0076] Figure 17B Another current source according to an embodiment of the present invention is shown.

[0077] Figure 18A It is explained that the sum of all time intervals is equal to the PWM period T used in the embodiment of the present invention.

[0078] Figure 18B FIG. 1 is a schematic diagram showing twelve current sources 1 for forming a 12-bit current DAC according to an embodiment of the present invention.

[0079] Figure 19A The current amplitude is explained as I during the time interval TLSB corresponding to LSB B0. max , and the current amplitude is I during the time interval 4*TLSB corresponding to the third bit B2 min .

[0080] Figure 19B Similar to Figure 18B , wherein according to an embodiment of the present invention, the input signal or control bit can be set high or low independently for each current source.

[0081] Figure 20 A circuit is shown that enables a less abrupt transition from low current mode to high current mode according to an embodiment of the present invention.

[0082] Figure 21 A PWM signal is shown in which the amplitude of the current pulse has a first value during a portion of the duty cycle and a second value during another portion of the duty cycle according to an embodiment of the present invention.

[0083] Figure 22 PWM signals with different pulse durations and different amplitudes according to an embodiment of the present invention are shown.

[0084] Figure 23 An example of a modified transition from a first pulse current amplitude Imin to a second pulse current amplitude is shown.

[0085] Figure 24 Shown is the non-linear function relationship between the relative brightness of two different LEDs (y-axis) and the forward current in the LED (x-axis).

[0086] Figure 25 ΔL / L as a function of L, and PQ curves such as a Barten curve for a ramp wave and a Barten curve for a square wave are shown.

[0087] Figure 26 and 27 Shown in I Tr and I Max is the curve ΔL / L obtained when the duty cycle is increased and the amplitude of the current pulse changes from I Tr Increase to I Max Things that happen when.

[0088] Figure 28 Shown Figure 25 details.

[0089] Definitions and acronyms.

[0090] Active Matrix. Active matrix is ​​a type of addressing scheme used in flat panel displays. In this method of switching individual elements (pixels), each pixel is attached to a transistor and capacitor that actively maintains the state of the pixel while other pixels are being addressed. A schematic diagram of the active matrix is ​​shown in Figure 1 Given above.

[0091] Grouping. Grouping means that multiple LEDs or OLEDs are connected to the same output of the LED or OLED driver. This has the advantage that the number of LED or OLED drivers can be significantly reduced. Common grouping numbers are 4, 8, 12, 16, and 32. Examples of grouping are shown in Figure 2 . In this example, the number of groups is 4. The output 30 of the LED or OLED driver 20 is connected to 4 LEDs or OLEDs instead of a single one. Each of the four LEDs or OLEDs (21, 22, 23, 24) is connected to a power supply 29 through a switch (25, 26, 27, 28). At any given time, only one of the switches is closed.

[0092] Barten (curve).

[0093] The Barten model is valid for foveal vision under photopic conditions. This has been used to cover the range from 0.0001 to 1000 cd / m 2 The "Barten curve" shows the convolution of contrast sensitivity with mean luminance.

[0094] The Barten curve is the threshold for visible brightness steps as a function of global brightness. All values ​​below this curve are not visible to the human eye.

[0095] Global brightness is defined based on a given test pattern displayed on the display under consideration, in particular, an LED or OLED display.

[0096] For example, the test pattern can be a sinusoidal distribution in one direction of the display (such as Figure 3 explained above). Figure 3 The brightness is shown as a function of position (usually a row or column of pixels or sub-pixels) in a given direction across the display surface.

[0097] The luminance (axis 31) varies like a sine wave (or more precisely, like a constant + a sine wave: L = L0 + A*sin(αx), where L0 is the constant or baseline luminance, x is the position in a given direction across the display surface, and A and α are constants), and is at a maximum value 33 (L Max ) and a minimum of 34(L Min ) varies between.

[0098] The contrast sensitivity function (CSF) is the inverse of the minimum modulation that can be seen by an observer of the display. In this example, the modulation is given by m=(L Max –L Min ) / (L Max +L Min ) is given.

[0099] The CSF is the inverse of the minimum visible modulation:

[0100] CSF=1 / m min

[0101] Instead of sine wave modulation, a square wave can be used. Figure 4 Shows the brightness across the test pattern (such as Figure 5 How the test pattern on (a) changes along direction D. Figure 5 A test pattern 50 is shown displayed on a display surface (either a complete display or a slice of a tiled display). Along direction "D", the brightness varies as if a square wave were superimposed on the average brightness, L(x) = L O + Square wave (x), where x is the position along direction "D". The square wave has a period P (51) covering at least two pixels but may contain more than 2 pixels.

[0102] The test pattern may also be a ramp wave with a square wave or a sine wave superimposed thereon.

[0103] This Barten pattern shows a mathematical representation of the human eye in terms of minimum detectable contrast steps as a function of brightness.

[0104]

[0105] in:

[0106]

[0107]

[0108]

[0109]

[0110] The following parameters may be used for the examples shown in this specification:

[0111]

[0112] m t is the minimum modulation threshold:

[0113]

[0114] and

[0115] ΔL=L max -L min

[0116]

[0117] in:

[0118]

[0119] According to the Barten curve, these are the minimum detectable contrast steps of the sinusoidal pattern.

[0120] For square or ramp waves:

[0121]

[0122]

[0123] These curves will be used as a benchmark of how well the system is performing. The square wave pattern will provide the lowest value followed by the sine pattern and ramp wave.

[0124] BD or bit depth. The number of bits used to encode, for example, the brightness, grayscale, color, etc. of a pixel.

[0125] Contrast, contrast, contrast sensitivity.

[0126] If the brightness difference is large, objects can generally be better distinguished from each other. Relative brightness differences are more important than absolute differences. This relative difference can be expressed as the ratio between two brightness values ​​and is called contrast. This relative difference can also be expressed as the difference between two brightness values ​​divided by their sum, which is called contrast. The reciprocal of the minimum contrast required for detection (by the human eye) is called contrast sensitivity.

[0127] Different test patterns can be used to evaluate contrast, contrast, and contrast sensitivity.

[0128] One of the test patterns is a sinusoidal brightness pattern. In this case, contrast is defined by the amplitude of the sinusoidal variation divided by the average brightness. This quantity is called the modulation depth or modulation. The minimum modulation required for detection of this pattern is called the modulation threshold.

[0129] When contrast sensitivity is measured using non-sinusoidal luminance variations, contrast is determined by dividing the difference between the maximum and minimum luminances by their sum. This is called Michelson contrast.

[0130] Digital-to-Analog Converter (DAC). A device used to convert binary or digital codes into analog signals. DAC devices convert abstract finite-precision numbers (usually fixed-point binary numbers) into definite physical variables such as voltage, pressure, brightness, or luminance.

