> I am building a moving message display. It is 7 rows x 80 columns. > Multiplexing will be accomplished by loading shift registers with column > data and the powering sequential rows for 1.64mS each. Therefore, each row > will be refreshed once every 11.5mS. > Is the refresh rate frequent enough to eliminate flicker? Can I > refresh at a slower rate? I sure could use the extra time for other tasks, > but I don't want a flickery display!! * EVERYTHING YOU EVER WANTED TO KNOW ABOUT TV'S, LED'S, REFRESH RATES, AND FLICKER (but were afraid to ask) * [plus lots of things you didn't] When trying to design a display to show moving images, there are two important and related properties to be considered: -1- Refresh rate -2- Update rate The first factor, refresh rate, indicates how many times per second each part of the display will "flash". For computer screens, this is typically 50-90Hz. For televion sets, it is typically 50-60Hz (for objects with low detail) or 25-30Hz (for objects with high detail). For movie screens it is typically 48Hz (North America) or 50HZ (Europe). For LED message displays, it is typi- cally around 60Hz, though some LED displays have a driver for each LED and thus illuminate continuously. Some people might wonder how a television set can have a different refresh rate for high- and low- detail objects. While this may seem impossible, it is a natural result of a process called interlacing. The electron beam in the set sweeps top-to-bottom 50 or 60 times a second, but each time it only scans the even or odd lines. An object which is large enough to appear on several con- secutive scan lines will thus be drawn on every scan, while a small object that only appears on one scan line will only be drawn on every other scan. Note that the worst flicker occurs on objects which are drawn equally brightly on two consecutive scan lines without appearing on the line above; such obj- ects will appear to "bounce" between the two locations 25 or 30 times per sec- ond; flicker may be mitigated, however, by ensuring that the overall bright- ness and apparent center of any object is the same in both fields. For exam- ple, if there is a bright line on row 26 and dim lines on rows 25 and 27, the overal effect will be a non-flickering, somewhat blurry line on row 26. The second major factor to consider for moving displays is update rate. This is the number of times per second that the information on the display is act- ually changed. For movies, this is typically 24 or 25Hz; for televion it may either be equal to the "field rate" (50 or 60Hz) or the "frame rate" (25 or 30Hz). Typically, moving subjects will appear best when the update rate is once per field, but updating the image every field may cause fine details in the picture to flicker (if a small object's position happens to be on an even scan line when the display is scanning the odd ones, and vice versa, that ob- ject will not appear) and a VCR which attempts a "freeze frame" will capture the even scan lines of one image and the odd scan lines of another (producing a double-exposure effect). While having an update rate slower than the refresh rate may produce jerky motion, it may also have another nasty effect: objects which are moving around the screen may appear as double- or triple- images (if the update rate is 1/2 or 1/3 the refresh rate). This effect may be demonstrated if you write a program to scroll a text screen at precisely half the monitor's refresh rate; if you watch the text as it scrolls up, you will see "phantom" lines of text between the real ones. To understand where the phantom images are "coming from", you need to remember that a person's eye will move as he/she watches a moving object. If a 7-row LED display is scanned at a rate of 100Hz (700 lines/second) and moves left by one dot every other frame, then the user's eye will travel left at a rate of 50 dot/second (i.e. 1/2 dot per frame, or 1/14 dot per scan line). Thus, the second time a frame is drawn it will appear 1/2 dot to the right of the first time (since the eye--and the retinal image of the first dot--will have shifted left 1/2 dot) and so the user will see two dots even though in reality there is only one. Note as well that even as the display moves from one scan line to the next the user's eye will move left, so the display may appear slightly skewed. The double-vision effect is the primary reason why most LED displays are scan- ned and scrolled at the same (approximately 60Hz) rate. By using the same rates for both, each dot will be made to appear precisely once on the display (if you watch, e.g. the top of an "E" it will appear as though all five dots are all moving, even though most of them are just staying put from one frame to the next). In addition, by scanning the display bottom-to-top, the skew of the text during a normal right-left scroll will be about a 14% lean to the right, which will appear fairly attractive. The primary difficulty in producing a nice scrolling LED display is making the thing big enough to allow the unit to be scanned fast enough to avoid flicker while maintaining a 1:1 scan/update ratio to avoid "double-vision". BTW, if you are designing a display primarily for use in scrolling mode, you may visually "compress" the display horizontally at some expense in refresh rate. If your display is wired for "row-scanning", your choices are limitted to 1:1, 2:1, 3:1, etc. with anything beyond 2:1 being rather dicey (and even 2:1 being iffy). If your display is wired for horizontal scanning (e.g. you use 5x7 display modules turned on their sides) then you may use a wide var- iety of compression ratios (this works best if the row modulus is prime). I have, for example, constructed a small "nametag" with a 28x5 display which can scroll messages quite nicely (it gets about 8-10 characters legibly on the screen at once, with an apparent screen width of 48 pixels). Really cool trick... Anyway, stay tuned for when I write the "everything ELSE you wanted to know..." article.