1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

questions about powering leds in large matrix like mikes macmux design

Discussion in 'General Electronics Chat' started by deuplonicus, Nov 9, 2010.

  1. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    using mike's design shown here I was trying to up the current driving. and was wondering if the 74hc595's with uln2803s in chain with them would work.

    Also if the Si2312BDS 5 amp N-fets would need to be better?

    The leds will be 20ma 3.1v blue 5mm, in the same amount as his design.
     
  2. Gayan Soyza

    Gayan Soyza Active Member

    Joined:
    Oct 23, 2006
    Messages:
    1,822
    Likes:
    19
    Location:
    Colombo
    NO.

    He is doing a row scan not a column scan.Positive supply to the LED's will comes from 74HC595 & flow current to GND via row side.So you don't need additional ULN drivers.

    I haven't use N-fets as he mentioned because those are very hard to find.

    Use darlington on row side.
     
  3. Mike - K8LH

    Mike - K8LH Well-Known Member

    Joined:
    Jan 22, 2005
    Messages:
    3,642
    Likes:
    109
    Location:
    Michigan, USA
    With that "cathode row" design, you could replace the 74HC595's with Micrel MIC5891 serial-to-parallel "sourcing" driver ICs but they're a bit expensive. Use two sets of Si2312BDS N-FET "sinking" row drivers (inputs connected in parallel) with outputs from each set driving the rows for a single 8x64 block.

    With an "anode row" design, you could use MIC5821 or TPIC6C595 "sinking" column driver ICs and P-FET "sourcing" row drivers. Or you could use the 74HC595's driving ULN2803A's you mentioned.
     
    Last edited: Nov 10, 2010
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. Gayan Soyza

    Gayan Soyza Active Member

    Joined:
    Oct 23, 2006
    Messages:
    1,822
    Likes:
    19
    Location:
    Colombo

    Good Day Mike

    Where were you all this time! I was waiting.

    I had a similar issue on the current design 16 X 64 multiplex.

    For a 16 rows design you are divided into two rows & do the scan in this case brightness will be 1/8.When its comes to real life adding two sets of column shift registers to the top row 8s (8, 595’s) & bottom row 8s (8, 595’s) becomes very hard.

    I planed to scan 1/16 using 8, 595’s by adding some 64 PNP transistors to the each 595 outputs.

    What’s your comment sir!!
     
  6. Gayan Soyza

    Gayan Soyza Active Member

    Joined:
    Oct 23, 2006
    Messages:
    1,822
    Likes:
    19
    Location:
    Colombo
    My current LED arrangement is "Cathode row design" that's why when the no of rows gets bigger the hardware part goes hell of a mess.

    Thinking to do my LED arrangement like "Anode row design" so all I have to place ULN with each 595 on column side & only need 16 current source transistors on row side.

    :) I need your comments Mike.
     
  7. Mike - K8LH

    Mike - K8LH Well-Known Member

    Joined:
    Jan 22, 2005
    Messages:
    3,642
    Likes:
    109
    Location:
    Michigan, USA
    Hi Gayan,

    I prefer to use a lower duty cycle to reduce "peak" current requirements. To get the brightness of 10-ma "average" current per LED you would need to provide about 160-ma "peak" current to each lighted LED on your 1/16th duty cycle design. This is beyond the typical 100-ma "peak" spec' for many LEDs.

    The "anode row" design provides a lot more choices for your sinking column drivers. A 74HC595 driving a ULN2803 should work fine. You can also replace that combination with a single but somewhat more expensive IC like the TPIC6C595 or MIC5821 (many others too).

    Regards, Mike
     
  8. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    Thanks Gayan and Mike,

    So is there any reason to use anode row driving vs cathode row driving aside from hardware differences?

    Also, would adding shift registers or demux's to the rows to reduce IO lines at the MCU cause a lower refresh rate?
     
  9. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    mike, another question, my LEDs are 20ma, someone on allaboutcircuits forum informs me that your design, as it says, only sources 4ma to the leds, how can I up this to 20ma?
     
  10. blueroomelectronics

    blueroomelectronics Well-Known Member

    Joined:
    Jan 21, 2007
    Messages:
    12,536
    Likes:
    170
    Location:
    Toronto, Canada
    You'll want to drive them at more than 20mA as they're only on 1/8 of the time. 60mA is common and may not damage them as quickly as 160mA. Also HE red are nice and bright with less current that standard red LEDs.
    One way is PNP driving transistors.
     
  11. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    so a pnp on the output of each shift register? or arrays if I can find them?

    would I also need to up the 5amp cathode row drivers?
     
  12. Mike - K8LH

    Mike - K8LH Well-Known Member

    Joined:
    Jan 22, 2005
    Messages:
    3,642
    Likes:
    109
    Location:
    Michigan, USA
    The design uses inexpensive 74HC595 ICs to provide about 32-ma "peak" current or about 4-ma "average" current to each LED. If you want 20-ma "average" current you'll need to provide about 160-ma "peak" current which is probably above the "peak" current spec' for your LEDs.