[0131] PWM as a digital-to-analog converter can be found in the third edition of Paul Horowitz's "The Art of Electronics." With a PWM DAC, the driven system typically acts as a low-pass filter that time-averages the modulated signal. In the example Horowitz gives, a PWM signal can drive a light-emitting diode. Specifically, he explains that a system driven by a PWM signal responds slowly (i.e., it acts as a low-pass filter). Therefore, a PWM DAC can be a time-averaging DAC. The human eye acts as a low-pass filter, so the device according to the present invention does not require such a filter.

[0132] For DACs (and specifically PWM DACs), see:

[0133] (a) “Digital-to-analog converter using pulse width modulation” see US4590457A

[0134] (b) “Pulse Width Modulation Digital to Analog Converter”, see US 6191722

[0135] The skilled artisan will also recognize that PWM units are known from DC motor control. The energy transferred is related to the average current; for example, the current never remains constant during a PWM cycle. Embodiments of the present invention can utilize conventional DACs that are typically sampled and held, i.e., the output current remains constant until a new binary code is input to the DAC and a conversion is requested. However, embodiments of the present invention can utilize pulsed or variable drive currents.

[0136] Duty cycle. The term duty cycle describes the ratio of 'on' time to a regular interval or 'period' of time; a low duty cycle corresponds to low power because the power is off for most of the time. Duty cycle is expressed as a percentage, with 100% being fully on.

[0137] EOTF Electro-Optical Transfer Function. It describes how digital codewords are converted into displayed brightness. Examples of EOTF are the Gamma curve based on CRT physics and the PQ curve introduced by Dolby for HDR (0 to 10,000 cd / m²). 2 Ideally, the EOTF should be defined based on a human visual response model.

[0138] The Gamma function has the following expression:

[0139] Y Gamma =L*V γ

[0140] 0≤V≤1

[0141] L is the maximum luminance of the display, and V is the normalized digital input value.

[0142] The PQ curve has the following expression:

[0143]

[0144] The following parameters are selected for the examples shown in the embodiments of the present invention:

[0145]

[0146] Standard PQ curve from 0cd / m 2 to 10,000 cd / m 2 If the maximum brightness limit on the display is 2500cd / m 2 , there are two options:

[0147] 1) Follows the standard PQ curve and clips at maximum brightness

[0148] 2) Between 0 and 2500 cd / m 2 Recalculate the standard PQ curve between

[0149] Option 1 has the advantage that when comparing two displays with the same input and PQ curve, they will show exactly the same content. However, above 2500 cd / m 2 All values ​​of will have the same lightness value.

[0150] Option 2 has the advantage that the full digital input range is used. However, the input-to-output relationship is lost, and on non-ideal displays, more low-brightness values ​​are discarded.

[0151] Flicker. Flicker is a visible dimming or reduction in brightness between two consecutive frames or more generally two consecutive cycles (eg, two consecutive cycles of a PWM signal).

[0152] Frame. A frame is one picture in a series of pictures, such as in a movie-length video. A frame can also mean a complete image used for display (on a display or a slice of a tiled display). In some contexts, a frame can also mean the time interval during which a frame is displayed.

[0153] JND. Just Noticeable Difference. In the branch of experimental psychology focused on sensation, perception, and sensation, a just noticeable difference, or JND, is the amount by which something must change for the difference to be noticeable, detectable, at least half the time (the absolute threshold). This threshold is also known as the difference threshold, differential threshold, or just noticeable difference. In LED or OLED displays, the smallest brightness step that will be perceived by the human visual system is of particular interest.

[0154] Weber's law is sometimes used alone or in combination with other laws to define JND. Weber's law defines that there is a constant relationship between JND and reference brightness.

[0155] Weber's law or fraction can be expressed as

[0156]

[0157] where K w is called the Weber fraction, and where ΔI is the intensity delta (or just noticeable increment of intensity) and I is the initial intensity. In other words, the Weber fraction provides a relationship between the initial intensity (or reference brightness) and the just noticeable increment of intensity (or JND). According to this law, just noticeable contrast (i.e., the inverse of contrast sensitivity) is constant regardless of brightness. It is believed that the ratio is between 1 / 50 and 1 / 100. However, it increases below and above a certain brightness.

[0158] According to Weber's law, it is possible to determine each JND for each brightness.

[0159] Weber's law may not be valid over the entire brightness range. For very low and very high intensities, the Weber fraction is too strict. Therefore, for low brightness, the De Vries-Rose square root law can be used:

[0160]

[0161] The origin of this is photon noise, that is, the number of photons in a light pulse is not completely fixed at a specific brightness, but varies "a little" (spreads) around a mean value according to a Poisson distribution. This means that the brightness difference or increment I must be large enough to be distinguished from fluctuations in background brightness. The Weber fraction is the most stringent over the entire range, so that if the Weber fraction is met, the De Vries-Rose square root law will also automatically be met. Therefore, Weber's law is the appropriate law to use when defining JND, but better solutions may exist by using other laws / rules.

[0162] LED display.

[0163] The following patents from the same applicant provide definitions of LED displays and related terms. They are hereby incorporated by reference for the definitions of these terms.

[0164] US7972032B2 "LED assembly",

[0165] US7176861B2 Pixel structure with optimized sub-pixel size for emissive displays,

[0166] US7450085 Intelligent light emitting module and method of operating such intelligent light emitting module,

[0167] US7071894 is a method and apparatus for displaying an image on a display device.

[0168] References to LED displays include OLED displays.

[0169] LSB. Least Significant Bit.

[0170] Luminance (L). The intensity of luminous intensity per unit area projected in a given direction. The SI unit is the candela per square meter, sometimes also called nits. In the literature, luminance and brightness are often used interchangeably, even though they are not the same thing. Whenever the term "brightness" is used herein, the inventors mean "brightness."

[0171] Of particular importance in designing LED or OLED displays is the functional relationship between brightness and the forward current through the LEDs or OLEDs used in the LED or OLED display. Figure 24The figure shows how the relative brightness varies as a function of the forward current in LEDs commercialized by companies such as Nichia. For example, the nonlinearity of a blue LED may be stronger than that of a red or white LED. However, in any display, there is a high probability that at least some OLEDs or LEDs will have relative brightness that varies nonlinearly as a function of the forward current. In such cases, the brightness or luminance must be determined for the forward current value, for example by optically measuring it using a pattern displayed on the display to determine this nonlinear relationship. Once these measurements have been made, a current can be selected that results in a factor of 2 increase in brightness that would not normally be achieved by a factor of 2 in the forward current.

[0172] MSB. Most Significant Bit.

[0173] Passive matrix. Passive matrix addressing is the addressing scheme used in early LCDs. This means that only m+n control signals are needed to address an m×n display. A pixel in a passive matrix must maintain its state without active drive circuitry until it can be refreshed again. Figure 6 Shows the principle schematic diagram of the passive matrix.

[0174] This figure is shown in US7262753 patent from the same applicant of the present application. Figure 2 The description of the drawings is hereby incorporated herein by reference.

