Patent Grant
11263956
The present disclosure relates generally to methods and devices for driving a display. More particularly, this disclosure relates to methods and devices that compensate image data to improve the refresh rate and the uniformity in brightness for an LED display.
Modern LED (light emitting diode) display panels require higher grayscale to accomplish higher color depth and higher visual refresh rate to reduce flickering. For example, a 16-bit grayscale for a RGB LED pixel allows 16-bit levels (216=65536) for red, green, and blue LEDs, respectively. Such a RGB LED pixel is capable of displaying a total of 655363 colors. One of the methods commonly employed to adjust LED grayscale is Pulse Width Modulation (“PWM”). Simply put, PWM generates a series of voltage pulses to drive an LED. When the voltage of the pulse is higher than the forward voltage of the LED, the LED is turned on. Otherwise, the LED remains off. Accordingly, when the pulse amplitude exceeds a threshold, the pulse duration (i.e., pulse width) of the PWM signal decides the on-time and off-time of the LED. The percentage of on-time over the sum of on-time and off-time (i.e., a PWM cycle) is the duty cycle, which determines the brightness of the LED. Configurations and operations of an exemplary LED display system, which includes LED topology, circuitry, PWM engines, etc., are explained in detail in U.S. Pat. No. 8,963,811, issued Feb. 24, 2015, as well as in the U.S. patent application Ser. No. 15/901,712, filed Feb. 21, 2018.
Another parameter for an LED display is the grayscale value, which is the level of brightness of the LED display. In a 16-bit resolution LED display, the grayscale value ranges from 0 (complete darkness) to 65535 (maximum brightness), corresponding to duty cycles from 0% to 100%. When the grayscale value is low, the brightness level of an LED is low. Conversely, when the grayscale is high, the brightness level is also high. LED displays often experience performance issues at low grayscale values.
A further parameter for the LED display is its Grayscale Clock (“GCLK”) frequency, which is related to the maximum number of GCLK cycles (“GCLKs”) in a data frame and the refresh rate of the display. In addition, a frame rate is the number of times a video source feeds an entire frame of new data to a display in one second. The refresh rate of an LED display is the number of times per second the LED display draws the data. The refresh rate equals the frame rate multiplied by the number of segments.
One of the advantages of PWM is that power loss in the switching devices is low. When a switch is turned off, there is practically no current. When the switch is turned on, there is almost no voltage drop across the switch. As a result, power losses in both scenarios are close to zero. On the other hand, PWM is defined by the duty cycle, switching frequency, and properties of the load. When the switching frequency is sufficiently high, the pulse train can be smoothed and the average analog waveform can be recovered. However, when the switching frequency is low, the off-time of LED will be noticeable and appears as flickers to a viewer.
Scrambled PWM (“S-PWM”) modifies a conventional PWM and enables a higher visual refresh rate. To accomplish that, S-PWM scrambles the on-time in a PWM cycle into a number of shorter PWM pulses that sequentially drive each scan line. In other words, a total grayscale value is scrambled into a number of PWM pulses across a PWM cycle. In a conventional PWM scheme, there may be only one PWM pulse so that the LED is lit continuously for a period of time, leaving the LED unlit for the remainder of the time. In contrast, S-PWM allows the LED to emit light in consecutive short pulses in the PWM cycle so that the light pulses spread across the PWM cycle more evenly, avoiding or reducing flickers.
One PWM cycle has a number of GCLK cycles that equal 2 to the power of the number of control bits:
Number_of_GCLKs=2NUMBER_OF_CONTROL_BITS.
For example, a 16-bit grayscale has 65536 GCLKs. Note that the number of GCLKs in one PWM cycle equals its grayscale value at the maximum brightness, i.e., the maximum pulse width. In some S-PWM, the total number of GCLKs can be divided into MSB (most significant bits) and LSB (least significant bits) of grayscale cycles. Each PWM cycle is divided into a number of segments (or sub-PWM cycles) according to the following equation:
Number_of_Segments=2NUMBER_OF_LSB.
