Reviewing the LM723

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
Sure, it is ancient. You can stock a bunch of more modern voltage regulators or you can just stock one that does positive or negative, low current or high current with just adding a cheap power transistor. Nonetheless the LM723 has some great lessons on how voltage regulators work. Operating voltage goes up to about 37 V. Not a bad deal for about $0.50 US. Old, yes,, but still very educational and useful.
The LM723 has two basic sections. It has a voltage reference section and a regulator section. The reference section puts out about 7 Volts. Absolute Maximum Rating on the spec sheet says 15 mA, maximum to be drawn. From 1 mA to 15 mA this output probably won't waver even 1 mV. The voltage regulator section can tolerate about 150 mA being pulled from it. Adding an external transistor adds to this. Most any transistor will do.
For operation below 7 Volts a voltage divider is added to the output of the voltage reference. The Voltage regulator section compares this voltage to the output voltage. For operation above 7 Volts you put the voltage divider on the output and compare the 7 V reference against a tap at 7 Volts on the divider.
 

Hypatia's Protege

Joined Mar 1, 2015
3,228
FWIW I'm partial to the LM338 (and it's ilk) for it's all around superior characteristics...

I offer a single caveat in regards to such devices -- Although promoted as 'adjustable regulators' ('programmable regulators' would seem more apt) one must exercise extreme caution when applying same to adjustable PSUs --- Stray outside the (less than intuitive) boundaries of the 'safe area' and your PSU output will resemble that of a malfunctioning sawtooth generator!:eek::eek::eek:

That said - IMNSHO properly applied linear regulation is the only choice where ultra low noise performance is required:)

Best regards
HP
 

AnalogKid

Joined Aug 1, 2013
10,986
That said - IMNSHO properly applied linear regulation is the only choice where ultra low noise performance is required:)
For a normal-sized benchtop device, absolutely. But for an application-specific power system, not so much.

At one time there were four companies in the US that made MRI scanners, and my company built sub-systems for three of them. One customer was moving to an upgraded design, and had a big SWaP (size, weight, and power) problem. A rackmount chassis with two 6U card cages had two multi-output switchers and four large single-output linears (two of them quite large), in the theory that they needed linear supplies for their low-noise signal processing section. I re-spun their signals backplane to have two small linear regulators to take POL regulators (and their heat) off one of their boards, and sprinkled frequency-specific decoupling at all backplane slot connector power pins, increased the power levels of both switchers, and powered the signals backplane directly off of them. Results, it reduced the chassis height by 3U, reduced weight by 30 pounds, reduced system power by 15%. But about the noise - it was less than 3 mV p-p at 100 MHz bandwidth, 4x below their previous best design, and 8x below their internal design spec. They had to rent a high performance test platform to measure it.

Why? Because while their linear supplies measured very well, they did not have 60 MHz decoupling at their outputs. And even if they had, the feet of wire to the backplane were excellent loop antennae. The majority of their interfering noise was radiated and conducted from the 1200 W of digital banging in the other card cage.

IMELHO properly applied switching supplies are an excellent choice where ultra low noise performance is required.

BTW, IMESHO MNSHO is ELH (even less humble) than yours, so there.

ak
 
Last edited:

Hypatia's Protege

Joined Mar 1, 2015
3,228
And even if they had, the feet of wire to the backplane were excellent loop antennae. The majority of their interfering noise was radiated and conducted from the 1200 W of digital banging in the other card cage.
An interesting observation and very compelling point!:)


Re: noise benefits of linear regulation
But for an application-specific power system, not so much.
---Emphasis added---
IMELHO properly applied switching supplies are an excellent choice where ultra low noise performance is required.
Bearing the previous quotation in mind --- Granted!:):):)

Very best regards
HP
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
Schematic examples. The LM723 is shown broken down into its inner parts for educational purposes.
FWIW I'm partial to the LM338 (and it's ilk) for it's all around superior characteristics...

I offer a single caveat in regards to such devices -- Although promoted as 'adjustable regulators' ('programmable regulators' would seem more apt) one must exercise extreme caution when applying same to adjustable PSUs --- Stray outside the (less than intuitive) boundaries of the 'safe area' and your PSU output will resemble that of a malfunctioning sawtooth generator!:eek::eek::eek:

That said - IMNSHO properly applied linear regulation is the only choice where ultra low noise performance is required:)

Best regards
HP
All very wise words. No, I am not suggesting the LM723 for new commercial devices. It is a suitable solution for education or home tinkering.
 