[0175] PQ curve or perceptual quantization curve. Figure 7 .

[0176] By Dolby TM The introduced standard has been published by SMPTE as SMPTE ST 2084. It defines transfer functions that allow the display of high dynamic range (HDR) video at brightness levels up to 10,000 cd / m2 and can be used with the Rec.2020 color space. PQ is a nonlinear electro-optical transfer function (EOTF).

[0177] Figure 7As shown in the following publication, from the SMPTE conference presentation "Perceptual Signal Coding for More Efficient Usage of Bit Codes", it is further described: "The ITU-R Rec. BT.1886 EOTF for television, often called "gamma coding", is often said to be perceptually linear. A recent ITU report related to ultra-high-definition television (UHDTV) (Report ITU-R BT.2246) uses a scaled Barten contrast sensitivity function (called "Barten (Ramp)") together with Schreiber's replacement threshold function to illustrate how the ITU-R Rec. BT.1886 EOTF for HDTV behaves similarly to human perception and is close to or below the visual detection threshold for 10 and 12-bit implementations. While this is roughly the case with a gamma curve of 100 cd / m2 (or 100 nits), as in [1]. Figure 1 As shown in , the 12-bit gamma curve rises quickly above both the Barten and Schreiber thresholds when using higher peak luma levels, suggesting that visible quantization artifacts, especially at the dark end of the luma range, may become apparent. While system accuracy can be increased by using higher bit depths, conventional infrastructure will struggle to go beyond 12 bits. In fact, most live production and broadcast environments still operate at 10 bits, so a system that can provide improved performance at these common bit depths would be ideal.

[0178] PWM (Pulse Width Modulation). Pulse Width Modulation uses a rectangular pulse wave whose pulse width is modulated, causing the average value of the waveform to vary. Figure 8 An example of such a rectangular pulse wave is shown.

[0179] A square wave has a period T, a lower limit I0 (usually 0 in our case), an upper limit I1, and a duty cycle D. The duration of one pulse P (the time the signal is at its upper limit) is D / 100*T (in the case where D is expressed in %). For example, if D=50%, the duration of the pulse is 1 / 2T.

[0180] In some cases, the shape of the pulse P is as shown in Figure 9 If the period T is "very long" or of the same order as the time constant of an important physical process, it may be advantageous to "split" the pulse into several sub-pulses spread over one period of the wave. Figure 9 In FIG. 1 , one period T has been divided into four sub-pulses SP1 , SP2 , SP3 and SP4 distributed across one period.

[0181] Depending on the application, it may be desirable to divide a cycle into more than 4 intervals.

[0182] In digital systems, the duration of a pulse is the clock period T cl For a given T and T cl The minimum achievable duty cycle is therefore T cl / T.

[0183] If the duty cycle is set to its minimum value T cl / T, the pulse width modulation signal will be Figure 10A If the duty cycle is further increased, T cl / T, the pulse P can be split into two or more sub-pulses, each sub-pulse occupies one of the intervals into which the period T is divided, such as Figure 10B Explained above.

[0184] As the duty cycle is further increased, each of the intervals is filled so that the total duration of the sub-pulses is equal to D*T.

[0185] When I0=0, the average current circulating in the diode yes:

[0186] =I1*D / 100 (where D is expressed in %) or

[0187] =I1*D (where D is expressed as a fraction of T, as a real number in the interval [0,1]). In an LED or OLED display, frames are displayed at a frequency of, for example, 60 Hz (corresponding to T = 1 / 60 s). When the LED or OLED is driven with a PWM signal, splitting the pulse into sub-pulses can reduce visible flicker (considering that anything below 100 Hz can be seen by the human eye). Splitting the pulse into sub-pulses can be seen as increasing the frequency by a factor of N, where N is the number of intervals into which the period is divided.

[0188] Even in these cases, the waveform of the current may not strictly be that of a PWM signal, but reference will be made to PWM when discussing LED or OLED current drive schemes. DETAILED DESCRIPTION

[0189] The present invention will be described with respect to specific embodiments and with reference to specific drawings, but the invention is not limited thereto and is limited only by the claims. The drawings described are merely illustrative and non-restrictive. In the drawings, for illustrative purposes, the sizes of some of the elements may be exaggerated and not drawn to scale. When the term "comprising" is used in this specification and claims, the term does not exclude other elements or steps. In addition, the terms first, second, third, etc. in the specification and claims are used to distinguish between similar elements and are not necessarily used to describe a sequence or chronological order. It should be understood that the terms used in this manner are interchangeable where appropriate, and the embodiments of the invention described herein can be operated in a different order than described or illustrated herein.

[0190] In the present invention, input image data is used to drive an active matrix display having pixels including drive transistors and light emitting devices, the input image data representing an image to be displayed in consecutive frames. This means that there is (2 n ) 3 It also means that the maximum display brightness is 2 n The brightness is divided into steps and this is therefore the minimum possible brightness that can be displayed using a linear electro-optical transfer function.

[0191] These n bits are given to a LUT in which the EOTF can be implemented. The output of this LUT is also limited in the number of bits (m). This limit is set by the display's hardware.

[0192] As discussed earlier, the duration of the shortest sub-pulse width is T cl This will determine the bit depth or number of brightness levels of the LEDs of the LED display.

[0193] For example, once the duration T of a frame is known, the maximum number of different values ​​of the duty cycle D is: T / T cl If T = 1 / 60s and T cl is 254ns, then the maximum number of brightness levels N is 65616, which is slightly larger than 2 16 In other words, this particular choice of T and Tcl is consistent with a 16-bit LED or OLED driver.

[0194] The lowest non-zero brightness level will be used as in Figure 10A This is achieved by the PWM signal represented in .

[0195] If the duration of the shortest sub-pulse is higher than T cl If the lower boundary of the current pulse amplitude is less than the maximum number of bits (254ns), then the use of a specific bit depth BD1 (such as 16 bits) will not be possible, and the number of brightness levels achievable using a constant value of the current pulse amplitude will be less than the maximum number of bits 2 BD1 -1, such as 2 16 -1.

[0196] If the number of bits is limited to a lower level BD2, such as 12 bits, then T c Will be limited, such as T cl ≈4μs (or less), where T=1 / 60s.

[0197] In order to be able to cl The same brightness as 254ns is achieved in ≈4μs, and the maximum amplitude of the current pulse I Max Must be lowered.

[0198] Thus, by reducing the current through the OLED or LED so that the brightness of the OLED or LED is 16 times lower, the brightness of a 16-bit display will be achieved. In other words, 4 extra bits are saved by dividing the current by a factor F, which is not necessarily equal to 16 (because of the nonlinear relationship between the forward current in the LED or OLED and the resulting brightness). Due to the nonlinearity of the current through the LED or OLED and the brightness of the LED or OLED, the ratio between I_min and I_max is not equal to 16. In order to obtain a ratio of L_min / L_max=1 / 16, one option is to measure the optical output of the LED or OLED display to determine which currents I_min and I_max provide this brightness ratio. The measurement can be made using a pattern displayed on a set of pixels, on a piece of the tiled display, or on the entire tiled display. The pattern can be a Barten pattern.