For a video source of a 60 Hz frame rate and a PWM cycle length of 8000 GCLKs, one may divide the PWM cycle into 32 segments (LSB=5) so that each segment has a pulse duration of 250 GCLKs. A total of grayscale value of 1600 GCLKs can be distributed into 32 segments at 50 GCLKs in each segment, potentially increasing the refresh rate up to 32 times. However, when the PWM pulse duration (i.e., pulse width) in the segment is shorter than the time it takes to raise the LED voltage above its forward voltage, the LED remains unlit. U.S. Pat. No. 9,390,647 provides a solution that extends the pulse duration by adding a fixed number of GCLKs to the pulse. However, such an S-PWM scheme results in large increments in the optical energy output at the low brightness level, as explained elsewhere in this disclosure. Other technical schemes may require a second power source to provide an additional driving current to extend the pulse duration, adding complexity and costs to the electrical system for the LED display.
Accordingly, there is a need for new systems and methods that improves image quality of the LED display without the shortcomings of the existing technologies.
An embodiment of the LED display system of this disclosure includes and LED display panel coupled to a driver circuitry. The driver circuitry includes a scrambled PWM generator, a register, and a memory. The scrambled PWM generator receives an image data of a grayscale value of (X+K). X is a grayscale value of a data from an external image source and K is a compensation value supplied by the driver circuitry.
According to one embodiment, the scrambled PWM generator distributes the grayscale value (X+K) into a plurality of segments according the following set of rules: when (X+K) equals or is smaller than G0*S0,
S=ceil((X+K)/G0) and R=mod(X+K,G0) (1);
when (X+K) is larger than G0*S0,
M=floor((X+K)/S0) and L=mod(X+K,S0) (2).
In Equations (1) and (2), G0 is a grouping number and S0 is a preset segment number stored in the driver circuitry. S is the number of output segments, among which S−1 segments has a pulse width of G0 GCLKs and one segment has a pulse width of R.
Further, L is the number of segments that each receives a pulse width of M+1. Each of the remaining S0−L segments receives a pulse width of M. Note that the unit of the pulse width or the grayscale value is GCLK. For example, a pulse width of M means a pulse width that has a time length of M GCLKs.
The group number G0 can be pre-determined based on experience or obtained by calibrating the LED display for flickering. It can be stored in a memory in the driver circuitry. The compensation value K is related to a first set of calibration data obtained at high brightness and a second set of calibration data obtained at low brightness of the LED display. For example, K=(floor(p*X)+q)−X, wherein p is derived from the first set of calibration data and q is derived from the second set of calibration data.
In some embodiments, the LED array in the LED display panel can be arranged in either the common cathode configuration or the common anode configuration. The LED display panel can be a large wall display for indoor or outdoor use. The LED display panel can also be a microdisplay for hand-held devices.
The current disclosure also provides a method for operating an LED display system. The LED display panel is coupled with a driver circuitry having a scrambled PWM generator. An image data X is to the driver circuitry. Data X is compensated by multiplying a calibration coefficient p in a multiplier. The data is further compensated by adding to it another constant q in an adder. As such, a total compensation value K is added to X so that the compensated image data has a value of (X+K).
The compensated image data (X+K) is then sent to the scrambled PWM generator. The scrambled PWM generator scrambles the image data into a number of segments to generate short PWM pulses to be sent to the power or current sources.
The current disclosure further provides a method for compensating image data for an LED display system. The LED display panel is driven by a driver circuitry having a scrambled PWM generator. The driver circuitry is connected to a video source. The input image data from the video source is X. The compensated image data is floor(p*X)+q. The values of p, or q, or both are obtained by calibration. For example, the display panel is calibrated at a high brightness level for uniformity to determine the value of p and calibrated at a low brightness level for uniformity to determine a value of q. Alternatively, the values of p, or q, or both are pre-determined empirical numbers.
The values of p, or q, or both can be independently determined for each individual LED in the LED display. Alternatively, q is a constant for LEDs of a same color in the LED display, p is a constant for LEDs of a same color in the LED display, or both.
The teachings of the present disclosure can be readily understood by considering the following detailed description in conjunction with the accompanying drawings.