#12

Joined Nov 30, 2010
18,224
FWIW I'm partial to the LM338
FWIW, I'm partial to the LM723 because that's what I cut my teeth on. Those years of experience allowed me to develop some versatility and I especially admire the ability of the chip to protect itself from excessive loads. When the company hired a, "real" engineer, his three pin regulator designs sounded like popcorn when subjected to the tests I used on my LM723 designs. Of course, that was AFTER I showed him how to find the limit on how many turns you can fit inside a torroid.:D

Some ten years later, I was denied a job designing an oven to temperature cycle new products because I was only a State Certified designer of heating and cooling machines. The, "real" engineer who got the job designed a machine that melted $3 million worth of products on its first run. I have the photograph.;)

I have virtually worshiped some great engineers during my career, but claiming you're a, "real" engineer doesn't impress me; your designs do. Our colleges are presently turning out thousands of, "real" engineers that couldn't design a clothes dryer without an Arduino in it, and they often struggle to do that because they don't know how to use a bjt as a switch or how to get a microcontroller chip to fire a triac!:mad:

/end rant
 

Hypatia's Protege

Joined Mar 1, 2015
3,228
Schematic examples. The LM723 is shown broken down into its inner parts for educational purposes.


All very wise words. No, I am not suggesting the LM723 for new commercial devices. It is a suitable solution for education or home tinkering.
Fair enough!:) --- I was merely reminiscingo_O

Best regards
HP:)
 

#12

Joined Nov 30, 2010
18,224
Perhaps you could take a 'snap' of it with a camera
Great idea! I have a PHD (press here, dummy) and it downloads with a USB cable!

Fail! I have the 1985 Paradyne Corporation Calendar wherein I won the February page with Mount Rushmore and the Cover Shot with a time lapse of the Snake River. I can't find the other calendar with a room full of melted modems for the cover shot.:(

Please forgive the flash hole. It's a photo of a photo inside a glass frame.
 

Attachments

Last edited:

#12

Joined Nov 30, 2010
18,224
You mean to say that they advertised their blunder?! -- And just when I'd abandoned all hope of corporate integrity!:confused::eek::D
Never fear. There was no integrity intended. The $3M fail was the last straw. They continued for another year by paying their employees in worthless, "stock", then parachuted out with all the pensions and profit sharing disappearing like a politicians promise.
 

Hypatia's Protege

Joined Mar 1, 2015
3,228
Never fear. There was no integrity intended. The $3M fail was the last straw. They continued for another year by paying their employees in worthless, "stock", then parachuted out with all the pensions and profit sharing disappearing like a politicians promise.
How did they manage to destroy $3 million worth of product prior to realizing they had a problem?:confused: I'm thinking Big oven or bad QC???:eek::eek::eek:

Best regards
HP
 

#12

Joined Nov 30, 2010
18,224
How did they manage to destroy $3 million worth of product prior to realizing they had a problem?:confused:
They hired a, "real" engineer! Real college degree = State of the Art Design, don't you know? Real engineer, big oven, and not a fail-safe in sight...not even a human watching the first full load, 24 hour run. Better to hire a college graduate than a blue collar person with 15 years experience and a State Certification to prove it. Real Engineers don't make mistakes like that, so there is no need to pay anybody to monitor the progress.

/sarc off
 

MrAl

Joined Jun 17, 2014
11,389
Hi,

The 732 went out of vogue a while ago because of its limitations and old method of voltage regulation.
Back then in the 1970's, it was probably impossible to find a voltage regulator chip WITH a band gap voltage reference. These days it's hard to find one WITHOUT a band gap reference.
Once the band gap references came in, the zener based references went out. That's because they are better over a wider temperature range.
Another drawback is the basic required voltage level. If you have a 7v zener then you need at least a 9v supply or so. If you dont have that, you cant use the 732.
One thing in its favor is the fairly simple current limit, but after all it does have a very wide variation over temperature because it uses the shunted base emitter diode scheme to limit current, and we all know the variation of base emitter diodes over temperature. Luckily, many applications dont need precision current limiting, not even Li-ion as long as there can be some reasonably safe max.
Also, it's only good for lower currents without added external devices.

One of the supplies i have is a switcher with a linear output. That's about the best you can get for low noise and clean output with high efficiency as well. But switchers also put out some EMI which can bother some designs.
 
Last edited:

#12

Joined Nov 30, 2010
18,224
it's only good for lower currents without added external devices.
Not a problem. I was blowing 35 amps per polarity through my linear designs in 1974, and the foldback circuit still worked!
It was about as fun and primitive as working with LM555 designs. Both chips have better versions available now, but the education is still applicable. (See my blog for a design that goes from zero volts to 1000 volts, based on the principles available in the LM723 datasheet.)
 
Top