APPLICAT ION NOT E T ips in measuring and reducing Philips Semiconductors
Abstract The image jitter and OSD jitter are mentioned in this application note. Jitter measuring instruction is also included. Basically this application note is to help engineers to check and avoid the jiter problems in monitor-set. Philips Electronics N.V. 1996 All rights are reserved. Reproduction in whole or in part is prohibited without the prior written consent of the copy-right owner. The information presented in this document does not form part of any quotation or contract, is believed to be accurate and reliable and may be changed without notice. No liability will be accepted by the publisher for any consequence of its use. Publication thereof does not convey nor imply any license under patent- or other indus-trial or intellectual property rights. 2
APPLICATION NOTE Tips in measuring and reducing Author(s): T. H. Wu Regional Sales Office - Southern East Asia Industrial Regional Application Laboratory Taipei-Taiwan Keywords Jitter Deflection On Screen Display etc. Date: 13 March, 1997 3
Summary The OSD and image jitters in monitor-set are mentioned, jitter measuring is also included in this report. Because of my background which is monitor application support, the accent lies on basic jitter measurement of separate blocks inside the monitor. 4
CONTENTS 1. INTRODUCTION... 6 2. DEFINITION OF JITTER... 6 3. CHARACTER JITTER... 8 4. MEASUREMENT SET-UP FOR MEASURING OSD JITTER... 9 5. THE VALUES OF OSD PLL LOOP FILTER... 12 6. OSD JITTER WITH DIFFERENT SOFTWARE CONFIGURATIONS... 12 7. JITTER PHENOMENA ON VIDEO IMAGE... 13 8. JITTER MEASURING INSTRUCTION FOR A MONITOR... 13 9. PERCEPTION LIMIT FOR JUST VISIBLE JITTER... 21 10. INFLUENCE OF PERCEPTION... 22 11. HOW TO AVOID JITTER PHENOMENA IN MONITOR-SET AT DESIGN STAGE... 22 12. REFERENCE... 23 5
1. INTRODUCTION The goal of this report is to help engineers measure jitter proformance on OSD jitter or on image jitter in monitor-set, then take actions to tackle the blocks in monitor which can be improved. 2. DEFINITION OF JITTER Considering a perfect and stable square wave, that means if you measure all the time intervals between two edges, they all have exactly the same value. This is our reference. Considering now an output signal from a real phase lock loop circuit, which is "locked" on our perfect reference signal. The time interval between the reference edges and the edges of the PLL output is not constant and is stochastically changing. We say "the output signal is jittering" Consider again the perfect reference and an output-signal of a PLL. Refer to figure 1. Fig. 1 6
The reference edges (in this case the trailing ones), are numbered from 0 to infinite, by I and are at locations 0.T, 1.T, 2.T,..., i.t. The trailing edges of the output signal can also be numbered from 1,... to infinite and w.r.t. the time reference they are at locations Bo, B1,...Bi,... The reference position, being the triggermoment of the instantaneous measurement, is called REF. We can now measure B0 or B1 or... Bi, for each about 300 instantaneous measurements. For each Bi, a histogram if Bi will look like this, figure 2. pp-value = the peak-to-peak value of Bi, for those 300 measurements Fig. 2 On the figure 2, a number quantifying jitter, is the pp-value or 6 times standard deviation value which calculated from a histogram measurement. 7
3. CHARACTER JITTER When displayed pictures becomes large, the real-time picture creation becomes visible due to the visibility of displayed pixels. Picture artefacts caused by, for instance, on screen display menus can therefore become very visible. One of these artefacts are caused by the PLL of the character generator. Because OSD menus are built up in a dot matrix structure, imperfections in these matrix structures become visible as a jittering of the characters, which results in vague and slightly moving characters called "character jitter". See figure 3. w/o jitter with jitter Fig. 3 8
4. MEASUREMENT SET-UP FOR MEASURING OSD JITTER Measurement set-up: Fig. 4 A Modulation Domain Analyser, such as HP53310A can be used to generate for jitter measurement. The jitter on the fast blanking(fb) has been measured with respect to the Hsync signal. The jitter is measured over several fields. In this situation, it is possible to see how much the jitter is fluctuating in time. See for example figure 5 and 6. 9
Fig. 5 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync at pin 6 of PCE84C886 ==> Fast Blanking at pin 1 of PCE84C886 9.5ns 832ps 10
Fig. 6 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync at pin 6 of PCE84C882 ==> Fast Blanking at pin 1 of PCE84C882 8.509ns 768.71ps 11
5. THE VALUES OF OSD PLL LOOP FILTER According to experimental results for different horizontal frequencies, a "saw-tooth" distortion occurs at different values of PLL loop filter. To get the best quality OSD for each video mode, the values of PLL loop filter should be selected carefully. Different values of PLL loop filter for different video modes can be selected to reach optimized OSD quality. For uc embeded PLLs, e.g. PCE84C886 or PCE84C882, they are sensitive. All possible noise should be minimized, especially ground noise. Any improper PCB layout may cause a serious noises problem which directly affects the picture quality of the OSD. 6. OSD JITTER WITH DIFFERENT SOFTWARE CONFIGURATIONS Though the OSD picture quality is mainly contributed by the hardware circuits and the OSD uc, e.g. PCE84C886/882 itself, there are still some performance differences between different software configurations. Therefore it's also very important to know how to improve the OSD picture quality from the software point of view. For more details, please refer to application note, Taiwan/AN9311. [2] 12
