LM723-based buck SMPS?

[Boris Gjenero]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea. It seems like the general
consensus is that the 723 is obsolete even for linear applications.
The SMPS circuit seems to use bang-bang rather than PWM control, and I
guess there's no current limit and instead the current limit feature
is used to regulate base current for the switching transistor... so
it's not great but it should work.

I guess all I'd really need to add would be some stuff at the input to
protect the supply from transients and filter RFI.

 

[Robert Baer]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea. It seems like the general
consensus is that the 723 is obsolete even for linear applications.
The SMPS circuit seems to use bang-bang rather than PWM control, and I
guess there's no current limit and instead the current limit feature
is used to regulate base current for the switching transistor... so
it's not great but it should work.

I guess all I'd really need to add would be some stuff at the input to
protect the supply from transients and filter RFI.

As your reference shows, the LM723 is far from obsolete.
Also, Jameco carries it at a very decent price, if all you are
interested in is a small quantity.

 

[Kevin McMurtrie]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea. It seems like the general
consensus is that the 723 is obsolete even for linear applications.
The SMPS circuit seems to use bang-bang rather than PWM control, and I
guess there's no current limit and instead the current limit feature
is used to regulate base current for the switching transistor... so
it's not great but it should work.

I guess all I'd really need to add would be some stuff at the input to
protect the supply from transients and filter RFI.

It will work if you don't mind the sawtooth noise. An op-amp and a 5V
zener diode could do the same trick.

 

[legg]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea. It seems like the general
consensus is that the 723 is obsolete even for linear applications.

This has had it's day.

It's nice to know that you can do it, and it's nice not to be flumoxed
by the task of repairing a circuit in which the 723 is used in this
way.

Few of the original designers would select the same component, if
given the same design task today. If you have no specific reason to do
so, then you should also follow your own common sense in this
decision.

Even low-power circuits can benefit in parts count, real estate,
operating performance and final total cost, through the use of
currently available application specific devices.

Feel free to crunch some numbers on this.

RL

 

[Ken Smith]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea.

If you only have one, its a bad idea. If you have a spare then it may be
a good idea. Even though the LM723 circuit in question is not as good as
more modern parts etc etc, if you are just making one and it will do the
job and you have the parts on hand, there really is no reason not to.

 

[John Popelish]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

I'm wondering if this is a bad idea. It seems like the general
consensus is that the 723 is obsolete even for linear applications.
The SMPS circuit seems to use bang-bang rather than PWM control, and I
guess there's no current limit and instead the current limit feature
is used to regulate base current for the switching transistor... so
it's not great but it should work.

I guess all I'd really need to add would be some stuff at the input to
protect the supply from transients and filter RFI.

I think you have an understanding of this circuit's limitations, so as
long as you are not expecting more from it than it delivers, it should
be fine. The lack of an effective output current limit is its biggest
weakness.

 

[James Meyer]

I'm considering using an LM723
http://www.national.com/pf/LM/LM723.html to build an SMPS to convert
automotive "12V" to regulated 5 V, mainly for 5 V logic. I'm
considering using the LM723 because I have one around and the data
sheet shows how it can be used as a step-down SMPS (fig. 9, page 9).

For an automotive application, any sort of simple voltage regulator is a
bad idea.

I say that because an automotive application is the "worst case"
definition of the term "ground loop". Making a power supply out of a simple
regulator, switching or linear, means that the common ground of your logic
circuit will be the same as where ever you connect the ground lead of the input
to the regulator.

That means that if you make any other connection to your logic circuit,
like perhaps to bring in a voltage from some other part of the car, you will
have created a ground loop. Cars are notorious for having transient voltages
running all over their electrical system. That includes the ability to produce
healthy voltage differences between two different points that are normally
referred to as "Ground".

My suggestion is to find a DC to DC converter, 12 volts in to 5 volts
out, that has an isolated output. An isolated output means that if you use an
ohmmeter between any of the output terminals, + or -, that you will read
infinity ohms back to any of the input terminals.

That way, you'll get good efficiency, voltage regulation, and eliminate
ground loops.

Jim

 

[Ken Smith]

I say that because an automotive application is the "worst case"
definition of the term "ground loop".

I'll see your automotive ground loop and raise you an aircraft ground
loop.

In a car the ground loops have most of the large currents under the hood.
In an aircraft they flow all over the place: out to the starters on the
wings back to some electronics in the tail and etc.

The biggest worry in the car is withstanding the 60V spikes on the so
called +12V.

 

[Boris Gjenero]

It will work if you don't mind the sawtooth noise. An op-amp and a 5V
zener diode could do the same trick.

Yeah, the sawtooth noise seems to be the worst part. Regulation of
the average value is good but the ripple I get is unacceptable. It can
be reduced but I'll have to see how far without making the circuit too
inefficient or unreliable. It clearly has to be there for the circuit
to oscillate.

 

[Joerg]

Hi Boris,

You could but the cost savings are easily eaten up by the larger size
inductor you are going to need. I'd take a modern switcher from National
or TI that runs at several 100KHz and can use a tiny inductor.

If you filter the input properly you can tackle the spikes on the 12V side.

Regards, Joerg