What kind of current can this ic handle?

Hi guys,

I intend to use this IC to source current for an array of leds:
http://www.allegromicro.com/en/Products/Part_Numbers/2981/2981.pdf

Looking at the current charts today however, I was a bit alarmed. They seemed to be saying that the chip could only handle 110mA with a 100% duty cycle on all 8 pins.

I then noticed however that the charts were the amount of current at 15v.

As I will be supplying only 5v, I'm guessing that the chip will be able to handle a lot more current at that voltage.

Using these calculators:
http://www.the12volt.com/ohm/page2.asp

I determined that with a 100% duty cycle on 8 output pins, using a total of 110ma at 15v, I would be putting 1.65 watts through the chip. (P = E x I)

I then calculated that if I was powering the chip with 5v, and I wanted to put the same amount of power through the chip, 1.65 watts, that the total current I could use on the outputs would be 330mA. (I = P / E)

What I want to know is, am I doing this correctly? Are my calcualtions right? Will the chip be able to handle outputting 3x as many milliamps if I cut the voltage by a third?

What I want to know is, am I doing this correctly? Are my calcualtions right? Will the chip be able to handle outputting 3x as many milliamps if I cut the voltage by a third?

I don't think so, the current limits are the current limits up to maximum rated voltage of the drivers. The higher power levels of 15v Vs 5v X current flow is what the load will see in power dissipation not the output transistors of this driver chip.

Keep in mind that 110ma limit only applies if you have all 8 outputs on continuously. If you are driving different numbers of outputs at any given instance or is you decrease the duty cycle by saying driving the inputs with PWM signals at different duty cycles you can push that spec higher as the duty cycle graphs show in the data sheet.

Lefty

What I want to know is, am I doing this correctly?

Nope: this is one of those times when you should focus on little details, and ignore the big picture :wink:

What you need to worry about is, not the total power, but the amount dissipated by the 2981, which is essentially the same at 5V as it is at 15V. What counts is the voltage difference between the power supply input and the driver output (the VCE(SAT) number in the datasheet), which increases some as current goes up, but doesn't change with different supplies. To get power dissipation, you multiply VCE(SAT) times the current, so reducing the supply doesn't help.

You should also note that, for the 2981, VCE(SAT) is in the 1.5-2V range, which is very significant when you're feeding it a 5V supply: you'll only get about 3.5V at the driver outputs. This tends to be an unpleasant surprise for newbies who are puzzled because their LEDs are dim or their motors run slow with what they think is a 5V supply.

The limit is 350mA per output at 100% duty, the absolute maximum rating 500mA.

If you see an absolute maximum rating then aim to be well below it, not at it - so here 350mA is the limit.

You will need to check the power dissipation since this is a Darlington driver it will drop 1.5V or more when switched on which is 0.5W or so per channel (4W total)

The total power dissipation seems about 1.5W from the graph they give, so you can't run all the channels on continuous at 350mA note, I suggest limiting the average total output current to 1A or so.

Keep in mind that 110ma limit only applies if you have all 8 outputs on continuously. If you are driving different numbers of outputs at any given instance or is you decrease the duty cycle by saying driving the inputs with PWM signals at different duty cycles you can push that spec higher as the duty cycle graphs show in the data sheet.

The problem is, while I won't usually have all 8 outputs on continously, there are times which that will indeed be the case. The prop I'm building is like a set of VU meters, adjustable with a knob, and it will be possible to max them out with that knob, so there will be times when they're all lit.

So if this chip can only put out 110mA total when all 8 outputs are on, then it's useless for my needs. The Arduino itself can put out more than that all at once. In fact, I've already rigged a test where I put 150mA through one of my bargraphs on a 1/8th duty cycle to see how it would look if I lit the whole array with that much current, and it was much dimmer than I would like.