Reference will now be made in detail to several embodiments of the present disclosure(s), examples of which are illustrated in the accompanying figures. It is noted that wherever practicable similar or like reference numbers may be used in the figures and may indicate similar or like functionality. The figures depict embodiments of the present disclosure for purposes of illustration only. One skilled in the art will readily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the disclosure described herein.
Used herein, the term “couple,” “couples,” “connect,” or “connects” means either an indirect or direct electrical connection unless otherwise noted. Thus, if a first device couples or connects to a second device, that connection may be through a direct electrical connection, or through an indirect electrical connection via other devices or connections.
In this disclosure, the term “low brightness” (i.e., low grayscale) generally refers to situations when the input signal length is low, e.g., less than 4 times the rise time of the LED, or less than 3 times the rise time of the LED. Conversely, the term “high brightness” (i.e., high grayscale) refers to situations when the input signal length is high, e.g., more than 4 times the rise time, or more than 6, 8, or 10 times the rise time of the LED.
The current disclosure provides an inventive S-PWM scheme C. For illustrative purposes, X is the grayscale value of the input image data in one grayscale input period; K is the compensation value added to the input image data; S0 is the segment number; and G0 is the length of each segment.
In S-PWM scheme C, when (X+K) equals or is smaller than G0*S0, S=ceil((X+K)/G0) and R=mod(X+K, G0). S is the number of output segments, among which S−1 segments has a pulse width of G0 GCLKs and one segment has a pulse width of R. R is a positive integer less than G0. Used herein, an output segment is a segment having at least 1 GCLK pulse width while a segment having no output pulse is hereby referred to as a “dark segment.” Accordingly, (S0−S) segments are dark segments.
In contrast, when (X+K) equals or is larger than G0*S0, M=floor((X+K)/S0) and L=mod(X+K, S0). L is the number of segments that each has a pulse width of M+1, while the remaining S0−L segments each has a pulse width of M.
Applying this rule to the scenario of distributing 1 to 320 GCLKs into 32 segments (S0=32), assuming the grouping number is 8 GCLKs (G0=8), the distribution of the grayscale value can be illustrated in Tables 1 and 2 below. Table 1 shows the case for distributing grayscale values from 1 to 256 GCLKs (e.g., grayscale value ≤S0×G0=256), while Table 2 shows the result for distributing grayscale values from 257 to 320 GCLKs.
Table 1 shows that, when the grayscale value is smaller or equal to S0*G0, the available grayscale data are first put into one single segment until the PWM pulse width in that segment reaches G0 before the remaining grayscale data is put into another segment that has less than G0 PWM pulse width. Accordingly, the maximum PWM pulse width in each segment is G0 (i.e., eight in this example). Consequently, at very low grayscale values, the priority is to fill individual segments until the segment has a pulse width G0 while the remaining segments receive no signal and remain dark. Note that when the grayscale value equals G0*S0, every segment has a pulse width of G0.
The rule of distribution changes when the grayscale value is larger than G0*S0. As shown in Table 2, the GCLK number in excess of G0*S0 is distributed 1 GCLK a time to a segment until all 32 segments have (G0+1) GCLKs. Then the excess GCLKs beyond (G0+1)*S0 is distributed one GCLK a time to each segment until all 32 segments have (G+2) GCLKs.
Accordingly, in this embodiment, the rule of distributing grayscale value into the segments when the grayscale value is larger than S0*G0 is the same as in the conventional S-PWM scheme. Nonetheless, when the grayscale value is low, i.e., less than S0*G0, this method maximizes the number of segments having at least a pulse width of G0.
Since S-PWM scheme B increases the PWM value in each of the 32 segments by the same number GLCKs, the LED is either on in all segments or remains unlit in all segments, which does not allow fine-tuning at low brightness. In contrast, S-PWM scheme C allows increasing the limited amount of PWM value in individual segments under certain conditions so that the LED emits light at least in some segments even at very low brightness levels. Accordingly, the S-PWM scheme B results in large increments in the optical energy output while the S-PWM scheme C allows fine-tuning of the optical energy output.
