led MATRIX CURRENT DRAW

[MrDEB]

I plan to use AA batteries or AAA. If I added right its 10v but thinking perhaps rechargeable?
I understand what your saying about the leds and not having alot of extra parts. The ULN2803 idea I got from Graham's Tetris on a pic project. He is using low voltage pic which on the end project I hope to use the LF or K version of the 18F4520. Am trying to think ahead but I get more issues as I go. I agree that I could eliminate the ULN2803 but I have then on the board at present. In my next prototype board I will make refinments and reconsider if the uLN2803 is really needed. KINGBRIGHT just notified me that they can ship me the brighter matrix but in common anode configuration. I would have to wait 2-3 weeks for the common cathode version. Thinking about it the common anode might be better?? Need to think on that but at present if I ever get back to the board, I hope to get it reconfigured.

 

[Mike - K8LH]

MrDEB, did you mention what it is you're building? We're very curious...

 

[MrDEB]

An LED 5 x 7 matrix that I want to be visible as far as 50 ft in daylight.
I have a brighter matrix on order from KINGBRIGHT.
I tested out another matrix (same model) using a 5v battery pak just out of curiosity.
one LED on with a 38ohm resistor I measured 10ma at best.
Went with two 5v packs (measured 8.4v) a 220 ohm resistor and I get 21ma (duh)
the 5v(measured at 4.80v) just won't drive the LED segments. In theory it should?? but its a no. I was just curious of how far I can push the LEDs at 5 v.
Taking into account that fresh batteries would MAYBE do the deed but it won't .
The data sheet says 4Vf but ?? it does not caculate in real testing.

 

[be80be]

Your doing something really wrong here. With a 40 ohm resistor you should beable to light them as bright as they get a led at a time

 

[JonSea]

If you look at the data sheet, the typical Vf is 4.0v for the red element and 4.4v for the green element but the maxiumu Vf for either the red or the green elements can be as high as 5 volts for the display (I think) you have. You should plan for the worst case conditions; forward voltage is also a function of temperature, so it may stray over even a greater range.

View attachment 67829

One way to know for sure with a particular LED is to illuminate it with a constant current source at the desired current at measure the voltage across the LED. With a sample of 35 LEDs, you may find most of them near the typical value but some will probably be at the Maximum value.

 

[be80be]

What don't add up is he said he's reading 21 at 8 volts that half what it should be
And so is the 10 there something going on here we don't know about

You like them green things I gave you some seeing we are lighting LEDs lol

 

[JonSea]

Well, reading MrDEB's post, I'm not really sure what a "5 volt battery pack" is.

A fresh alkaline battery starts out at 1.3 volts and fairly quickly drops to 1.0 volts. So the first question is what is the battery voltage when the LED is illuminated? If the batteries are nearly dead, the open-circuit voltage might read ok, but under load, decrease to nothing.

Remember too that LEDs aren't linear. The forward voltage across the LED varies with the current.

 

[be80be]

The nominal voltage of a fresh alkaline cell is 1.5 V

from https://en.wikipedia.org/wiki/Alkaline_battery

So 4 start at 6 volts and would drop to 4.8 when dead and give about 10 hours of use.

MrDebs real life figures have something going on that we don't no like leds are getting power but not enough to light but for one a short comes to mind here.

I played with one of these a long time back and was getting funny readings like this and it was a short I could light one led but not fully on and with the lights off you could barely see it was trying to light a whole row so I would test this Like jon saying and figure what you really need to power your leds

 

[MrDEB]

Today I am going to solder together a perf-board to eliminate the mass of jumper wires with alligator clips just to "organize" the whole mess. Also want to see in real life if an NPN or a PNP transistor really performs differently as in using a computer simulation (used TINA and LTSPICE). One consideration that has yet to be discussed is the voltage drop across the resistor? should be considered??
The battery pack is a holder w/ 3 AAA batteries. I put two in series for the 8.40 volts. Batteries are not FRESH. Need more 220 ohm resistors so a FRESH 9v battery is in order as well or assemble a 9v power supply using a 7809.
Another though is a constant current source. Just a thought to maintain a constant brilliance from the LEDs.

 

[JonSea]

Before you get too far off the rails, maybe you should consider laying out a board with individual LED? Something like this pattern with all the LEDs having their own resistor and wired in parallel groups? This would eliminate the multiplexing issue AND the Vf issue if the proper LEDs were selected. You could use as many LEDs in each group, limited only by the current draw on the battery and the current-handling ability of a Uln2003.

I would probably make three groups, the right arrow pieces, the left arrow pieces and the bar. Then you can have arrow head plus bar for direction, and perhaps everything for stop?

Individual LEDs can be blindingly bright on 20 mA with a Vf in the less than 3 volt region.

