TI release uln2003lv
jmg
Posts: 15,173
Since there have been recent threads on LED drive, and Port Init, this chip probably derserves its own thread
I see TI has a new variant, in a very old lineage - as you can guess from the part number
http://www.ti.com/product/uln2003lv
This is very cheap, as it is essentially just 7 matched FETS of ~ 2-3 ohms, (so ~10x the drive of a Prop pin) and it also has 300K pull downs, good for that init time, and includes the usual clamp diodes.
It is however, only 8V rated - so could drive LED strings with care, and certainly can Multiplex drive LEDs like 7 segments, with less drop that the darlington model.
- TI rates the GND pin at a AbsMax of 1A,
Being matched, one mode they show, is a current mirror.
I see TI has a new variant, in a very old lineage - as you can guess from the part number
http://www.ti.com/product/uln2003lv
This is very cheap, as it is essentially just 7 matched FETS of ~ 2-3 ohms, (so ~10x the drive of a Prop pin) and it also has 300K pull downs, good for that init time, and includes the usual clamp diodes.
It is however, only 8V rated - so could drive LED strings with care, and certainly can Multiplex drive LEDs like 7 segments, with less drop that the darlington model.
- TI rates the GND pin at a AbsMax of 1A,
Being matched, one mode they show, is a current mirror.
Comments
It uses darlington transistors.
Duane J
That does not say that the new model is also Darlington. In fact, TI is completely coy about the output structure. Even when they show outputs that are paralleled, they don't make any claims about increased current capacity (as one might with MOSFETs), only that it's a wired NOR configuration. However, the saturation voltage vs. output current graphs suggest a resistive output consistent with a MOSFET structure.
-Phil
Looking at figure 6 it doesn't really look resistive as in many MOSFETs except in very low current types.
And, Vout at 50mA of 0.24V is pretty low but darlingtons can do this to at low currents.
However, Iin(on) is 25uA max which does suggest a MOSFET.
And, Vout at 50mA of 0.24V does look like a MOSFET.
And, they have a 300K pull down on the input which does suggest a MOSFET.
I guess I don't know. Like you TI is kind of cryptic. If it is a MOSFET why keep it a secret? MOSFETs are superior to darlingtons in most ways.
I guess I have to agree with you that its a MOSFET albeit a kind of weak one though.
There are many MOSFET chips with much better specs than this.
The basic advantage this family of parts have is the integral free wheeling diode.
Duane J
So, I'm not sure why anybody would pick this one...
Is the on state resistance of uln2003lv less than 2003a?
Also, I just noticed Figure 13, which suggests that paralleling two outputs can increase current capacity, which you would not do with Darlingtons -- at least not without balancing resistors.
-Phil
-Phil
Yes, significantly, especially the voltage drop under load.
As it is an array of MOSFETS, the Voltage Output limit drops to 8V, which could be an issue, but the ON voltage drop is much lower than a Darlington, and the Drive current is also much lower. ( ~ 12uA) - ideal for 3V CMOS.
So it will not replace all ULN2003, but I'd expect most low rail LED drivers to change to this.
Its the switching delay of 15ns that clinches it for me - Darlington version is 250ns. That the switch off time is larger for 5V (compared to 3.3V) is also a clue!
My own PropRGB board uses 3 ULN2003A, so I'll have to think about if it's worth switching...
Would you need balancing resistors here as the individual devices would be a pretty close match and the thermal effects would be similar for each transistor? It has been a long time, but I thought the runaway issue was related to the transistor running away would heat more than the other in parallel causing further current hogging until it failed then letting magic smoke out of the other one left holding the current bag.
The Vbe of a transistor falls as the die temperature increases. Thus when bipolar transistors are connected with bases and emitters commoned up there is a risk of this type of thermal runaway. It can be reduced by mounting the devices on the same heatsink so the temperature difference is reduced. It is even a problem within one transistor die if the power levels are high enough (operating in linear mode with high collector voltages - this mode of failure is called "secondary breakdown" and what happens is that hot-spots on the chip start to carry a greater share of the current (and thus heat up more and more)
In a darlington array there is no direct commoning of bases (check the schematics) so this kind of thermal runaway is less likely (commoning outputs does tie the bases together in some sense, but a more indirect sense - in effect there is some series resistance involved via the Vsat of one of the transistors of the pair).
http://www.mouser.com/Search/Refine.aspx?Keyword=ULN2003V12
The above is 8v,
but if you have 15v source and put 4 white LEDs in a string (3v drop each) would you still not be safe with under 8v at sink gate?
Though Com pin is also rated 8v, maybe it can not handle reverse voltage above 8v? (leave it unconnected?)
Just use the newer ULN2003V12 instead.
It is cheaper, and has the same peak currents, but a slightly higher Rds drop at that levels.
(0.6v vs 0.4v) MAX levels.
http://www.ti.com/lit/ds/symlink/drv777.pdf
TPL7407L 40V
http://www.ti.com/lit/ds/symlink/tpl7407l.pdf