Total current on ULN2803A

A

Andres Rodriguez

New Member
I am using a ULN2803 to drive eight LED strips. Each strip draws 200mA. According to the datasheet (attached) the maximum continuous collector current is 500mA. My question is if that maximum current is for the entire chip or for each individual Darlington pair.
 

Attachments

A

AllyCat

Senior Member
Hi,

Yes, 8 chains running at 200mA each looks just out of specification, but much will depend on the application, are you using PWM control and is the circuit enclosed (in a box), etc.? You could probably run 6 chains safely at 100% or all 8 at perhaps 60% duty cycle. In practice, in "free air", with a decent PCB copper area and perhaps a small heatsink glued on the chip, eight would probably work perfectly well at 100%.

I don't know what those FETs cost, but 30 Amps for a 200mA load seems rather overkill. IMHO eight cheap bipolars such as BC548 (or better BC337) each with 10mA base drive through a 470 ohm resistor from the PICaxe pin will do the job. With 10mA from each PICaxe pin a darlington configuration is hardly necessary and the "overswing" diodes shouldn't be needed for a LED load.

Cheers, Alan.
 
P

PaulRB

Senior Member
AllyCat said:
I don't know what those FETs cost, but 30 Amps for a 200mA load seems rather overkill.
Click to expand...
You're right Allan, they are overkill, but I've used them before to drive 12V LED strips, and I know they work. They're about 50p each, so 8 would cost more than the 2803, and a lot more than your BC548s!

Marmitas, trust Allan's advice on the transistors, but if you ever need to run longer strips and/or higher voltages, remember those FETs.
 
P

PaulRB

Senior Member
Not sure that exact ic would be suitable, Pete. Marmitas needs 200mA per output. Not sure what voltage is being used, but the problem with both the 2803 and the ic you suggested may be total dissipation.
 
A

Andres Rodriguez

New Member
PaulRB said:
Not sure that exact ic would be suitable, Pete. Marmitas needs 200mA per output. Not sure what voltage is being used, but the problem with both the 2803 and the ic you suggested may be total dissipation.
Click to expand...
Sorry, I did not include the voltage. It is 200mA and 12 Volts.
 
J

jedynakiewicz

Senior Member
You can gang up the inputs/outputs on the 2803; therefore you could use two 2803s connected to provide four outputs on each chip. I seem to remember someone suggesting on this forum the piggy-backing of two 2803s, one directly on top of the other and all pins connected, as another way of providing a higher output. Overall, I think the best solution is the string of BC548s as suggested by Alan.
 
D

Dippy

Moderator
Well, unless I've missed something (highly likely) I wouldn't recommend the BC548 by Alan.

I agree that discrete trans are a good idea.
Though a quick squiz at (Fairchild & ON Semi) BC548 on Farnell shows recommended IC max cont = 100mA.
I think I'd give them a miss.

The 337 are much more butch.
Always read data sheets before diving in.

Just one 'off-tangent' comment.
When going on about MOSFETs people seem to jump straight to big lumps.
There are tons of 'old faithful' small ones kicking about.
Granted SMT/SMD is taking over, but you can still get tiddlers like 2N7000, BS170/270 that would possibly
OK here - even though I'm sure we haven't got the the full details.
 
W

westaust55

Moderator
Dippy said:
Well, unless I've missed something (highly likely) I wouldn't recommend the BC548 by Alan.

When going on about MOSFETs people seem to jump straight to big lumps.
but you can still get tiddlers like 2N7000, BS170/270 that would possibly
OK here - even though I'm sure we haven't got the the full details.
Click to expand...

Nothing missed to my knowledge. BC547/8/9 and BC557/8/9 and all rated IC = 100 mA max.

BC337 and 2N2222 are rated IC = 800 mA max.

The 2N7000 (which I use for i2c level shifting and 1-Wire interface for adding EEPROM programming) are rated ID = 200 mA max continuous and 500 mA pulsed
 
A

AllyCat

Senior Member
Hi,

Yes, it appears that all BC548s are not created equal. Fairchild appear to be happy to rate theirs at 500mA ( "This device is designed for use as general purpose amplifiers and switches requiring collector currents to 300 mA." ), but TBH I'd rather believe ON-dig/Philips and the BC337 (or one of many other types) would be a far more appropriate choice. Certainly I wouldn't buy BC548s specifically for this application.

Marmitas, I doubt if 12 volts (versus 5 volts) makes any significant difference. With a 5 volt rail, bipolar (or logic level FETs) can have a significant advantage over Darlingtons which have ~750mV saturation voltage (i.e. ~15% "wasted" voltage, or 30% potential power), but there's little difference from 12 volts, particularly if the base current is coming via a 7805 regulator from 12 volts anyway.

The amount of heat (power dissipation) transferred by conduction via the IC pins and convection from the IC package should not be under-estimated (if you think about it, that's almost the only way for the heat to escape), so where power dissipation is the limiting factor (as is the case here) then "piggy-backing" ICs is unlikely to help much. I don't know the physical difference between the two devices in the specification linked in #10 by Rick100, but the higher rated device may have a "heat speader" (e.g. a small lump of copper) inside tha package to help heat to escape, so a small heat sink "flag" glued to the package could be all that's needed.

IMHO probably any of the above solutions will work well enough, it really depends what components you have easily or cheaply available and how hard you might later "push" the design (to higher current LEDs, PCB enclosed in a box, etc.).

Cheers, Alan.
 
D

Dippy

Moderator
Slight tangent....
It was interesting reading the link given by Alan.

If you check the dates you'll see quite a difference - maybe Alan got his from his old PC grimoire?

Check Fairchild's site and you'll find the Farnell one is the latest - albeit 2002.
http://www.fairchildsemi.com/ds/BC/BC548.pdf
And it confirms the 100mA figure.

I know not what happened between 1997 and 2002, but it is a lesson to check the latest data sheet.
Maybe Fairchild bought a new multimeter :)
Quite an eye-opener for all I suspect.
And lesson 2 - go to manuf's site where possible.
 
W

westaust55

Moderator
Another aspect to be mindful of is the maximum power dissipation.

To look at a few generic examples:
BC108 IC = 200 mA max and 300 mW max (old metal can type)
BC548 IC = 100 mA max and 625 mW max
BC337 IC = 800 mA max and 625 mW max
2N7000 ID = 200 mA max and 200 mW max with Tamb = 25 ºC


For the BC3548, to utilise the full 100 mA current rating and not exceed the 625 mW rating, the VCE must be less than 6.25 Volts which means working in the linear region is possible (though not the best in terms of heat dissipation/losses/volt drop).

For the BC337, to utilise the full 800 mA current rating and not exceed the 625 mW rating, the VCE must be less than 0.78 Volts which means working in the saturated region as a switch.

The 2N7000 has a typical Drain-Source ON resistance of 2.4 Ohms when VGS = 10V but for logic applicaitons with VGS = 5V that rises to 3.2 Ohms.
So typical ON resistance is not a problem but if your sample has an ON resistance of 5 Ohms then you are right on the power dissipation limit.

Thus as Dippy indicates, it is essential to consult the datasheets. One should also consider more than just the collector/drain current.
 
You must log in or register to reply here.
Top