 

[MrDEB]

comparing transistors

Assembled a separate board and using one RED and one GREEN led segment so any multiplexing etc has NO INFLUENCE on readings. see attached schematic. I swapped connections for the NPN and PNP using same battery etc. the resistors connected to the cathodes are not connected to any other connections except what is pictured. The anodes were swapped from one emitter to the other with a DMM in series as pictured. Using ohms law the calculations are not exact.
It appears that the PNP switch applies more voltage and more current to the LEDs. IMO this application is more efficient using a PNP verses an NPN. Correct me if I am wrong, please before this train crashes.
This testing is pretty much what Jonsea suggested but I had it completed before I read his suggestion. I guess I was one step ahead??--lol
as far as controlling the matrix, am going to connect one resistor per colum(10) and one PNP per row (7) and as suggested one led at a time. Considering perhaps a constant current source to supply the anodes along with the 10v or ? battery. A LDO regulator and CCS in a to92 package. End design will be ALL smd.



Comparing a PNP switch to an NPN switch

Using the attached schematic / inserting an NPN in place of the PNP

Battery voltage = 8.56v nothing connected
resistor = 327.9 ohms
LED Vf = 4
measured current draw as pictured in series with the LED anode
This is a Bi- Colored matrix
NPN with base connected to +
RED current draw = 12.7ma
GREEN = 11.1ma
voltage at emitter = 7.74v

PNP with base connected to ground
RED = 22ma
GREEN = 20.8ma
voltage at emitter = 8.47v

 

[JonSea]

What was I thinking? I don't even have tickets for this train! Drop me off at the next stop.

 

[MrDEB]

I used an actual led matrix so REAL TIME measurements can be made. This is one of three that I have on hand.. Taking into consideration cold weather, battery life etc. I am considering the worst case scenario hopefully and using actual components that will be utilized. No guessing but then Murphy's law!!
Now about them tickets, they are free but the seats may come loose from the floor--LOL



"A DIYer spends hours looking at a circuit and spends days talking about it""

 

[MrDEB]

forgot to ask, whats your take on using a CCS to power the anodes via the collector on the transistors? This should help in regulating the current between the GREEN and RED LEDs hopefully??

 

[MrDEB]

using an LED matrix already assembled solves the durability issue some what. I already figured out somewhat how to reconfigure the code to enable only one LED at a time w/ a delay for POV

 

[JonSea]

My sincere recommendation is that you use the simplest combination of hardware and software possible to make this happen, since you seem to have trouble with both. That would be the parallel LEDs and resistors I described above, controlled by several gates of a ULN2003.

This is easy to understand and to control and offers you the greatest chance of success.

Oh look..it's my stop. I'm just going to jump onto that feight train going anyhere else.

 

[be80be]

Jon now your on my train lol simple would be the best here

 

[MrDEB]

Now I am getting lost?? I am following the suggestion to make it simple. Am eliminating the ULN2803's, using transistors on the anodes and enabling only one LED segment at a time. The LEDs in the matrix are two chips per segment which would explain the 4vf. Now you mention parallel resistors and leds?? unless I missed something??
The schematic I posted to describe my testing is as simple as I think it can get.
If I go with individual LEDs I louse the ruggedness of the matrix block. The code to light a single LED in sequence is easy, just a for next loop with PORTD.BITS(x) and a 10ms delay between each bit enable (this is what was suggested at the start of this thread. One LED at a time

 

[JonSea]

See my post #50. This eliminates the need for muxing, the problem with needing greater than 5 volts to drive the LEDs and many programming issues. Each LED is driven directly from 5 volts and is on 100% of the time.

If you mux with a single LED on at a time and you have seven LEDs, each LED is only on 1/7 of the time, so it will be dimmer. You may be able to drive the LEDs harder, but the max current is only 30 or 40 mA, so even if you CAN drive it harder, it won't make much difference in perceived brightness.

You're laying out a circuit board anyway; if you use individual LEDs you have all kinds of options.

Just my opinion of course but I've seen enough of your level of understanding to believe my method might lead you to success.

My train is pulling away...no wifi on this freig...........

 

[MrDEB]

I get your idea BUT!

I am using a 5 x 7 matrix block not individual LEDs which would be easier as you stated. The matrix has all the LEDs in a secure package that is practically indestructible.
The CCS idea is not viable but I wish it was seeing how I need 20ma per segment at or above 7volts. Going to try out yours and Burts idea of one LED on at a time but I may end up inserting the ULN2803's back in and use a higher voltage to drive the LEDs. The 5v just won't cut the mustard.
Back to the drawing board.
Thinking about using the transistors on the anodes instead?
I appreciate your input as it gets me to rethinking my off the wall ideas but looking at some LED drivers and how they incorporate them I got to wondering about another idea for a different project. But this train needs to get back on track.
QUESTION is this an electric diesel train or steam?