I thought this chip was gonna save me a lot of soldering, but now I don't really see the point of it. I could source more current from the Arduino's pins directly. I wanted the chip because rather than putting 15mA through each LED I wanted to be able to put 30 or 40ma through each led 1/8th of the time. And that means the chip would have to be able to source at least 300mA. I knew PNP transistors were up to the job of handling as much as 500mA EACH, so I expected this one chip with 16 of them to be able to handle at least that much in total, and initially it looked that way from the max specs on the datasheet. :frowning:

The limit is 350mA per output at 100% duty, the absolute maximum rating 500mA.

Wait, so you're saying the chip CAN put out 350mA per output pin?

These charts are so confusing. How do you know that? The data sheet says:

500 mA output source current capability

Recommended for high-side switching applications that benefit
from separate logic and load grounds, these devices encompass
load supply voltages to 50 V and output currents to -500 mA.

Output Current IOUT –500 mA

Output Source Current (Outputs Open) lOUT VIN = 2.4 V, VCE = 2.0 V 2 -350 mA

Allowable peak collector current as a function of duty cycle

None of this seems to indicate whether it's total output or per pin. Or am I missing something here in the annotation?

Hm, but wait; there's this too:

The suffix “A” indicates an 18-lead plastic dual in-line package
with copper lead frame for optimum power dissipation. Under
normal operating conditions, these devices will sustain 120 mA
continuously for each of the eight outputs at an ambient
temperature of +50°C and a supply of 15 V.

Hm... well that bit there does seem to indicate that the chart is in fact the output per pin. But if that's the case, why don't they advertise the total anywhere? You'd think they'd want to advertise that their chip can output 960mA at once?

Well, unless someone tells me different, I'm gonna go under the assumption that this thing CAN supply 120mA per output pin when all 8 pins are in use with a 100% duty cycle. That seems to be what the datasheet says. And Mark says that the mA in that chart are per output.

In that case, that's good. I'm back in business. That's more than enough current for my needs.

Oh, and thanks Ran for pointing out the voltage drop. I don't know nearly enough about transistors yet, and I wouldn't have taken that into account when doing my resistor calculations.

Now if only I could determine from this data sheet what the safe maximum to pulse the LED's with is:
http://www.us.kingbright.com/images/catalog/SPEC/DC10GWA.pdf

It says 140mA on a 1/10 duty cycle, but it also says that the pulses have to be less than 0.1ms which is 100 nanoseconds. That's much faster than the 1785 nanoseconds I was going to multiplex my leds at. (70hz update of 8 10-segment displays = switch between bargraphs 560 times a second)

It's possible that 0.1ms is a mistake, but given how much current that is, I think it may not be. Regardless, I dont know how I should plot that on a graph to figure out what it is safe to pulse these leds with for 1.7 milliseconds. It does list a power dissipation there, and I suspect that might be key to figuring this out, but I havwn't been able to determine what to do with it.

Hey Ran,

You mentioned that voltage drop from the first transistor array of 1.5-2v... It just occured to me that the LEDs themselves wiill drop the voltage further. And then the second transistor array, the sink, will drop it even more. Is this going to be a problem?

Here again is the spec sheet for the source array:
http://www.allegromicro.com/en/Products/Part_Numbers/2981/2981.pdf

And this is the spec sheet for the leds:
http://www.us.kingbright.com/images/catalog/SPEC/DC10GWA.pdf

And this is for the darlington array I'll be using to sink the current:
http://focus.ti.com/lit/ds/symlink/uln2803a.pdf

It says 140mA on a 1/10 duty cycle, but it also says that the pulses have to be less than 0.1ms which is 100 nanoseconds.

Keep in mind... those are extreme test conditions and supplied as a reference... not a "requirement".

At "full on" you only NEED to supply 20ma to light an LED in the bar. If you don't EXCEED the guidelines (140ma) in the datasheet... you should be fine. Do some tests... with those limits in mind.

chip would have to be able to source at least 300mA

Keep in mind... just like the LED wants a MAX of like 20ma CONTINUOUS current... the CHIP can likely handle more if it's also... not continuous. Though... my opinion is that 1/8th duty cycle is just gonna be DIM unless it's REAL FAST.

