Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
02/13/05 16:34
Modified:
  02/13/05 16:49

Read: times


 
#87276 - Additional facts
Responding to: ???'s previous message
In a former reply I recommended the use of sub micros and a RS485 bus joining them. I want to point out why this seems to be necessary to me.

LED intensity
Some years ago a build a little toy using diffuse shining ultra high efficiency LEDs, running at only 2mA, because the toy was battery powered and was very small. At normal room brightness the display had sufficient contrast, but when I went with the toy outdoor and the sun was shining bright, then I even couldn't decide whether the display was on or not!
This is not surprising of course, as enormeous differencies in the brightness exist, which is overcome by a logarithmic scale of our eyes. Unfortunately this also means, that for an outdoor application a display must be extremely much brighter than for indoor use!!!
Some numbers: In a bright room enlightend with normal lamps brightness is about 1000lx, which differs of course. Under the direct sun up to 50000lx is present, depending where you live. So, your outdoor display must be about 50 times brighter to give the same contrast as in a room. And because brightness of a LED is rather proportional to current flowing through it, then about 50 times more current is needed, in our example 100mA!

So, for outdoor use extreme LED currents might be necessary. And, because our eyes have a logarithmic scale, a relevant increase of brightness of a display can only be achieved by increasing the LED current by an order of magnitude, means by a factor of ten! Anything less will only slightly increase the by our eyes noticed brightness!!

Multiplexing
You told us, that you divided the display into 8 modules, each consisting of a matrix of 32 x 32 LEDs and that you did a multiplexing by driving 1 of 32 columns at each module, means that 8 columns are driven at a time. You also told us, that the LED current is sinked by ULN2803/4.
If you keep in mind, that maximum current per output is only 200mA of this chip, which is even optimistic at high ambient temperatures, then only 200mA / 32 = 6.25mA is present per LED. Please note, that this current is much too little to give adequate brightness for outdoor use!!
So, if you insist on using this ULN2803/4, which seems ok, because costs are very little with this chip, then you must decide to decrease the multiplexing ratio. If you drive 1 of 8 columns for example, then you get 25mA per LED, which is already a relevant increase in intensity.

BUT: If you use a multplexing ratio of 1 : 8, then you would need to actualize 4 columns per module, means 4 x 8 = 32 columns in total. And this is a bit difficult, when doing this by the use of only one micro.
In such a case having a sub micro on each 32 x 32 matrix modul would be very helpful. Each micro should have a big SRAM, where the whole display pattern is stored. This display pattern is once transmitted by the master micro by the use of a RS485 bus, which consists of maybe three twisted pair lines: 'data', 'clock' and 'strobe', or so. 'Data' and 'clock' is needed to transmit data, while 'strobe' is used to synchronize all the modules in a suited way. The master micro is the one, that recieves the input of keyboard and converts it into the display pattern.
This use of a sub micro on each matrix modul would also heavily ease the adapting efforts if you decide to change the display!

Each 32 x 32 martrix modul contains 16 74HC595 + ULN2804 to sink the LED currents and one 74HC138 decoder followed by 4 suited column drivers to source the LED currents.

Make no mistake, it's not that I want to recommend the use of ULN2803/4 sinks. But if you want to use them, then above approach seems more suited to me.

Current driver
How difficult it is to design a suited driver to source the LED currents can be seen, if you have a look at datasheet of BD676, which once was used in a similar application. BD676 is a Darlington, which would ease the driving of this transistor. But, unfortunately, collector emitter saturation voltage of such a Darlington driver is up to 2.8V at 4A! As the voltage drop across LEDs is much smaller than 2.8V, you see, that such a Darlington would even waste more power, than needed to let the LEDs shine!
Also, if the voltage drop across driver transistor is so unpredictable high, how to manage equal currents through all the columns?

Ok, then what about the use of a single bipolar transistor, not a Darlington?
As 32 x 200mA = 6.4A needs to be driven per column, and collector emitter saturation voltage shall be adequately low, then a tenth of collector current is needed flowing through the base, which is about 0.64A in our example. This means, that you need a driver for the driver...

Then using a PMOSFET?
This is also difficult, because many of these FETs need more than -5V gate source voltage to fully turn-on.

Anyhow you solve this problem, LED column drivers must be designed very properly! Maybe it's clever to divide the column and use two drivers for every 16 LEDs? Important is, that voltage drop across driver transistor is not only low, but also equal between the individual drivers, otherwise you will notice intensity dfferencies between the columns.