[0199] More generally, according to the present invention, if the input image data is encoded using BD1 bits, then each input image is converted into a (BD2 + 1)-bit digital image by truncating (BD1 - BD2) least significant bits or most significant bits of the BD1-bit digital image for each pixel, where BD2 < BD1, and the indication is encoded in (BD2 + 1) bits.

[0200] If (BD1 - BD2) least significant bits are truncated, a first drive current (Imax) is provided to each pixel of the (BD2 + 1)-bit digital image, and if (BD1 - BD2) most significant bits are truncated, a second drive current (I Min ) is provided to each pixel of the (BD2 + 1)-bit digital image, where the ratio of the first drive current (I Max ) to the second drive current (I Min ) is such that it creates a brightness ratio of 2^(BD1 - BD2), and where the indication provides information related to the drive current to be used.

[0201] On the other hand, for the new value I Min of the current pulse amplitude, the maximum brightness will be 16 times lower than in the case of the initial value I Max .

[0202] Throughout this specification, it is assumed that the ratio of I Min to I Max is equal to the ratio of the brightness of the LEDs or OLEDs. However, in practice, this is not the case. In this example, this brightness ratio of low current and high current is 1 / 16. Due to the efficiency (brightness <-> current) of the LEDs or OLEDs, the current ratio (I Max <-> I Min ) does not have a ratio of 1 / 16. In the example shown, a ratio of approximately 1 / 10 is used.

[0203] In a first aspect of the present invention, the required brightness range is achieved with fewer bits (e.g., BD2 instead of BD1, such as 12 bits instead of 16 bits) by using a first current value I max as the maximum amplitude of the current pulse when the LEDs or OLEDs must reach a higher brightness range and a second, lower current value I min as the maximum amplitude of the current pulse when the LED or OLED pixels are used in a lower brightness range.

[0204] Instead of having to send BD1 (such as 16) information bits related to the average current flowing through the LEDs or OLEDs, BD2 (such as 12) bits can be used to encode the average amplitude of the current flowing through the LEDs or OLEDs (by means of, for example, the duty cycle), and 1 bit for encoding I max or I min Which of BD2 (such as 12) bits must be used as the amplitude of the current pulse. When the LED or OLED is driven with a pulse width modulated current, BD2 (such as 12) bits can be used, for example, to encode the duty cycle.

[0205] For example, if BD2 plus 1 bit (such as bit 13) is a logic 1, the current pulse will have an amplitude of I max , as in Figure 11A Explained above. Figure 11A In the example, a 25% duty cycle corresponds to the 12-bit sequence MSB=b11=0, b10=1, b9=0, b8=0, b7=0, b6=0, b5=0, b4=0, b3=0, b2=0, b1=0, b0=0=LSB (where b11 is the MSB and b0 is the LSB).

[0206] If the 13th bit is a logic 0, the current pulse will have an amplitude of I min , as in Figure 11B Explained above. Figure 11B In the example, a 25% duty cycle corresponds to the 12-bit sequence b11=1, b10=0, b9=0, b8=0, b7=0, b6=0, b5=0, b4=0, b3=0, b2=0, b1=0, b0=0 (where b11 is the MSB and b0 is the LSB).

[0207] In embodiments of the present invention, data sent to LED or OLED displays according to these embodiments of the present invention can still be encoded and stored with BD1 (ie 16) bits (per color), but is truncated before being sent to the LED or OLED current drivers.

[0208] Figure 12 The steps taken to truncate BD1 by (eg, 16) bits according to an embodiment of the present invention are shown.

[0209] In an embodiment of the present invention, the values ​​of the first S most significant bits (such as the four most significant bits) and / or the L least significant bits (such as the four least significant bits) may indicate whether the corresponding LED or OLED has been driven to achieve a brightness in the highest or lowest brightness range.

[0210] Comparators may be provided to compare the first S or four most significant bits ( b15 , b14 , b13 and b12 ) of the (such as 16) bits of BD1 .

[0211] If the S or four bits are zero, they are truncated. This leaves a BD 2-bit word, such as the 12-bit word [b11-b0]. Current pulse amplitude bit b cpa (BD2 plus 1 bit or the "13th" bit) is set to zero, indicating that the amplitude of the current pulse is I Min , which corresponds to the lowest brightness range.

[0212] Figure 13 The original BD2 (eg, 16) bit word and the result truncated to BD2 (eg, 12) bits are shown, as well as the state of the current pulse amplitude bits when the first S (eg, 4) most significant bits b15, b14, b13, and b12 are zero.

[0213] If any of the first S (such as four) most significant bits are different from zero, indicating that the LED or OLED is to be driven in a higher brightness range, then the L (such as 4) least significant bits (b3, b2, b1, and b0) of the 16-bit word will be truncated. This leaves BD2 (such as a 12-bit word [b15-b4]. Current pulse amplitude bits b cpa (BD2 plus 1 bit, or the "13th" bit) is set to 1, indicating that the amplitude of the current pulse is I Max , which corresponds to the highest brightness range.

[0214] Figure 14 The original BD1 (e.g. 16) bit word and the result of truncating it to BD2 (e.g. 12) bits are shown, as well as the state of the current pulse amplitude bits when at least one of the first S (e.g. 4) most significant bits b15, b14, b13 and b12 is 1 (this is indicated by Figure 12 The state Y on the top is represented).

[0215] In the attached Figure 13 and 14 In the example above, state X means either 1 or 0 and indicates that truncation is done independently of the value of those bits in the X state.

[0216] Switching between low current and high current can be achieved as follows:

[0217] - Between 0-4095, the pixel is driven with low current (low light (LL) in the table below),

[0218] - Between 4096-65535, pixels are driven with high current in 12-bit steps (High Light (HL) in the table below).

[0219] So, if a 16-bit value image is received and bits 15-12 are zero, the lower 12 bits are selected to be displayed with a low current. If bits 15-12 are not zero, the first 12 bits are selected to be displayed with a high current:

[0220] Table 1:

[0221]

[0222] In the high current implementation, a brightness step of 16 is taken (due to the higher current), as in Figure 16a Explained above.

[0223] It would be preferable if the step change from lower current to higher current was not visible to the human eye. For example, a Barten curve could be used to indicate whether this transition is visible. As described above, the Barten curve shows the threshold for visible brightness steps as a function of global brightness. All values ​​below this curve are not visible to the human eye.

