Upgrading the standard: a better way to drive peripherals

Our world is becoming automated. We see a strong initiative for more automation in our everyday lives, from smarter homes (AC, lighting and white goods) to easier and better travel in automobiles.  This requires a lot of processors and logic devices!  But how is logic controlling all of those motors, LEDs and relays? Peripheral, motor and low-side drivers are integral parts in making this happen. You may already know about a very standard driver used in most applications, the Darlington transistor. But as we strive to build innovative, better solutions, I feel compelled to ask: how can we make the standard even better?

What does the standard driver look like?

The simplest, yet the most common peripheral driver today is the Darlington transistor array. This low-side driver enables logic devices to drive or control a device with a higher power demand (shown in Figure 1):

Figure 1: Darlington low-side driver

In current systems, designers use arrays consisting of multiple Darlington pairs for control of a full system. This type of system typically allows logic devices with TTL or 5V CMOS to drive devices with up to 50V and 500mA per channel. Whenever current demand is too high for a single channel to drive, paralleling the channels helps distribute the current load evenly (shown in Figure 2).

Figure 2: Darlington Array Driver

However, using this type of architecture has its own tradeoffs and constraints.  One of the biggest problems is the increased board size whenever most, if not all, channels of the peripheral driver are overloaded. This then requires the use of an additional driver to divide the current demand among these.  Another setback is the increased power dissipation this device adds to your system.  The voltage in the low-side of the output for this device is increased, due to the stacked NPN transistors, by about 0.7V.  The dissipated power by this system will now look like:


PD = (~0.7V + 2Ω*IO) * Io

How to make the standard better?

One solution for these tradeoffs is to use an NMOS transistor instead of the Darlington pair. This low-side driver architecture has reduced dissipated power and can support inputs for all GPIO levels, from 1.8V to 5V.

Figure 3: NMOS low-side driver

This configuration allows us to drive peripherals the same way as with a Darlington pair with significantly lower power dissipation:


PD = (2Ω*IO) * Io

TI’s new peripheral driver, the TPL7407L, is a seven-channel, NMOS low-side driver array that replicates this architecture. This device allows us to replace any standard seven-channel Darlington based driver, while keeping dissipated power lower than the standard solution. This device also has an increased current support, allowing higher current demands to be allotted to a single channel or fewer than the standard device.

Figure 4: 7CH NMOS Low-Side Driver

Peripheral driving is used heavily in high-voltage applications such as white goods, HVAC, automobiles and building automation.  If you have or are designing a system that uses a Darlington transistor array as a peripheral driver, improve your system without having to go through an extensive redesign with this device.  This simple change can take your design to a whole new level and make your system that much better!

Want to learn more?

-          Watch a video on how to lower your power dissipation using the TPL7407L

-          Get a sample of the TPL7407L

-          Work with an evaluation module for the TPL7407L

-          Talk directly with engineers on TI’s E2E technical forums

Stay tuned to #TISpinsMotors on our social channels or go to ti.com/motor to learn more about our complete motor portfolio.

  • Hi Watcher, the TPL7407L was released earlier this year, this device was designed to offer a lower power dissipation when compared to ULN2003A. TPL7407 also allows higher currents per channel with lower heat dissipation at higher duty cycles.

  • I don't see why you didn't replicate the ULN2803A instead. I've been in the business for over 30 years and I don't think I've ever seen a 7407 used in a design.

    To add to Gene Norris' comments; power dissipation is only improved if your switching frequency is low enough as the doubled switching time will significantly increase switching losses. Add that to the higher thermal resistance and it is not a simple "drop in" replacement.

  • ULN 2.5A Total Current - ~12mA = 2.48A (which is greater than 2A) allows over 300mA/ch on all 8 channels.

    You would have made a better argument with total power:

    Although the ULN has a better θJA = 62.66 versus the TPL @ 91.9. For room temperature and SOIC packages, and

    Tj=100C (In case we want to boil water :))

    Pd(max)=(100-25)/θJA for the

    TPL = 0.816 and ULN = 1.19

          = 117mW/ch      = 149mW/ch

    But for the TPL Vo will be about ~0.5 V so it could support 234 mA,

    the ULN Vo would be ~1.2 V so it could support 125 mA.on all eight channels (140mA on 7)

    The power generation is somewhat idealized, but you get the picture, the TPL generates

    less heat for the same current...As I mentioned before.

  • Thanks for your comment Gene.  The improvement of this device vs. our other relay drivers is mostly for paralleled output applications. All of our ULN and the TPL7407L devices have a 2A rated current to ground, this means that although ULN2803 has more channels it can actually drive a lower total current. If you were to use this device to drive a motor or a high current relay the power saving differences would be significant. Whenever driving high duty cycles, say 100%, the ULN can only do ~80mA maximum when using all 7 channels, TPL7407L can do ~200mA in the same conditions.

  • Compare Parts TPL7407L ULN2803A


    Drivers Per Package 7 8

    Switching Voltage (Max) (V) 40 50

    Peak Output Current (mA) 500 500

    Delay Time (Typ) (ns) 250 130

    Output Voltage (Max) (V) 40 50

    Input Compatibility CMOS


    Vol@Lowest Spec Current (Typ)


    200 900

    Iout/ch (Max) (mA) 600 500

    Iout_off (Typ) (uA) 0.01 50

    So fewer drivers, less switching voltage, slower, but the same Iout* quite an improvement.

    Yes the the chip will heat up less, but it has to because its smaller...

    *Yes the recommended Iout(max) and clamp(max) are both actually 500mA.