Would like advice on the powering of my new MMD project

[zuisti]

Three months working on a commercial led matrix MMD project.
I'm using a 28 pin PIC18F2620 and 2x12 8x8 led matrices.
It's far from ready, even to a DS1307 clock drivers ...

But, to test and debug my main scrolling routines, I made
a simplified ISIS simulation and it's working fine yet.

It presents and describes almost all possible functions.

Some features:

 horizontal - vertical scroll
 left - right scroll
 big (double) chars too
 settable CLOCK
 max 623 chars/message (limited by the PIC's RAM)
 max 90 fix and one user modifiable message
 insert a running clock into a fix message

Measured in ISIS the program is fast enough, at 16 MHz PIC clock
the whole refresh rate is better than 240 Hz. In ISIS I'm using
a slower (4 Mhz) clock, so the refresh rate is better than 60 Hz.

I do not intend to make public the project, you must this understand.
But for those who are interested I uploaded a video which shows the
operation of the simulation:
 
 http://ifile.it/m0c2rwd/MyAvi.rar
 pw: sonsivri

Sorry for the quality of the video. It's caused by the asynchronity
of the two frame rates (speeds): the Proteus and the video capture.



And now here is my problem:

- how much average current should be for a good LED brightness,
  (in my case 16 rows are multiplexed, all with 96 cols)

- if the led lits as well while its average current is 10 mA:

  at worst case the max necessary current is 16 x 10 = 160 mA
  on column side (I'm using 96 x BS170 Mosfet drivers here)

  but all rows eat (at worst case) 96 x 160 mA = 15.36 A !!
  (I'm using 16 pieces TIP147 pnp darlington transistors here)

  Seems the peak currents are horrible ...

  I think the column scanning is not better, well, only
  16 leds lit at a time but multiplexed by 96 !   

  Is there a 8x8 led matrix wich lits well with 5 mA or less?    
  

Please help me!

zuisti

[SuperDrive]

How about row scanning? 96 x 10 mA = 960mA at worst.

[zuisti]

How about row scanning? 96 x 10 mA = 960mA at worst.

Yes but due the multiplexing by 16 the average current will be 1/16th, unfortunately.
So we get 16 x 10 x 96 =  15.36 A peak current at worst (at row side).
Or ... am I wrong?

[solutions]

Yes but due the multiplexing by 16 the average current will be 1/16th, unfortunately.
So we get 16 x 10 x 96 =  15.36 A peak current at worst (at row side).
Or ... am I wrong?

You are wrong (you asked...  ).  He suggested you run 10mA per 96 dot row at 1/16 the duty cycle (I'd even go with 20mA for each dot...you need to look at the data sheet for max pulsed current per dot).  You never consume more than 96*10 or 96*20 mA since you are only lighting ONE row at a time of the 16.

[zuisti]

You are wrong (you asked...  ).  He suggested you run 10mA per 96 dot row at 1/16 the duty cycle
(I'd even go with 20mA for each dot...you need to look at the data sheet for max pulsed current per dot).
You never consume more than 96*10 or 96*20 mA since you are only lighting ONE row at a time of the 16.

Sorry but I not agree..

You are right about this: "you are only lighting ONE row at a time of the 16"
But it also means that the average LED current (and also the brightness) will be 1/16th of the 10 (20)
mA. This is 10/16 = 0.625 (or 20/16 = 1.25) mA and it's far not enough...

Therefore  160 (16 x 10) mA current is needed for all column leds for a good visibility.
And yes, I must choose a led matrix with so high max pulsed current.
 
zuisti   

[carbontracks]

Honestly, 15A isn't that much current considering how many LEDs you're driving.  Your supply voltage is probably no more than 5V, so that's not even 100W.  You'll just need to make a decent buck regulator.  Look up Linear Tech or National Semiconductor for good switchmode controller ICs.