In some embodiments of the disclosure, the compensation value K is obtained by calibration. For example, the calibration is carried out through photo capturing and adjusting of the brightness of individual LEDs in the LED display. This calibration is normally carried out at high brightness. The purpose is to achieve uniformity in brightness across the display. In such a calibration, each individual LEDs in the LED display receives that same image data—same X value. A first photo of the LED display is taken, which shows variations of brightness of the LEDs. A first data is added to the image data and sent to the LEDs. A second photo is taken. Adjustments of the input image data are made and photos are taken until the uniformity in brightness meets the pre-determined criteria. The corresponding compensation value K for this LED display is set in this matter.
In a specific embodiment, each LED pixel is a RGB LED pixel that contains a red LED, a blue LED, and a green LED, each receiving its respective input image data Xi and obtaining a coefficient pi, i=r, g, or b. The coefficient pi obtained from the calibration for each individual LED is then stored in, e.g., a look-up table in a memory, such as a SRAM. The memory can be built on the same chip together with the driver circuitry or on a different chip coupled to the driver circuitry chip. The coefficient is retrieved when needed, e.g., at the power-up of the LED to preload the calibration data to a register in the driver circuitry.
In a further embodiment, the calibration process is carried out both under a high brightness level to obtain a first set of coefficients pi and under a low brightness condition to obtain a second set of coefficients qi. In some embodiments, the performance characteristic at low brightness is flickering of the LED display, which can be monitored by visual inspection. Assuming, at a low brightness condition, an individual LED receives an input image data Xi and is assigned a coefficient qi after the calibration process. Alternatively, qi can be stored in a memory in the driver circuit. Accordingly, coefficient pi, qi, or both are assigned to each individual LED.
For a 1920×1080 pixel color LED display, there can be up to six matrices of calibration data—Pr, Pb, Pg, Qr, Qb, and Qg—each being of 1920×1080 in size. The elements in these matrices are pimn or qimn, in which i=r, g, or b, while m is the row number and n is the column number of the color pixel in the LED array.
Used herein, p, pi, and pimn may be used interchangeably. Each of them is a coefficient assigned to an LED in the LED array while p refers to a coefficient assigned to an LED in general, pi emphasizes the color of the LED, and pimn indicates the color and position of the LED. The coefficient p may differ from one LED to another. Alternatively, the same coefficient p may be applied to all LEDs of the same color in the LED array. Likewise, q, qi, and qimn may be used in the same manner. Further, a matrix of p, pi, or pimn is designated as coefficient matrix P while a matrix of q, qi, or qimn is designated as coefficient matrix Q.
One of the calibration methods in this disclosure is an open loop process that derives coefficients p and q using two or more images of the LED array. First, the calibration is carried out at a high brightness level. For each red, green, or blue color, the LED array is provided with a uniform input data set. E.g., when pwm_R=65535, pwm_G=0, pwm_B=0, the red LEDs in the LED array are lit. An image of the LED array is taken using a camera. The image is then processed to deduce the brightness of each red LED to obtain an intensity matrix A. The lighting efficiency of each red LED is proportional to the corresponding matrix element. A red LED at column m and row n has the intensity of Armn while the mean intensity of all red LEDs in the LED array is mean(A). The corresponding coefficient pimn can be derived according to the equation:
pimn=Armn/mean(A) (3)
The same process is repeated when the input to the LED array is green or blue. As such, red, green, and blue color each has a corresponding coefficient pimn. The coefficient matrix P is a matrix of coefficients pimn.
The second step of the calibration process may be carried out at a low brightness level. Likewise, for each red, green, or blue color, a uniform data set is sent to all LEDs of the same color in the LED array but at a low brightness level. For example, the data set (pwm_R=200, pwm_G=0, pwm_B=0) lit all red LEDs in the array at a low brightness level. An image of the LED array is taken and an intensity matrix B is extracted from the image. Likewise, each of red, green, and blue color is calibrated to obtain its corresponding matrix B.