Keep in mind... those are extreme test conditions and supplied as a reference... not a "requirement"

I know that the 140mA is not a requirement. I'm sure the leds will run fine with as little as 10mA. However, if I multiplex them on a 1/8 duty cycle with only 10mA, they'll be way too dim.

So I need to know what I CAN supply to them, and for how long. And the problem is, while that 140mA on a 1/10 duty cycle at 10000hz tells me I can put more than the 25mA absolute maximum listed for the leds through them, I don't know how to interpolate the data. I can't tell from those numbers if say putting 50mA through each led for 1.5ms on a 1/8th duty cycle is safe.

Do some tests... with those limits in mind.

Tests will only tell me if I'm putting so much current through them that they'll burn out in a few seconds, or a few minutes. They won't tell me if the leds are gonna burn out a month after I've sold this prop to someone. I sold a bunch of props in which I may have put quite a bit of current through 4017 chips when I didn't know better. So far they've worked fine, but I've heard those chips can't supply much. I'm worried in a year or two if I'll start having angry customers coming to me complaining their props no longer work.

Though... my opinion is that 1/8th duty cycle is just gonna be DIM unless it's REAL FAST.

Actually, I've found that the faster you pulse the LEDs, the dimmer they are. So if you run them on a 1/8th duty cycle, it's better to update the display at 70hz than to update it as fast as possible. I think this is because the LED doesn't get a chance to reach full brightness for the amount of current going through it. Or maybe it's your eye which is too slow to react. I know perception comes into play here. I saw a paper which said 50% current 100% of the time appears dimmer to your eye than 100% current 50% of the time.

Oh and 15mA per led on a 1/8th duty cycle isn't that bad with these leds I've got. They're only 16mcd leds too. They're not super bright by any means, but they look okay. Just not spectacular. And probably too dim when outdoors. That's why I want to put a bit more current through them. If I can just double the brightness, I think I could be satisified with them.

Hm... I just realised something. I won't be running those darlington arrays off 5v. I'll be running them off 9v, won't I? I'll be putting 9v into the arduino, and the regulator will regulate that down to 5v, but the whole point of using the arrays is to put more current through the leds than the arduino can supply. So I think I'll have +9v on the source ic input, and 0v on the sink ic ground... so 9v will flow through the leds in the array, controlled by the 5v coming off the arduino pins which also makes its way to ground through the leds.

In which case, I need to calculate the resistors I'll need for the leds for 9v minus whatever the voltage drop was. 1.5v I think Ran said.

0.1ms which is 100 nanoseconds

Not in this part of the galaxy, it isn't.

So I think I'll have +9v on the source ic input

Maybe you will, maybe you won't: the odds are pretty good that your 9V supply is unregulated. You need to make sure that the supply feeding your LED drivers is regulated, because relatively small changes in LED voltage can make destructive differences in current.

And don't forget that whatever you use for a sink driver will have a voltage drop across it, just as the source driver does.

0.1ms which is 100 nanoseconds

Not in this part of the galaxy, it isn't.

Sorry, I meant microseconds. The Arduino doesn't have a nanoseconds timer. :slight_smile:

Maybe you will, maybe you won't: the odds are pretty good that your 9V supply is unregulated. You need to make sure that the supply feeding your LED drivers is regulated, because relatively small changes in LED voltage can make destructive differences in current.

So I suppose I should pick up a 5v regulator that can handle my current needs?

My supply will be a couple of 9v batteries btw. I'm not sure how their voltage dropping as they discharge could be destructive, though it might mean the leds dim or go out ealier than I'd planned.

And don't forget that whatever you use for a sink driver will have a voltage drop across it, just as the source driver does.

I asked you about this. Why does it matter? If the voltage drop across the sink driver is too great does that mean my led won't light at all? So does this mean that 5v may not be enough for both these darlington arrays and the LED? What do I do? Get a 6v regulator? A 9v regulator probably isn't an option with a 9v supply...