[0224] The Barten curves (for square wave patterns and for ramp wave patterns) are Figure 16b . It also shows the PQ curve (EOTF) for a 16-bit display and a 12-bit display. Increasing the number of output bits (from BD2 to BD1, or from 12 to 16) causes the curve to shift towards a lower minimum detectable contrast when the number of output bits increases. This behavior is expected because for BD1 (such as 16)-bit encoding, the maximum luminance of the display is divided by a power of 2 more than for BD2 (such as 12)-bit encoding, and thus causes a minimum luminance step for BD1 (such as 16)-bit encoding, which is less visible than for BD2 (such as 12)-bit encoding.

[0225] Different ways of performing truncation can be implemented, and each is included as an embodiment of the present invention.

[0226] From I min Switch to I max This may be accompanied by visual artifacts. However, truncation is preferably performed so that the transition from the lowest luminance range, determined by Imin, to the highest luminance range, determined by Imax, is as smooth as possible with respect to human visual perception. The human visual perception model can be represented by a Barten slope, or, as described above, can be defined by a gamma curve, or even more preferably, by a PQ curve.

[0227] It appears that the brightness of the transition from one reference current amplitude to another will have a decisive influence on whether the artifact is too large (ie, too noticeable to the human eye to be acceptable). min Increase to I max (corresponding to the brightness L min and L max ), one LSB corresponds to N (where N = L max / L min , or more generally, a 2^(BD1-BD2)) times higher brightness increase.

[0228] The current circulating through the LED or OLED (averaged over period T) is =D / 100*I pulse , where I pulse is the amplitude of the current pulse (i.e., I min or I max ).

[0229] If the duty cycle is increased by one LSB, the change in the current circulating through the LED or OLED averaged over a period T is Δ =1 / M*I pulse , where M is the number of duty cycle steps ΔD that can be encoded using the chosen number of bits. When using 12 bits to encode the duty cycle, Δ =1 / 4095*I pulse .

[0230] Assume that LED or OLED uses I pulse =I min to drive, and then the duty cycle D increases from 0 to 100% in steps of one LSB. The current will step Δ =1 / 4095*I min Increase until the duty cycle reaches 100% and =I min After that, to further increase the brightness, the amplitude of the current pulse must be increased to I max .

[0231] In the case of N=16, and still using BD2 (e.g., 12) bits to encode the duty cycle D, the duty cycle required to have the same brightness (averaged over period T) in the LED or OLED would be encoded as follows: MSB=b11=0, b10=0, b9=0, b8=0, b7=1, b6=1, b5=1, b4=1, b3=1, b2=1, b1=1, b0=1=LSB. Indeed, this encoding represents the number b11*2048+b10*1024+b9*512+b8*256+b7*128+b6*64+b5*32+b4*16+b3*8+b2*4+b1*2+b0*1=255. This encodes a duty cycle D=(255 / 4095)*100%≈1 / 16*100%.

[0232] For BD1=16, the average brightness of the LED or OLED will be:

[0233] <l>=L max *1 / 16

[0234] The amplitude Imax of the current pulse has a corresponding brightness Lmax that is 16 times higher than Lmin using the corresponding current Imin, and equation (2) will be written as follows:

[0235] <l>=L max *1 / 16=16*L min *1 / 16=L min .

[0236] Afterwards, in I pulse =I max In the case of =1 / 4095*I max Increase, which is greater than I pulse =I min The case is 16 times higher stepping.

[0237] Because the brightness step Δ <l>Because the current becomes larger by a factor of N (16 in this example), it can cause visual artifacts. The origin of the visual artifact can be spatial or temporal. For example, the increase in current from one frame to the next in an LED or OLED corresponding to one LSB may be too high to be compatible with standards such as DICOM. The brightness difference between two adjacent LEDs or OLEDs (this brightness difference of the LEDs or OLEDs corresponding to a single LSB) may be too high to be compatible with standards such as DICOM.

[0238] It is therefore advantageous to choose N such that a luminance step L = N / 4095*Lmax will correspond to a just noticeable difference (JND).

[0239] More generally, we will seek to have the same M -1)*L min The corresponding just noticeable step in brightness, where M is the number of bits used to encode the duty cycle (or more generally, 2 M -1 is the number of duty cycle steps (between the minimum value (eg, 0%) and the maximum value (eg, 100%), and hence the number of steps in brightness).

[0240] As mentioned above, there are different types of curves to model or represent contrast sensitivity as a function of brightness. Based on these models, lookup tables can be derived to improve the calculation time, i.e., to reduce it. For example, the Barten model (which takes into account the dependence of contrast sensitivity on screen brightness, field of view, spatial frequency of the image, etc.). The Barten model is widely used and is cited by many electronic imaging studies and standards. Based on this model, the PQ curve has been designed to provide a closer fit to the human visual response curve, as explained in the publication "Perceptual Signal Coding for More Efficient Use of Bit Codes", SMPTE conference presentation, 2012, Scott Miller, Mahdi Newamabadi, Scott Daly.

[0241] Figure 15 Perceptual quantization curves (representing contrast sensitivity as a function of luminance) are shown.

[0242] Curve 150 shows the contrast variation as a function of brightness obtained when a linear current driver with a bit depth of 16 bits is used to drive the LEDs or OLEDs of an LED or OLED display.

[0243] Curve 151a shows the contrast variation as a function of brightness obtained when driving the LEDs or OLEDs of an LED or OLED display using the same linear current driver with a bit depth of 12 bits and the same amplitude of the current pulses used for pulse width modulation.

[0244] Curve 151b shows the comparative variation of the 12-bit bit depth obtained by truncating the 16-bit encoding according to the present invention (ie, at the first amplitude I Min The 4 MSBs are zero when the 12 LSBs are used, and at the second amplitude for the current pulse I Max 12 MSBs are used when at least one of the 4 MSBs of is not zero).

[0245] As long as the first amplitude of the current pulse I Min , the curve 151b follows the curve 150. Thereafter, when the second amplitude I is used for the current pulse Max , the curve ΔL / L stops at the end of 151b and starts to follow the curve 151a.

[0246] Curves 150, 151a and 151b show how the contrast varies as a function of L by ΔL / L, where L is the (average) brightness on the display.

[0247] The current is encoded using, for example, 12 bits. As the bit code increases by 1 LSB, the brightness increases by ΔL and curves 150, 151a, and 151b can be constructed point by point. At low brightness, ΔL / L changes faster than at high brightness.

[0248] and Figure 16b In comparison, the 12-bit (or BD2) HL and LL are also shown in dark grey, superimposed on the 16-bit PQ curve (150) and the second part of the 12-bit PQ curve (151b, from 156 cd / m 2 start). Figure 15 Shown for I cpa =I min (curve 151a) and I cpa =I max (Curve 151b) Evolution of contrast sensitivity using BD1 (e.g. 12) bit coding, where I cpa is the amplitude of the current pulse used for pulse width modulation. In this example, BD 2-bit (such as 12-bit) encoding has been described earlier, ie 12 bits are used to encode the duty cycle of the pulse width modulated current flowing through the LED or OLED.