Assuming the input data of high brightness is XH and input data of low brightness is XL. The following equations can be obtained:
Bimn=(XL−qimn)*pimn*XL/XH (4), and
qimn=XL−Bimn*XH/XL/pimn (5)
qimn is the coefficient a color LED obtained at a low intensity. The matrix of coefficients qimn is designated as matrix Q.
The calculations above assume the imaging efficiency at the high brightness level and at the low brightness level are the same, which may not be accurate. Used herein, the term “imaging efficiency” means the ratio between the brightness extracted from the image of the LED to the actual brightness of the LED. To solve this problem, in some embodiments, qimn is obtained using images of the LED array at two different but low brightness levels. Two intensity matrix B1 and B2 are derived from two images taken at two different low brightness levels having input data X1 and X2, respectively. In this case qimn can be obtained by solving two linear equations for each color LED as follows:
B1imn=(X1−qimn)*pimn; (6)
B2imn=(X2−qimn)*pimn, (7)
qimn=X1−B1imn(X1−X2)/(B1imn−B2imn) (8).
Another exemplary calibration method is a closed loop or iterative process, which uses an imaging system to adjust coefficient q for each LED to obtain a uniform brightness. This process uses an adjustment loop in which the value of q is modified to reduce the measured image variations until the displayed image is at or below a pre-determined level. Details of this embodiment is described below, referencing
As shown in
X′=(X+Q)*P (9)
The matrix X is a uniform matrix. As such, Q applies an adjustment to the matrix X. Used herein, a uniform matrix is a matrix in which all elements are of a same value.
After the adjustment, in S4, an image is taken and an intensity matrix B is extracted using the image. The difference between B and a uniform matrix is calculated as the error matrix E in S5 according to Equation 8.
E=B−mean(B) (10)
In Equation (10), mean(B) is the matrix of the mean intensity value of LEDs of in the LED array, which is a uniform matrix.
In S6, the error matrix E is compared with a pre-determined threshold value. If E is at or smaller than the threshold value, Q is output as the result of the calibration. The threshold value can be a small fraction of mean(B), e.g., 1% of mean(B) or 0.5% of mean(B). Each element q in the matrix Q is the coefficient obtained at a low brightness less or the equivalent thereof. If E is larger than the threshold value, the process continues to S7.
A compensation matrix C is calculated using the error matrix E in S7.
C=−k*E (11)
k is a constant smaller than the lighting efficiency of the LED array, i.e., 50% of the average lighting efficiency of the LED array.
In S8, the compensation matrix C is added to matrix Q to obtain an adjusted Qnew.
Qnew=Q+C=Q−k*E=Q−k*(B−mean(B)) (12)
Qnew is assigned to Q in S2 to calculate a new input data matrix X′ to the LED array and start a new iteration.
When the observed intensity of an LED pixel in the array is high, its corresponding element in the error matrix E for the LED array would also be large. To that the LED pixel requires a more significant compensation c for that pixel (C=−k*E). Consequently, the new data input to that pixel qnew is smaller (Qnew=Q+C), which results in a lower output brightness. Q is further adjusted using the new output intensity matrix B and a smaller k. As such, every iteration modifies the input data to give brighter LED pixels smaller input values so that the output image becomes more uniform. The iteration continues until the variation of the images reaches a pre-determined level or does not have any further noticeable reduction. The resulting Q matrix provides q, which is used in calculating the compensation value K.
In certain embodiments, e.g., when light emitting from LEDs are consistent and uniform, it may not be necessary to apply a different q to each individual LED. Instead, all LEDs of the same color in the LED array can use one set of calibration data at low brightness, high brightness, or both. I.e., at low brightness, all red LEDs use the same qr, all blue LEDs use the same qb, and all green LEDs use the same qg, thereby reducing qimn, (three matrices qr, qb, and qg, each being of 1920×1080 in size) to three numbers. Independently from what values of qr, qb, and qg are used for low brightness, at high brightness, all red LEDs may use the same Pr, all blue LEDs use the same pb, all green LEDs use the same pg, thereby reducing pimn (three matrices pr, pb, and pg, each being of 1920×1080 in size) to three numbers. Such simplifications reduce the size of the memory needed for storing the calibration data. In these embodiments, the values of q and p can be selected based on experience or obtained by calibration.