    As I mentioned earlier, if you want a brighter display then you should either replace the 74HC595 ICs with higher current sourcing driver ICs, or as Bill (blueroom') mentioned, you could add PNP or P-FET sourcing driver transistors between the column lines and the 74HC595 outputs. The sourcing driver ICs are easier but they're also more expensive (unless you 'sample' them as I did).

    Good luck on your project.

    Cheerful regards, Mike

    <added>

    You got a link to that AllAboutCircuits thread? I can't find it.
     
    Last edited: Nov 11, 2010
  13. Ubergeek63

    Ubergeek63 Well-Known Member

    Joined:
    Apr 16, 2008
    Messages:
    1,886
    Likes:
    37
    74ABT574s driven byte serial speeds up the interface and can pull 100mA per pin to ground
     
  14. Mike - K8LH

    Mike - K8LH Well-Known Member

    Joined:
    Jan 22, 2005
    Messages:
    3,642
    Likes:
    109
    Location:
    Michigan, USA
    How exactly would the 74ABT574's speed up the interface?

    This particular MacMux design loads the 16 column driver ICs (with 128 bits of column data) during each row update interval by writing 16 bytes and clocks onto the 10 pin buss in just 64 instruction cycles. I'm not sure how you might do any better than that with daisy-chained 74ABT574's.

    Other than your claim of a faster interface, the 74ABT574 "sinking" driver might be a good choice for cathode column drivers in an anode-row display. If you could really get 100-ma "sinking" current (the spec' sheet I downloaded says 64-ma), you wouldn't be violating your LED "peak" current spec' and you might get about 12.5-ma "average" current, which is pretty darned bright.
     
    Last edited: Nov 11, 2010
  15. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0

    Thanks for the help mike, here's is the AAC forum thread: large irregular shaped LED matrixs - All About Circuits Forum

    I don't want to risk breaking anything so if 160ma is too risky, and everyone seems to suggest 60ma, then I would prefer to use something not risky. so I can add a source after the 74hc595s pnps or whatever sampling is always nice, good idea.

    My only question left at this point would be the Nfets on the cathode rows, would they also have to be able to sink more given a source of 60-100ma per led?
     
  16. Mike - K8LH

    Mike - K8LH Well-Known Member

    Joined:
    Jan 22, 2005
    Messages:
    3,642
    Likes:
    109
    Location:
    Michigan, USA
    Yes. Figure it out. How much current does the N-FET row driver need to sink if all 128 LEDs in that row are lighted at 60-ma? at 100-ma?
     
    Last edited: Nov 11, 2010
  17. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    hehe, thanks, I just don't want to figure it out wrong. The help of people like you, gayan and tom66 (on ACC) and others around me, are invaluable in understanding the concepts behind these circuits.

    through you guys answering my silly questions I learn a lot in a much shorter time period.

    one last silly question since I'm kind of new to using sources and sinkers. I do need resistors on the cloumns as per the voltage being fed to them and the wanted ampheres per led, correct? I can figure out the numbers from there.

    Thanks! Soon as I place my order, I will start the project thread with tons of fun building pics (as puttine this many LEDs in a sign and soldering will take a while)
     
    Last edited: Nov 11, 2010
  18. Ubergeek63

    Ubergeek63 Well-Known Member

    Joined:
    Apr 16, 2008
    Messages:
    1,886
    Likes:
    37
    my bad ... i was thinking the circuit was running bit serial.

    the spec gives 64mA while maintaining correct logic levels, it gives as max 128mA per output if you do not care about logic levels.

    we do it here but i think we only pull 50mA per pin on a 1000 LED matrix
     
  19. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    Out of curiosity, are you running a matrix from serial over a computer through the mcu? What kinds of speeds do you achieve?


    Mike, looks like I need 12.8A of N fet power for the rows if I use 100ma currents per LED

    and if I use 60ma * 128 = 7.68 amps of N fet current ability

    I'm going to start searching for some suitable N fets, What should I look for? low Rds? I am not really sure what made you pick the Si2312BDS, so if you could inform me I would love to learn why.

    I would really like to use the MIC5891, as it seems ideal and I actually find it to be cheaper to use these than drive transistors with 74hc595s..... however it would inihibit me to 60ma of output as that would be a 480ma load on each mic5891 which each can sink 500ma of current.



    SO, right now I'm going to send out some 12.5% duty cycles out to my one of my LEDs with some transistors I have laying around at 60ma and 100ma. I will report back to you after I decide what I truely need.
     
    Last edited: Nov 11, 2010
  20. Ubergeek63

    Ubergeek63 Well-Known Member

    Joined:
    Apr 16, 2008
    Messages:
    1,886
    Likes:
    37
    10% of an 8051 at 12MHz drives the display at 95% duty cycle refreshing a 64x10 array 100 times a second. (column off, send 8 bytes, column on, wait, repeat)

    8051 also runs a PID and a keyboard interface and scrolls the display
     
  21. deuplonicus

    deuplonicus New Member

    Joined:
    Sep 14, 2010
    Messages:
    60
    Likes:
    0
    how in the world did you manage that amazing refresh rate/duty cycle?
     

Share This Page