L293D driver and SN754410 substitution

[jedynakiewicz]

I have been using L293D driver ICs to operate small DC motors. I need to increase the current with some new motors and I intend substituting SN754410 chips for the L293D chips. I understand that this is a direct one-for-one substitute but with a higher current rating. The L293D drivers have built-in back-EMF protection diodes and the structural schematic in the datasheet (page 2) (http://www.rapidonline.com/pdf/82-0203.pdf) for the SN754410 appears to show that these are present in this chip also. The application circuit diagram shows external protection diodes however (datasheet page 6). Could the experts enlighten me as to what I actually need to do to power 12v 800mA DC motors (brushed) with these chips? Do I need the external protection diodes? If I do, what diode is recommended?

 

[JimPerry]

http://www.picaxeforum.co.uk/showthread.php?22629-sn754410ne-h-brige-problem shows some potential problems

 

[Technical]

Don't use them if you can avoid it. Lots of PICAXE users have had issues with these chips, which were nearly always resolved by going back to the L293D.
They definately do not have a good back emf internal diode as the L293Ds do. A common 1N4001 would do as an external.

 

[jedynakiewicz]

Wow! It is not often that one gets such a definitive approach as that... Thank you Technical and Jim. I shall revert to using L293Ds with the outputs ganged up in parallel in order to get the current rating that I need; I am glad that I asked.

 

[inglewoodpete]

Perhaps the L298 would do you better if you have to gang up outputs. Up to 2 amps (or 4 amps with outputs paralleled).

 

[jedynakiewicz]

Perhaps the L298 would do you better if you have to gang up outputs. Up to 2 amps (or 4 amps with outputs paralleled).

Thank you, I had considered that option; I have a tube of L298s but I generally work with 0.1" strip-board and the pinouts on the L298 are a bit of a pig to work with. The layout of the L293D is much easier and indeed with strip-board the pins are so arranged that ganging up the bridges is only a question of not cutting the copper across between the relevent pins.

 

[westaust55]

Many websites and forums advise to use external flyback dioses with the Sn754110.

Here is the details/recommendations from just one of the many webpages:

Your worries about the internal clamping diodes on this component are definitely well-placed. On page 2 of the datasheet (http://focus.ti.com/lit/ds/symlink/sn754410.pdf), it is clear that internal diodes do exist on the outputs of this IC. For your own reference, such diodes placed across an inductive load are called flyback diodes, meant to protect the circuitry from voltage spikes. So at first glance it appears that you would not need external flyback diodes with this IC, unlike some other H-bridge ICs.

However, based on the application circuit on page 6 of the datasheet, it appears that external diodes are indeed needed. Generally, good idea for most ICs is to Google them and look for application websites. In this case, some of the sites I found indicated that the internal diodes on this component are only used for electrostatic discharge, even though they may work as flyback diodes. Here's one example site: http://www.ladyada.net/make/mshield/use.html. The site states:

"After careful reading of the datasheet and discussion with TI tech support and power engineers it appears that the output diodes were designed for ESD protection only and that using them as [flyback]-protection is a hack and not guaranteed for performance."

same as my own thoughts/comments in another post relating to SN754110 chips only a few days ago:
http://www.picaxeforum.co.uk/showthread.php?22629-sn754410ne-h-brige-problem#post222934

 

[jedynakiewicz]

Westy, thank you. That very much clarifies what I had concluded; that the external inductive transient protection is essential; in principle, the SN754410 is a direct replacement for the L293 which has no inbuilt external inductive transient protection diodes but not the L293D which has. In fact, the 1amp rating also matches the L293 specification.

What surprises me a little is that @Technical recommended 1N4001 diodes (post #3), which are standard recovery diodes, for this role; I would have thought a fast recovery diode was needed in this application. I rather thought to use UF4001 diodes instead which is why I asked the question. Opinions?

 

[inglewoodpete]

Thank you, I had considered that option; I have a tube of L298s but I generally work with 0.1" strip-board and the pinouts on the L298 are a bit of a pig to work with. The layout of the L293D is much easier and indeed with strip-board the pins are so arranged that ganging up the bridges is only a question of not cutting the copper across between the relevent pins.

Sure the pinout for the L298 is not as convenient as the L293 but it works easily with stripboard. The L298 legs are springy enough to fit into the 0.1" grid. I have not had any trouble with them to date.

 

[SAborn]

For about $7.00 on ebay you can get the L298 on a premade circuit board with screw terminals for motor connections and header pins for data pins, for that cost i doubt if you could buy the chip and build a board as cheap.

If you are making a board and can do double sided boards one trick i used with the L298 is to edge mount it, so one row of pins go to the bottom layer and the other row goes to the top layer.
All that needs to be done is tracks on the edge like Vero board on both sides, then slide the chip on and solder, no holes needed to mount the chip this way.

 

[jedynakiewicz]

If you are making a board and can do double sided boards one trick i used with the L298 is to edge mount it, so one row of pins go to the bottom layer and the other row goes to the top layer.
All that needs to be done is tracks on the edge like Vero board on both sides, then slide the chip on and solder, no holes needed to mount the chip this way.

That is a most ingenious solution and I can think of several possible applications with similarly-pinned chips where that approach would be most useful, thank you.

I have photographed the method that I use with the L298; the pins can be bent carefully using two pairs of fine needle-nosed pliers so that they conform to the 0.1" perfboard as per @inglewoodpete (I prefer to bend them rather than just spring them); the L293 pinouts form a more logical (simpler) layout for the circuit design though which is where my comment was really coming from.



Any thoughts, though, on the need for fast recovery as opposed to standard recovery diodes for inductive transient protection?

 

[hippy]

Any thoughts, though, on the need for fast recovery as opposed to standard recovery diodes for inductive transient protection?

Standard diodes are found in many circuits so it would seem these are generally fine but specific cases may require specific solutions.

Without knowing the full characteristics of the motor being driven it's probably impossible to say or even to model. The best one can do is build a prototype and measure it, adjust as appropriate.