[zuisti]

Thanks carbontracks;

I'm planning a secondary, separated power supply for this big current. this will powered the
16 row driver darlingtons only.
I thought also for a secondary but unstabilized power supply, but discarded it because of the
brightness fluctuations.

And .. a watchdog function is needed because if the multiplexing stops for any reason,
the actual leds (the whole row) go bust ...

zuisti

[SuperDrive]

So you want to drive it at 160mA (pulsed). You should read the datasheet carefully if driving it at 160 mA will give a brighter output than let say at 40mA, you may wasting power into heat not light.

Some newer LED give a lot more lumens per watt than the older one.

[carbontracks]

When it comes to driving leds with PWM vs continuous currents, there' are a lot of conflicting viewpoints out there, even among experts.  Some people observe significant gains in efficiency when pulsing LEDs at very low duty cycles, while some observe the opposite.  It really depends on exactly what kind of LED it is and what brightness is desired.  I wouldn't worry about it; you're probably safe assuming that brightness will depend on average current, regardless of the duty cycle or whatever.  Of course you have to keep things within the max ratings of the LED.  Unfortunately, manufacturers don't often spec pulsed current ratings, so you may never know for sure...

[zuisti]

After a small market research (on the net : - ) I found this:
     (may be that others would be interested in)

LEDM88G 8x8 green led matrix (from Futurlec)
Price: $2.50 ($2.20 in qty 25+)

Forward Voltage: 2.1 - 2.5V
Forward Current (If): 30 mA
Peak Forward Current (Ifp): 150 mA

The link:  http://www.futurlec.com/LED/LEDM88G.shtml

This model seems appropriate, but I am doing preliminary experiments
to determine the min. peak current (at 1:16 duty cicle) to reach the
necessary brightness with the lowest power consumption.

Thanks for all replies.
zuisti

[Amon]

Hi:

It seems a good price. Is Futurlec any good? Also, how much are shipping times to Europe if anyone knows?
Thanks.

[Jagi]

When it comes to driving leds with PWM vs continuous currents, there' are a lot of conflicting viewpoints out there, even among experts.... you're probably safe assuming that brightness will depend on average current, regardless of the duty cycle or whatever.  Of course you have to keep things within the max ratings of the LED.  Unfortunately, manufacturers don't often spec pulsed current ratings, so you may never know for sure...

carbontracks, you are correct only partially. Unfortunately higher frequencies of PWM result in higher losses, because of the characteristics of the Mosfet's in themselves. Yes, you are also correct that the perceived brightness of an LED is a factor of the current thru it. I had been playing with some LEDs changing the current thru them and I found out that typically @ 18mA (continuous current) most LED's reach their max brightness. Any further increase in the current only results in more heat and less luminuous output until @ about 23mA the LED's tend to give more heat and @ 25mA they are pretty much dead in a few seconds

[shams_iqbal]

hi dear would u plz tell me what exactly scanning frequency u r using i mean 16 row are meltiplexd in second time becaused i have some practical knolwdge on these sort of displays and one of my friend is engaged on these sort display menufacturing for ur probblem plz do these thing
      in proteus u cannot analyse displays performance so plz dont consider it
increse the scanning rate as u can and scanning should be intrept driven timing based on higest perorty
     set all other task on low perorties so u will be able to get exact scanning timing
      other is plz provide ur hardware schemitic it seems that there is some trouble in ur hardware also regarding to current and one fact is that current would be less then that one rows totel current not the sum of all leds and there might be problem with ur row scanning code
       set a minor delay between each row turning on i mean add blank on displays betwen shifting rows to turn on this will help u better display

still have any problem then discuss here i will resolve ur issue because i have worked on these displays

[zuisti]

....
in proteus u cannot analyse displays performance so plz dont consider it
....

Hi "shams_iqbal";

Thanks for your reply.
My answer is here:
http://www.sonsivri.com/forum/index.php?topic=36369.msg113622#msg113622

Welcome to the board
zuisti