7. JITTER PHENOMENA ON VIDEO IMAGE 8. JITTER MEASURING INSTRUCTION FOR A MONITOR The jitter phenomenon is an effect where some noise is visible on a vertical line. As the human eye is not an objective measuring instrument, it is very important to have a measuring set up that gives reproducible figures. These figures are important for two reasons: 1. To be able to measure the influence of every modification made in the application. 2. To exchange measuring data between verious laboratories. Finally the customers can judge the performance on the screen. But for debugging and improvement, is vital. 13
For measuring setup, following are needed. 1. Modutation domain analyzer HP 55310A 2. AC current probe Tektronix A6302 3. Selfmade terminator for A6302 Measuring setup configuration is shown as in figure 8. - The trigger input of the 55310A is connected to the vertical sync. - Input A is connected to the horizontal sync. - Input B is connected to the test points of horizontal deflection. Fig. 8 14
With the correct setting of the HP55310A, the time delay between Hsync and the zero crossing of the deflection current can be measured. Check with an oscilloscope that the slopes are free from ringing etc. Select a slicing level at the inputs of the HP55310A that corresponds with the slicing level of the IC or circuitry concerned. Due to e.g. storage time modulation, pin balance correction, EW modulation, cross talk etc., the center line can be curved and will show a large figures at the measuring result from the HP55310A. To avoid this, the HP55310A must be triggered externally with the Vsync. In the horizontal menu a time delay of 1/2 vertical period must be set (e.g. 6.5ms ; 72Hz). The time scale must be set at 0.1ms/div. This setting will result in a measurement during 1ms in the middle of the screen. Now only the real jitter is measured. Now switch to the histogram mode. Set accumulate one and measure 1000 samples and note the standard deviation and peak to peak value. As there is always some tolerance on the jitter figures measured (up to 20-30% can occur), it is desirable to take the average of 5 measurements of the standard deviation and peak to peak values. As a rule of thumb, the ppvalue is about 6x the standard deviation-value. Normally the standard deviation-value for working 30kHz and 82kHz horizontal frequency is about 5ns and 3ns. In case to high jitter values are observed, this measuring system canalso be used to measure the jitter of separate blocks inside the monitor as in figure 9. Suggestions are: 1. Incoming sync --> incoming sync (orientation value 0.1/0.6 ns (stdev/pp)) 2. Incoming sync --> sync at inputsync processor. In case the sync is processed by the micro controller. 3. Incoming sync --> H-drive output. 4. H-drive outpuyt --> collector pulse. 15
Fig. 9 16
Following are the measuring results for a monitor, Fig. 10 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync ==> Incoming Hsync 4.143ns 564.56ps 17
Fig. 11 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync ==> H-DRV 5.429ns 715.87ps 18
Fig. 12 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync ==> H-out collector pulse 5.214ns 736.34ps 19
Fig. 13 Video timing: Measured point: Jitter pp-value: Jitter Standard deviation: VESA 800x600/72Hz Incoming Hsync ==> H-deflection Yoke 5.429ns 741.58ps 20
9. PERCEPTION LIMIT FOR JUST VISIBLE JITTER The limit for visible jitter has still to be investigated by means of perception test. The image shall appear to be stable. This can be accomplished by insuring the peak-to-peak variation in the geometric location of picture elements which does not exceed 0.0002mm per mm of designated viewing distance for the frequency range of 0.5Hz to 30Hz. If the viewing distance is 500mm, the acceptable jittering should be little than 0.1mm. Fig. 14 21
10. INFLUENCE OF PERCEPTION The jitter performance of a monitor-set is very dependent on several factors. For instance, it is possible that two different monitor sets can be evaluated totally different, although they have the same jitter! Some parameters which influence certain performance are summaried as follows. - A set with a good focus performance, will result in a better visibility of jitter and demands a better jitter performance. - 480 line display systems are more sensitive for jitter than 768 line display systems, e.g. XGA mode. This could be explained by an optical integration by the eye of jitter in the same line. - A multiburst is more critical for jitter than one vertical line. On screen, the jitter of character "I" can easily be displayed very stable while jitter of character "M" is more critical to be seen. - The greater the contrast, the jitter is observed easier. - Jitter at alternative lines is much more visible than low frequent jitter. 11. SOME TIPS REDUCE JITTER PHENOMENA IN MONITOR-SET AT DESIGN STAGE - Short tracks at sync input of sync processer - no sync processing at input - To avoid mismatch in base drive - To avoid using proportional base drive - To avoid using PWM to control horizontal phase shift - To avoid crosstalk between horizontal drive and B+ control drive - To avoid jitter caused by EHT compensation feedback loop - Refer to AN96129, Monitor ICs FAQs for application hints 22
12. REFERENCE [1]. AR6-641027JSJ, High line-frequency On Screen Display character generators. by J. van Stratum [2]. Taiwan/AN9311, OSD Microcontroller Optimization Techniques. by Albert Hu, Ben Chang, Richard Chen and T.N. Yang [3]. Jitter measuring instruction. by Han Misdom 23