Power consumption
Assume you have this Darlington driver presenting a voltage drop of 2.8V. Further assume that ULN2803/4 shows a voltage drop of 1.3V. If 1.8V drops across the LED, probably a bit more, and 2V is needed to accomplish adequate current controlling, then 7.9V is needed for the LED circuitry. Assuming a current of 32 x 200mA = 6.4A per column and 4 columns per 32 x 32 matrix modul, then 7.9V x 6.4A x 4 = 202W is dissipated by the LED stuff of only one 32 x 32 matrix modul!
This gives a total power dissipation of 1.6kW for a display containing 8 32 x 32 matrix modules.

If you would use ST2221 for all the LEDs instead, avoiding any multiplexing, then you could use a LED supply voltage of only about 3V, because ST2221 already contains the LED current controlling components (constant current sources)! Then, power dissipation per 32 x 32 matrix modul would be 25mA x 32 x 32 x 3V = 77W.
This gives a total power dissipation of 616W for a display containing 8 32 x 32 matrix moduls. This is only 39% of the power needed with above approach!!

A big advantage of using this genius ST2221 is, that you can also feed them serially and that you only need to actualize their data if you want to scroll the pattern! Here I would also recommend the use of an individual sub micro on each matrix module.

Kai

List of 242 messages in thread
TopicAuthorDate
BIG I/O expansion            01/01/70 00:00      
   Like I said            01/01/70 00:00      
      Carry on            01/01/70 00:00      
   As i said before            01/01/70 00:00      
   Fanout            01/01/70 00:00      
      Great idea, but 2 things            01/01/70 00:00      
         Thanx Sasha            01/01/70 00:00      
   Why?            01/01/70 00:00      
      Yep, nice guess            01/01/70 00:00      
   outputs            01/01/70 00:00      
      I really need'em and can't multiplex'em            01/01/70 00:00      
         But why??            01/01/70 00:00      
         Can't you go serial?            01/01/70 00:00      
            Serial??            01/01/70 00:00      
            output of a CPLD            01/01/70 00:00      
               using CPLD outputs ..            01/01/70 00:00      
                  To buffer or not to buffer            01/01/70 00:00      
                     to buffer or not            01/01/70 00:00      
   timing problems            01/01/70 00:00      
   what on earth are you doing            01/01/70 00:00      
      Exactly!            01/01/70 00:00      
   complaints and more info.            01/01/70 00:00      
      Don't worry            01/01/70 00:00      
      Consensus - not conspiracy!            01/01/70 00:00      
      Head Bashing            01/01/70 00:00      
         Head banging            01/01/70 00:00      
      more questions            01/01/70 00:00      
      Summary of data and a suggestion            01/01/70 00:00      
         A typo correction.            01/01/70 00:00      
   well            01/01/70 00:00      
   More info. on the project            01/01/70 00:00      
      More info?            01/01/70 00:00      
      thevenin terminations on LE signals            01/01/70 00:00      
         Cable            01/01/70 00:00      
            Termination must exactly match!            01/01/70 00:00      
               Explanation            01/01/70 00:00      
   Stranger and stranger            01/01/70 00:00      
      Humor            01/01/70 00:00      
         Russel            01/01/70 00:00      
            tolerance goes both ways            01/01/70 00:00      
   Bingo !            01/01/70 00:00      
   Bingo !            01/01/70 00:00      
      repeating myself            01/01/70 00:00      
         bare CPLD            01/01/70 00:00      
            Not professional?            01/01/70 00:00      
   Forget about the second Bingo ! :)            01/01/70 00:00      
      Payam            01/01/70 00:00      
         The new decoder            01/01/70 00:00      
   Cos i am just too good to you :)            01/01/70 00:00      
      Wouldn't this be a bit expensive?            01/01/70 00:00      
   Trifle expensive            01/01/70 00:00      
      synchronous design            01/01/70 00:00      
         synchronicity            01/01/70 00:00      
   Not really            01/01/70 00:00      
      ...but there are clocks            01/01/70 00:00      
   a total mess            01/01/70 00:00      
      Dear Eric            01/01/70 00:00      
         Why didn't you just say....            01/01/70 00:00      
            Sub-micros and RS485            01/01/70 00:00      
         wrong approach            01/01/70 00:00      
            here are the links (I forgot)            01/01/70 00:00      
            Eric            01/01/70 00:00      
               serial/parallel            01/01/70 00:00      
                  Cables, Multiplexing            01/01/70 00:00      
                     more on MPX            01/01/70 00:00      
                        sunlight            01/01/70 00:00      
                           more on sunlight            01/01/70 00:00      
                              Could you please explain more?            01/01/70 00:00      
                                 Where is the current limiting from            01/01/70 00:00      
                                 more explaining            01/01/70 00:00      
                           LEDs are in danger!            01/01/70 00:00      
                              4538ing is nice            01/01/70 00:00      
                              What about this            01/01/70 00:00      
                              instead of 4538            01/01/70 00:00      