[0249] Figure 15 Barten curves are also shown (curve 152 corresponds to a square wave pattern, while curve 153 corresponds to a ramp wave pattern).

[0250] So in low light the 16 bit PQ curve is followed perfectly (by curve 151a) however at a certain moment there is a jump to HL drive (at 156 cd / m 2 When the amplitudes of the current pulses (Imin and Imax) are chosen according to the invention, this curve is still below Barten and thus the transition will not be visible (or not noticed).

[0251] The circled area 154 corresponds to the transition region, where the 2 From I min to I max .

[0252] Each current step (corresponding to an LSB) is accompanied by a step in contrast sensitivity (ΔL / L, where ΔL is the increase in brightness corresponding to an increase in current through the LED or OLED, and L is the average current corresponding brightness).

[0253] The brightness step below the Barten curve is not noticeable. Specifically, the brightness change in region 154 (where the amplitude of the current pulse changes from I min Change to I max ) is not visible.

[0254] Even if the brightness step corresponding to one LSB is increased by a factor of N (eg, N=16), if the transition occurs below the Barten curve, the corresponding brightness increase ΔL will not be perceptible.

[0255] exist Figure 15 In, from I Min to I Max The jump is completed at a brightness of +-156cd / m2.

[0256] exist Figure 23 In the example of FIG. 1 , a first current value Imin is used for the current DAC (reference Figure 17A and 18A ), the first 8 bits encoded are converted into a current. Instead of using the last 4 bits for Imin, bits 3, 2, and 1 are set to zero, and bit 0 is implemented using a current Imax corresponding to Lmax=N*Lmin (N=16 in this example, for example). The overall brightness generated by the LED or OLED is the same as if all bits were set to 1.

[0257] Embodiments of a DAC that can achieve a more gradual transition from a first current amplitude to a second current amplitude for a PWM drive system for an LED or OLED display will now be described. Figure 17A An example of a circuit that can implement the present invention is shown. In this or any other embodiment of the present invention, a low-pass filter can be used. However, remember that brightness is the same as luminance, but is perceived by the human eye, which acts as a low-pass filter that averages the sequence of light pulses. So for PWM drive pulses, the brightness or average brightness during one PWM cycle can be filtered by the eye's LPF (low-pass filter).

[0258] A bit sequence encoding the desired current in LED or OLED 180 is fed to signal generator 171. Signal generator 171 generates a first signal 172A that is a function of the bit sequence. First signal 172A determines which of switches 175 or 176 is closed or open. When closed, first switch 175 connects first current source 173 to node 178. When closed, second switch 176 connects second current source 174 to node 178. Signal 172A can, for example, drive first switch 175. Second switch 176 can be driven by the output of a logic inverter 177 that uses signal 172A as an input. Signal generator 171 also generates a second signal 172B that controls a third switch 179. Second signal 172B is used to modulate the current in LED or OLED 180 according to a PWM scheme. When closed, third switch 179 connects LED or OLED 180 to both switches 175 and 176. When the third switch 179 is open, the current in the LED or OLED 180 is zero. When the third switch 179 is closed, the current in the LED or OLED 180 is equal to the current I generated by the first current source 173 when the first switch 175 is closed and the second switch 176 is open. min When the third switch 179 is closed, the current in the LED or OLED 180 is equal to the current I generated by the second current source 174 when the first switch 175 is open and the second switch 176 is closed. max The anode of the LED or OLED 180 is connected to the third switch 179 and the cathode of the LED or OLED 180 is connected to the cathode node 182 at the potential Vk. The signal generator may be implemented, for example, on an FPGA or an ASIC.

[0259] The second signal 172B may be generated in different ways. In one specific example, for example, the LSB of the truncated bit sequence may correspond to the duration T LSB , this T LSB Equal to, for example, the frame rate divided by (2 n -1), where n is the number of bits of the truncated bit sequence. If bit B0 (LSB) of the truncated bit sequence is 1, the signal generator sets the second signal 172B to "high" to generate a signal at a time equal to T LSB The third switch 179 is closed during the time period of 172 and allows current to flow through the LED or OLED 180. If the second bit B1 is 0, the signal generator sets the second signal 172B to "low" to equal 2*T LSB During the time period of 4*T, the third switch 179 is turned on and current is prevented from flowing through the LED or OLED 180. If the third bit B2 is 1, the signal generator sets the second signal 172B to "high" to LSB During the time period of n bit B, the third switch 179 is closed and current is allowed to flow through the LED or OLED 180. N , the second signal 172B changes due to B N The time interval set to H or L period is equal to 2 (N-1) *T LSB The sum of all time intervals is equal to the PWM period T. Figure 18A Shown above.

[0260] When the PWM signal is generated as described above, it is even possible to change the state of the first signal 172A (and thus the amplitude of the PWM signal) for any bit Bi of the truncated bit sequence. Figure 19A As shown above, the amplitude of the current is I during the time interval TLSB corresponding to the LSB B0. max , and the current amplitude is I during the time interval 4*TLSB corresponding to the third bit B2 min .

[0261] Now let's explain how a display designer will determine Figure 17A I of the circuit explained above Min and the amplitude of Imax.

[0262] Assume that the display must have a maximum brightness of 2500 cd m -2 The image data is 16 bits per pixel (per color) per frame. For the reasons discussed earlier, the chosen bit depth is 12 bits.

[0263] Based on industry know-how, LED or OLED characteristics (e.g., optical power as a function of forward current in the LED or OLED) and / or a current emulator (in its simplest form, a lookup table giving measured values ​​of L as a function of current), the desired L is achieved. Max The forward current I Max This is done for each color, i.e., the I MaxRed 、Green LED I MaxGreen and the blue LED MaxBlue For simplicity, the difference will be ignored. Once this is known, the equation can be calculated for the range from zero to L. Max The curve ΔL / L is calculated for each current step (each current step corresponds to a duty cycle increase equal to TLSB) and ΔL is calculated accordingly. This is illustrated in Table 2.

[0264] For the data in Table 2, a curve 250 representing ΔL / L as a function of L can be traced as shown in Figure 25 Explained above. Figure 25 Also shown are PQ curves 251 (Barten curve for ramp wave) and 252 (Barten curve for square wave).

[0265] Figure 28 Shown Figure 25 A close up of curve 250 of . Because of the nonlinearity and because the current (and therefore the brightness) is increased in steps (one LSB at a time), there is a discontinuity. Contrary to appearances, the function ΔL / L is not multi-valued. Figure 28 This becomes apparent when looking at the jumps from point A32 to point B32, from point B32 to point C32, ... corresponding to each single LSB increase in duty cycle.

[0266] Curve 250 is selected to be lower than the curve of the L value of the Barten curve. The designer or display still has some freedom to select the brightness below the Barten curve. Depending on the performance requirements, the brightness can be selected so that the curve ΔL / L is between curves 251 and 252 or lower than curve 252. In this case, for example, the value of L can be selected so that ΔL / L is lower than the Barten curve L Tr =107cd m -2 As done for Lmax, the display designer can determine the brightness L Tr =107cd m -2 Required forward current L Tr .