I wanted to figure out how much current I could put through my leds and for how long, and that power dissipation listed in the datasheet seemed to be the key to this, so I did some calculations.

The power dissipation listed is 62.5mW. And I had the equation P = I*E to work with. But what voltage should I use?

I first tried 5v, but I got 12.5 mA out of the equation. This didn't make sense, since I knew the leds could handle 25 mW.

On a hunch, I then tried the typical forward voltage of 2.2v. This gave me 28.4mA which was greater than the above rating.

Then I tried the max forward voltage. Aha! 25mA exactly.

Sadly, though I now see how they calculated the power dissipation, I'm back where I started. The power dissipation doesn't really tell me anything that the max current rating didn't, so I still don't know how much current I can safely put through the led on a 1/8th duty cycle with a 1.2ms pulse width.

I do have some bounds for how much current I can put through the leds due to some of my other calculations.

Here's what I've come up with:

The 2803 can sink 2500mA total.
But only 500mA per pin.
And only 312mA per pin if all are active, because 2500mA / 8 pins = 312mA.

But, the 2803 will be on the columns of my array, and only one column will be lit at a time.

So, I can sink 500mA per column. (Absolute max)

If I can only sink 500mA per column, and there are 10 leds per column, then that means I am limited to 50mA at most per LED, or rather, 50mA per row.

Sourcing the rows, I have the 2982. It can handle 120mA per output, continuously. It can therefore handle 50mA per row easily. So no problem there.

But going back to that 50mA per led... That 500mA per pin on the 2803 is an absolute maximum, and the data sheet for it (http://focus.ti.com/lit/ds/symlink/uln2803a.pdf) doesn't list a reccomended value. So I'm kinda stuck. Though at least I know I shouldn't exceed 50mA per led.

Well, I've decided to scale back my design a bit. :cry:

I started working on a schematic in Fritzing:
http://raccoonrocket.com/gb/egbschematic.png

But after getting it around 75% complete, I noticed a serious problem. Focusing on all the little details, I'd missed a big one and miscalculated the number of pins I need. In addition to the 18 I needed for the array, I also needed one for a piezo speaker, one for a switch, and one for a potentiometer. I'd forgotten the potentiometer. In addition, I'd decided the other day that I wanted to be able to pulse the brightness of two additional leds which would be on the ends of some antennas. That brings the total number of pins I'd need up to 22.

I'm aware that I could use a shift register to reduce the number of pins I need by 7 or so, but at this point it's just getting too complicated. Three chips to gain an extra two rows was already getting ridiculous, four is simply too much.

So, rather than have four 20 led bargraphs, broken up into eight sets of 10 leds, I am instead going to leave some of the leds unused and have four 16 led bargraphs, broken up into eight sets of 8 leds.

Through each of the 8 leds in a column, I will put 20mA, and I will run them on a 1/8th durty cycle with a 70hz or greater display update. I have already tested this out with a single bargraph being blanked 1/8th of the time, and while the display is not blindingly bright it is quite acceptable.

This will leave me with only 40mA to go before I exceed the 200mA limit, but I think I can power a piezo and a couple more leds off that, especially since the other leds should be really bright. There's a couple more leds on my illuminated pushbuttons which need to be lit as well, but thise I can run directly from the 9v battery and so won't contribute to the 200mA limit.

On the upside, limiting my current to 200mA means I'll more than double the run time I'd calculated that I'd get from two 9v batteries, so I should now get over 6 hours out of them. And it may in fact be more like 12, considering that less than half the display will be lit most of the time.

And my wiring will be much simpler as well, for now I will need only the 2803 to sink the current of each column. No more dual 2981's to source current. It'll still be a pain in the ass to solder ribbon cable to all those bargraphs to make my array, but it'll be slightly less of a pain in the ass now. Also, I no longer need to worry about whether the voltage drop between the two ICs the the leds will be too great with just 5v drving them.

Seems like a fair trade for 16 leds.