Both q and p are used in determining the compensation value K so that optimal compensation of the LED can be accomplished in the full range of brightness levels.
In another embodiment of this disclosure, the grouping number G0 and the segment number S0 can be determined based on experience or obtained by calibration. S0 and G0 are stored in the driver circuitry of the LED display, e.g., in a register. In the calibration process, an initial G0 value (e.g., 8) and/or an initial S0 (e.g., 32) values are set in the driver circuitry, the LED display is run at various brightness levels, especially low brightness levels, to test performance characteristics such as flickering and brightness uniformity. The G0 and S0 can be adjusted until the performance meets or exceeds a pre-determined criteria.
Note that the values of pi, qi, G0, and S0 can be obtained through calibration of the LED array or can be per-determined without calibration, e.g., based on experience.
Data from the multiplier enters an adder where the second set of calibration data, qi, is added. The second set of calibration data is obtained under a low brightness condition, i.e., low brightness calibration. Assuming the calibration data adds qi GCLKs to N1, the output data N2 from the adder equals (N1+qi) or (floor(pi*X)+qi). As such, the compensation value Ki=(floor(pi*X)+qi)−X. Therefore, the compensation value Ki is informed by both the high brightness calibration and the low brightness calibration, corresponding to the curves shown in Panel C of
The calibrated image data (X+K) is sent to a S-PWM engine, which receives a preset segment number S0 and a preset grouping number G0 from a register and generates digital PWM signals. The digital PWM signals are sent to a plurality of power sources. The power sources in turn drive a scan-type LED display panel, which may be either a common anode configuration or a common cathode configuration.
In the common anode configuration, the LED display panel has an array of RGB LED pixels arranged in rows and columns. The LED array has a plurality of common anode nodes. Each of the plurality common anode nodes operably connects anodes of LEDs of a same color in a row to a corresponding scan switch. The cathodes of the LED pixels in a same column are connected to a power source.
In the cathode configuration, the LED array has a plurality of common cathode nodes. Each of the plurality common cathode nodes operably connects cathodes of LEDs in a row to a corresponding scan switch. The anodes of LEDs of a same color in a column of LED pixels are connected to a current source.
Many modifications and other embodiments of the disclosure will come to the mind of one skilled in the art having the benefit of the teaching presented in the forgoing descriptions and the associated drawings. For example, the driver circuit can be used to drive an LED array in either common cathode or common anode configuration. Elements in the LED array can be single color LEDs or RGB units or any other forms of LEDs available. The driver circuit can be scaled up or scaled down to drive LED arrays of various sizes. Multiple driver circuits may be employed to drive a plurality of LED arrays in a LED display system. The components in the driver can either be integrated on a single chip or on more than one chip or on the PCB board. Further, the display can be any suitable display, including large outdoor display panel or small micro display for cell phones. Such variations are within the scope of this disclosure. It is to be understood that the disclosure is not to be limited to the specific embodiments disclosed, and that the modifications and embodiments are intended to be included within the scope of the dependent claims.
This application is a continuation in part of U.S. patent application Ser. No. 15/945,497, filed on Apr. 4, 2018, the content of which is incorporated herein by reference in its entirety.
| Number | Name | Date | Kind |
|---|---|---|---|
| 8963811 | Li et al. | Feb 2015 | B2 |
| 20030006994 | Suzuki | Jan 2003 | A1 |
| 20100168957 | Takahashi | Jul 2010 | A1 |
| 20110074799 | Wu et al. | Mar 2011 | A1 |
| 20150294642 | Kurokawa | Oct 2015 | A1 |
| 20160019829 | Li et al. | Jan 2016 | A1 |
| 20160027385 | Li et al. | Jan 2016 | A1 |
| 20180240409 | Li et al. | Aug 2018 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 20200184881 A1 | Jun 2020 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15945497 | Apr 2018 | US |
| Child | 16791864 | US |