                                 all over the globe            01/01/70 00:00      
                                    RE: to do what            01/01/70 00:00      
                                       no need if non-mpx            01/01/70 00:00      
                  Question to Erik            01/01/70 00:00      
                     ST2221            01/01/70 00:00      
               EMI!            01/01/70 00:00      
                  not out, but in            01/01/70 00:00      
                  Transmitting parallel            01/01/70 00:00      
                     Series termination            01/01/70 00:00      
                        correct impedance of series termination            01/01/70 00:00      
                           termination            01/01/70 00:00      
                              Output current            01/01/70 00:00      
                                 Kai, is that right?            01/01/70 00:00      
                                    Series termination methode            01/01/70 00:00      
                                       serial termination, got it            01/01/70 00:00      
                                    series termination            01/01/70 00:00      
                                 cable of 100Ohm            01/01/70 00:00      
                                    Of course!            01/01/70 00:00      
                              used for parallel of thevenin terminatio            01/01/70 00:00      
                           Impedance of flat ribbon cable            01/01/70 00:00      
                              anyhow, termination may not be needed            01/01/70 00:00      
                                 Is needed for 2m long cables!            01/01/70 00:00      
                                    that long not needed            01/01/70 00:00      
                                       Agreed            01/01/70 00:00      
                                 re: termination may not be needed            01/01/70 00:00      
                              Thanx Kai            01/01/70 00:00      
                              Flat cable characteristic impedance            01/01/70 00:00      
                                 Look closely...            01/01/70 00:00      
                                    the the            01/01/70 00:00      
                                    Doh!            01/01/70 00:00      
                                       Initial Misconception            01/01/70 00:00      
                                          I can't even spell that...            01/01/70 00:00      
                        ferrite            01/01/70 00:00      
                           ground is part of signal path            01/01/70 00:00      
                              Vcc IS part of the signal path            01/01/70 00:00      
                                 Vcc PIN is part of signal path            01/01/70 00:00      
                           Depends on what you want to filter            01/01/70 00:00      
                              PC power supply            01/01/70 00:00      
                                 4700micro is too high            01/01/70 00:00      
                                    paralell            01/01/70 00:00      
                                       This cannot work!            01/01/70 00:00      
                                          we have a problem            01/01/70 00:00      
                                             Kai, Eric            01/01/70 00:00      
                                                you need to think            01/01/70 00:00      
                                                paralell resistor current limiter !            01/01/70 00:00      
                                                   Thanks Farshid, Reinforcing PC power sup            01/01/70 00:00      
                                                      Reinforcing PC power            01/01/70 00:00      
   Jan            01/01/70 00:00      
      We don't understand each other            01/01/70 00:00      
         What if only BIG I expansion            01/01/70 00:00      
            Your Own Thread            01/01/70 00:00      
      Metastability            01/01/70 00:00      
   how about            01/01/70 00:00      
   Maybe            01/01/70 00:00      
      Divide the problem            01/01/70 00:00      
      Own ideas            01/01/70 00:00      
      sidesteps            01/01/70 00:00      
   Thanks            01/01/70 00:00      
   I dunno Kai..            01/01/70 00:00      
      I dunnoJez            01/01/70 00:00      
      Agreed            01/01/70 00:00      
         CE radiation            01/01/70 00:00      
   Thanx for criticisims from Eric & Kai            01/01/70 00:00      
      the original poster            01/01/70 00:00      
         Eric, Kai            01/01/70 00:00      
            if you can't explain            01/01/70 00:00      
               paralell resistor current limiter !            01/01/70 00:00      
      Then, why don't you discuss it?            01/01/70 00:00      
         taking our warnings seriously and discus            01/01/70 00:00      
      world record            01/01/70 00:00      
         world record +1            01/01/70 00:00      
   Eric            01/01/70 00:00      
      Cost Is Final Issue            01/01/70 00:00      
         lowest cost            01/01/70 00:00      
            Yeah.......            01/01/70 00:00      
            now drop the multiplex            01/01/70 00:00      
      Let's compare parallel vs. serial            01/01/70 00:00      
         Time to confess            01/01/70 00:00      
            simple bitbang            01/01/70 00:00      