[0267] Knowing this, the curve ΔL / L can be traced as previously done. This is illustrated in Table 3. L' will only be used for luminance to avoid confusion between Tables 2 and 3.

[0268] Figure 26 shows that the amplitude of the current pulse is I Tr (curve 260) and I Max The curve ΔL / L obtained in the case of (curve 250).

[0269] Figure 26 The figure shows the visualization of the amplitude of the current pulse when the duty cycle increases from I Tr Increase to I Max Things that happen when.

[0270] Starting from point A, the duty cycle of the PWM signal increases (one LSB at a time). The brightness L increases and the ratio ΔL / L decreases. Around point B (i.e., near L=LTr), the ratio ΔL / L is constrained below the Barten curve 252. Once the average current in the LED or OLED reaches I Tr , on curve 260, the brightness does not increase any more. Curve 260 corresponds to the "low light" or LL mode, and the current I Tr Corresponding to the previously described current I Min If you need to increase the brightness further, you can use a Max (Instead of I Min ) pulse to start.

[0271] The duty cycle D is set to a value D0 such that D*I Max As close as possible (ideally equal to) I Min Now this is on curve 250, below the Barten curve 251, which means that any brightness change caused by the slight change in forward current during the transition (of the current pulse's amplitude) from IMin to IMax is not noticeable. The ratio ΔL / L jumps from point B on curve 260 to point C on curve 250.

[0272] When the duty cycle further increases from D0 to 1, the ratio ΔL / L changes from point C to point D along the curve 250. Since the duty cycle D0 is different from zero, this means that some of the bit combinations are not used.

[0273] Figure 27 A similar curve is shown. In this case, the demand is L Max =1000Cd m -2 .

[0274] In this example, LTR ≈ 43 Cd m -2 .

[0275] exist Figure 23 In the example of FIG. 1 , a first current value Imin is used for the current DAC (reference Figure 17B and 18B ), the first 8 bits encoded are converted into a current. Instead of using the last 4 bits with Imin, bits 3, 2, and 1 are set to zero, while bit 0 is implemented as a current of Imax = 16*Imin. The total current generated by the DAC is the same as if all bits were set to 1.

[0276] Now, another embodiment of a DAC that can optionally implement a more gradual transition from a first current amplitude to a second current amplitude for a PWM drive system for an LED or OLED display will be described. In any embodiment of the present invention, a low-pass filter may be used. However, remember that brightness is the same as luminance, but is perceived by the human eye, which acts as a low-pass filter that averages the sequence of light pulses. Therefore, for PWM drive pulses, the brightness or average brightness during one PWM cycle may be filtered by the eye's LPF (low-pass filter). Figure 17B A current source 170 is shown. This circuit can be implemented, for example, on an FPGA or an ASIC. The output current I Out is a function of the input signal 171. The input signal 171 is, for example, a binary signal (also referred to as b S "bit selection"). Figure 17B The embodiment shown in shows an embodiment utilizing dual current sources. Figure 17B An embodiment of the invention can be seen in Figure 18B 1 is used to illustrate how to make a "regular" (i.e., non-PWM) current DAC that can be used to drive an LED or OLED. However, it can provide current to a PWM current drive unit. In practice, PWM driven pixels are known in the art and a PWM switch can be provided between the current source and the LED or OLED, see for example US20170025064. Figure 5 , which is incorporated herein by reference.

[0277] Output current at b S When it is high (H), it is I Out =I Ref1 , or in b S When it is low (L), it is I Out =I Ref2 The output current of the current source 170 is determined by the signal b S The switch 175 or 176 is closed. Figure 17B In the example, the signal bs drives the switch 175, and the signal non-b S (NOT b S By using the inverter 177 to S (obtained by inverting the phase) to drive switch 176. For example, if the signal b S = is high, switch 175 is closed and switch 176 is open, and the first current source 173 is connected to the output terminal 172 of the current source 170. S is low, switch 175 is open and switch 176 is closed, and the second current source 174 is connected to the output 172 of the current source 170 .

[0278] Specifically, I Ref2 =N*I Ref1 .

[0279] Several current sources 170A, 170B, 170C, etc., like current source 170, may be used in parallel to form a DAC current.

[0280] exist Figure 18B In the example shown, 12 current sources 1801, 1802, ..., 1812 are used to form a 12-bit current DAC. The outputs of the 12 current sources are connected to a common node 182 via a network 180 of switches 1821 to 1832, where the currents are summed. Switches 1821 to 1832 are controlled by 12 bits (B1, B2, ...) that encode the amplitude of the current flowing through, for example, an LED or OLED connected to the common node 182.

[0281] Reference current I Ref1 and I Ref2 Changing from one current source to another is as follows.

[0282] For the current source 1801, for example, there is I Ref2 1801 =N*I Ref1 1801 =1 / 2I max .

[0283] For current source 1802: I Ref2 1802 =N*I Ref1 1802 =1 / 2I Ref2 1801 =1 / 4I max .

[0284] For current source 1803: I Ref2 1803 =N*I Ref1 1803 =1 / 2I Ref2 1802 =1 / 8I max etc,……

[0285] For current source 1812: I Ref2 1812 =N*I Ref1 1812 =1 / 2I Ref2 1811 =1 / 2 12 I max .

[0286] exist Figure 18B In the example of , the input signal of each current source is connected to the signal input bit 181 (which is thus an example of a current DAC that can be used in the first example of an embodiment of the present invention).

[0287] exist Figure 19B In the example, the input signal or control bit can be set high or low independently for each current source 1801, 1802, etc.

[0288] Specifically, the control bits can be determined by the control block 1901 as a function of the bit code B[12:1] sent to the current DAC. The bit code B[12:1] is used to control the states of the switches 1821, 1822, ..., 1832 of the switch network 1833 and to determine the selection bit b of each of the current sources 1801, 1802, ..., 1812. S status.

[0289] For example, control block 1901 allows for a less abrupt transition from low current mode to high current mode. Figure 19B The current DAC illustrated above allows synthesis of currents corresponding to FIG. 16 .

[0290] Another example of a circuit that enables a less abrupt transition from low current mode to high current mode is in Figure 20 As given above. The (binary) down counter 200 is preloaded with a 12-bit code that encodes, for example, the duty cycle. This 12-bit code B[12:1] may include an indicator bit. At the beginning of the PWM cycle, the counter decrements at each clock cycle. As long as the counter is not zero, the output signal is, for example, ON or High. The output signal is OUT[12:1]. As soon as the counter reaches zero, the output is OFF or Low and remains so until the end of the PWM cycle. The output signal activates a switch that connects the output of a current source, for example, 1800, to an LED or OLED.