               something ain't mising here?            01/01/70 00:00      
                  more            01/01/70 00:00      
               not even that...            01/01/70 00:00      
                  you can, but            01/01/70 00:00      
                  Cool            01/01/70 00:00      
      urband legends            01/01/70 00:00      
      No influence            01/01/70 00:00      
   This heart of mine on fire            01/01/70 00:00      
      HC595's and ULN2803's            01/01/70 00:00      
         typpo            01/01/70 00:00      
         column xistors            01/01/70 00:00      
            Instantaneous!            01/01/70 00:00      
               saturation and off state not on and off            01/01/70 00:00      
                  What is a 'xistor'??            01/01/70 00:00      
                     Okay, transistor            01/01/70 00:00      
                     xistor            01/01/70 00:00      
                  Saturation and Gain            01/01/70 00:00      
                     FETs and gate capacitance            01/01/70 00:00      
                        Fets Vs Transistors            01/01/70 00:00      
                           Which MOSFET            01/01/70 00:00      
                           More Qs on P-MOSFETs            01/01/70 00:00      
                              Best text ...            01/01/70 00:00      
                              Mosfets            01/01/70 00:00      
                              Low 'Ugs' and low 'Ron' is crucial!            01/01/70 00:00      
                                 Bingo!            01/01/70 00:00      
                                 Soft-start circuit            01/01/70 00:00      
      answers and comments            01/01/70 00:00      
         ghosting            01/01/70 00:00      
            what            01/01/70 00:00      
               ghosting            01/01/70 00:00      
                  ok            01/01/70 00:00      
                     this is probably not the case with Payam            01/01/70 00:00      
                        no, worse            01/01/70 00:00      
            P-MOSFETs            01/01/70 00:00      
         typicals            01/01/70 00:00      
            These probably might go            01/01/70 00:00      
            yes, you are and so what            01/01/70 00:00      
               "them" explained            01/01/70 00:00      
   another solution            01/01/70 00:00      
   Serial is a must!            01/01/70 00:00      
      Serial is not a must!            01/01/70 00:00      
   Additional facts            01/01/70 00:00      
      Multiprocessor?            01/01/70 00:00      
      To ULN or not to ULN            01/01/70 00:00      
         ULN2803, human eye response            01/01/70 00:00      
            typical again            01/01/70 00:00      
         consider everything            01/01/70 00:00      
         Equivalent brightness with pulsing            01/01/70 00:00      
      Re: Additional facts            01/01/70 00:00      
         You put many of them...            01/01/70 00:00      
            Re: You put many of them            01/01/70 00:00      
               the matrix revisited            01/01/70 00:00      
   here are some more additional facts            01/01/70 00:00      
      NFI            01/01/70 00:00      
         Me, not subtle ???            01/01/70 00:00      
            Subtle as a brick!            01/01/70 00:00      
   one little detail            01/01/70 00:00      
      30° is it enough?            01/01/70 00:00      
   Calculating ... please wait!            01/01/70 00:00      
      answers/comments            01/01/70 00:00      
         six-layer boards?            01/01/70 00:00      
            no 2 layers possible            01/01/70 00:00      
               6 layer for 20$???            01/01/70 00:00      
                  here            01/01/70 00:00      
                     PCB discussion extracted            01/01/70 00:00      
               a trace about 1 inch            01/01/70 00:00      
                  20A on a 2mm track!            01/01/70 00:00      
                  holy smoke            01/01/70 00:00      
                     Okay.            01/01/70 00:00      
                        again            01/01/70 00:00      
                           colloquialism            01/01/70 00:00      
               Calculating a bit...            01/01/70 00:00      
                  'Three' layer design            01/01/70 00:00      
                     layers - LED - EMI            01/01/70 00:00      
                        LEDs don't understand ground bounce            01/01/70 00:00      
                           But the four 74HC595 and ULN2804 do!            01/01/70 00:00      
                              595s don't, ULN ... not important            01/01/70 00:00      
                                 not so            01/01/70 00:00      
                           Ground bounce            01/01/70 00:00      
                              Sure, shall take care of it.            01/01/70 00:00      
      Don't underestimate heat dissipation!            01/01/70 00:00      
   Kai            01/01/70 00:00      
      Payam            01/01/70 00:00      
         less complex/less expensive            01/01/70 00:00      
         Did you try the multiplex?            01/01/70 00:00      
            Yes            01/01/70 00:00      
            go ahead and try            01/01/70 00:00      
            Square root increase            01/01/70 00:00      
               Distributed power supply            01/01/70 00:00      

Back to Subject List