[0291] At the same time, the comparator 201 compares the output of the down counter 200 (OUT[12:1]) with a predetermined binary value and changes the selection bit b of the current source 1800. S status.

[0292] It is thus possible to generate a PWM signal in which the amplitude of the current pulses will have a first value during a portion of the duty cycle and a second value during another portion of the duty cycle. Figure 21 On the commentary.

[0293] Instead of comparator 201, the output of down counter 200 can be fed to a logic lookup table. This generation allows synthesis as in Figure 22 The PWM signal on the

[0294] Any embodiment of the present invention can be used with either an OLED display or an LED display. In any embodiment of the present invention, a low-pass filter can be used. However, remember that brightness is the same as luminance, but is perceived by the human eye, which acts as a low-pass filter that averages the sequence of light pulses. Therefore, for PWM drive pulses, the brightness or average brightness during one PWM cycle can be filtered by the eye's LPF (low-pass filter).

[0295] Although the present invention has been described above with reference to specific embodiments, this is done to illustrate rather than limit the present invention. It will be appreciated by those skilled in the art that various modifications and different combinations of the disclosed features are possible without departing from the scope of the present invention.

[0296] Tables 2 and 3

[0297] Table 2

[0298]

[0299] Table 3

[0300] < / l> < / l> < / l>

Claims

1. A method of driving an active matrix display having pixels comprising a drive transistor and a light emitting device using image data, the image data representing an image to be displayed in consecutive frames, the method comprising: n-bit data is converted into (m+1)-bit data, where each frame is represented by n-bit data, where m is less than n and m and n are integers. The conversion method includes: Starting from the least significant bit or the most significant bit of the digital image data having n bits, truncating the digital image data having n bits by (nm) bits to form digital image data having m bits, One bit is added to the digital image data having m bits to form digital image data having (m+1) bits, the one bit being used as an indication, When the indication indicates a first condition, a first driving current is provided to a pixel corresponding to the digital image data having (m+1) bits, wherein the first condition is when the truncation is completed starting from the least significant bit, and When the indication indicates a second condition, a second driving current is provided to the pixel corresponding to the digital image data having (m+1) bits, wherein the second condition is when the truncation is completed starting from the most significant bit, wherein the ratio of the first drive current to the second drive current is such that it results in a brightness ratio of 2 to the power of (nm), and Among them, the (m-(nm)) bits of the first driving current corresponding to a subframe for driving a certain frame, which remain after the (nm) least significant bits have been truncated, and the (m-(nm)) bits of the second driving current corresponding to another subframe for driving the frame, which remain after the (nm) most significant bits have been truncated, and which are adjacent to the (nm) most significant bits before truncated, are in the same position in the n-bit data, and the subframe and the other subframe have the same data format.

2. The method of claim 1 , wherein: The first and second drive currents that produce a brightness ratio of 2 to the power of (nm) are determined by optical measurement of the output of the active matrix display.

3. The method of claim 2, wherein: Optical measurements are optical measurements of the set of pixels used to display the test pattern.

4. The method of claim 1 , wherein: The relationship between brightness and drive current is at least partially non-linear.

5. The method of claim 3, wherein: The optical measurements were made according to Barten using a set of pixels of the active matrix display to display a test pattern.

6. The method of claim 1 , wherein: The conversion is performed using an electro-optical transfer function, such as a gamma curve or a PQ curve, or a lookup table derived from the electro-optical transfer function.

7. The method of claim 1 , wherein: The transition from the first drive current to the second drive current is performed such that it results in a brightness increase below a minimum detectable contrast evaluated according to a model of the human visual system.

8. The method of claim 7, wherein: The model of the human visual system is based on at least one of: just noticeable difference, Barten model, Weber's law, De Vries-Rose square root law, PQ curve.

9. The method of claim 1 , wherein: The pixels of the active matrix display are LED or OLED pixels.

10. The method of claim 1, wherein: Each frame can be divided into subframes.

11. A digital drive circuit for driving an active matrix display having a pixel, the pixel comprising a drive transistor, a light emitting device, and a digital-to-analog converter, the digital drive circuit further comprising: An apparatus for converting n-bit data into (m+1)-bit data, each frame being represented by n-bit data, wherein m is less than n and m and n are integers, comprising: means for truncating the digital image data having n bits by (nm) bits starting from the least significant bit or the most significant bit of the digital image data having n bits to form digital image data having m bits, and means for adding one bit to the digital image data having m bits to form digital image data having (m+1) bits, the one bit being used as an indication, When the indication indicates a first condition, a first driving current is provided to a pixel corresponding to the digital image data having (m+1) bits, wherein the first condition is when the truncation is completed starting from the least significant bit, and When the indication indicates a second condition, a second driving current is provided to the pixel corresponding to the digital image data having (m+1) bits, wherein the second condition is when the truncation is completed starting from the most significant bit, wherein the ratio of the first drive current to the second drive current is such that it results in a brightness ratio of 2 to the power of (nm), and Among them, the (m-(nm)) bits of the first driving current corresponding to a subframe for driving a certain frame, which remain after the (nm) least significant bits have been truncated, and the (m-(nm)) bits of the second driving current corresponding to another subframe for driving the frame, which remain after the (nm) most significant bits have been truncated, and which are adjacent to the (nm) most significant bits before truncated, are in the same position in the n-bit data, and the subframe and the other subframe have the same data format.

12. The digital driving circuit for driving an active matrix display having pixels according to claim 11, wherein: The means for converting comprises an electro-optical transfer function such as a gamma curve or a PQ curve or a look-up table derived from the electro-optical transfer function.

13. The digital driving circuit for driving an active matrix display having pixels according to claim 11, wherein: The transition from the first drive current to the second drive current is performed such that it results in a brightness increase below a minimum detectable contrast evaluated according to a model of the human visual system.

14. The digital driving circuit for driving an active matrix display having pixels according to claim 13, wherein: The model of the human visual system is based on at least one of: just noticeable difference, Barten model, Weber's law, De Vries-Rose square root law, PQ curve.

15. The digital driving circuit for driving an active matrix display having pixels according to claim 11, wherein: The pixels of the active matrix display are LED or OLED pixels.

16. The digital driving circuit for driving an active matrix display having pixels according to claim 11, wherein: Each frame can be divided into subframes.

17. An LED or OLED display comprising the digital driving circuit according to claim 11.

18. The LED or OLED display according to claim 17, wherein: It is adapted to generate an output of the active matrix display for optical measurement to determine first and second drive currents that produce a brightness ratio of 2 to the power of (nm).

19. The LED or OLED display according to claim 17, wherein: It is adapted to provide a drive current to a set of pixels to display a test pattern.

20. The LED or OLED display according to claim 19, wherein: It is adapted to drive a set of pixels of the display according to Barten to display a test pattern.

21. The LED or OLED display according to claim 19, wherein: The set of pixels is a slice of a tiled display or the tiled display.

Images