uA723 voltage regulator viable in 21st century?

[David Hess]

The discussions about 723 obsolescence have the tendency to repeat themselves. Fact is that
there is no equivalent or better part for the range of specs (40V, 150mA, ...). There *are* better parts for less voltage and current, but thats like apples and oranges. The reason why it still sold after 50 years is that there is *no* direct replacement, obviously. 

There have been improved functional replacements like the Motorola MC1466L which includes a second differential pair for current regulation, but they never became popular enough to replace the 723 or even continue production.

[Wolfgang]

The discussions about 723 obsolescence have the tendency to repeat themselves. Fact is that
there is no equivalent or better part for the range of specs (40V, 150mA, ...). There *are* better parts for less voltage and current, but thats like apples and oranges. The reason why it still sold after 50 years is that there is *no* direct replacement, obviously. 

There have been improved functional replacements like the Motorola MC1466L which includes a second differential pair for current regulation, but they never became popular enough to replace the 723 or even continue production.

MC1466 has spec advantages over LM723, but it is less flexible because it only works in a floating architecture. Moreover, its not made anymore since many years, along with SG3532 and L146 and other competitors.

I am prepared to donate a cask of the finest bavarian beer for anyone who points to a *real* improved LM723 equivalent (no crippled low voltage, current or power versions, please) that is in volume production *today*. 

[DC1MC]

...snip...
I am prepared to donate a cask of the finest bavarian beer for anyone who points to a *real* improved LM723 equivalent (no crippled low voltage, current or power versions, please) that is in volume production *today*. 

I'm entering the contest with TPS7B7701-Q1 https://www.ti.com/product/TPS7B7701-Q1 

Cheers,
DC1MC

[Wolfgang]

Only 20V maximum output voltage ? Another try, please. 

[bd139]

LT3081

[KE5FX]

I'd say the LT3042/LT3045 would be good candidates.  They are much quieter than the LT3080 series, and requiring >15V output seems like an unnecessary constraint for 95% of use cases.

If you need 40V, dunno what to suggest other than a discrete component solution.

[Wolfgang]

LT3081

Much more noise, no long term drift spec, ... This part is for heavy lifting, rather than precision. 

[Wolfgang]

I'd say the LT3042/LT3045 would be good candidates.  They are much quieter than the LT3080 series, and requiring >15V output seems like an unnecessary constraint for 95% of use cases.

If you need 40V, dunno what to suggest other than a discrete component solution.

I agree that there good low noise regulators for 15 to 20V, some better than 723. But again, the comparison is not fair because good high voltage analog is much harder to make than low voltage only. So - still waiting for the 723 replacement part.

[Wolfgang]

I'd say the LT3042/LT3045 would be good candidates.  They are much quieter than the LT3080 series, and requiring >15V output seems like an unnecessary constraint for 95% of use cases.

If you need 40V, dunno what to suggest other than a discrete component solution.

"Real" precision stuff uses precision references (LM399,LTZ1000), precision ultra low noise op amps and discrete power components. Its probably too hard to integrate power transistors *and* precision circuitry in a single chip. Not seen any of those, they always try to keep heat generation and precision regulating circuits apart. 

[KE5FX]

How about an LT3042 or LT3045 with an external pass transistor?  You would probably have needed that with the 723 anyway.  It might be feasible to use a Sziklai pair with a 2:1 resistive divider to return the emitter (collector) voltage to the OUTS pin.

[RoGeorge]

555! I dont see what its good for except flashing LEDs and making crap monostables.

What?!  No.  555 is for making AM radio. 
http://tubetime.us/index.php/2011/02/23/555-contest-entry/

[rsjsouza]

Things that people do for a case of beer...

I would throw a few that might be a replacement with very near characteristics, such as the TPS7A39, TPS7A47, TPS7B4254, etc. 

A parametric table at:
https://www.ti.com/power-management/linear-regulators-ldo/products.html#p238max=30.01;200

[Wolfgang]

How about an LT3042 or LT3045 with an external pass transistor?  You would probably have needed that with the 723 anyway.  It might be feasible to use a Sziklai pair with a 2:1 resistive divider to return the emitter (collector) voltage to the OUTS pin.

Possible, but again an unfair comparison due to only half the LM723 voltage range.

[Wolfgang]

Things that people do for a case of beer...

I would throw a few that might be a replacement with very near characteristics, such as the TPS7A39, TPS7A47, TPS7B4254, etc. 

A parametric table at:
https://www.ti.com/power-management/linear-regulators-ldo/products.html#p238max=30.01;200

... getting better, but the parts have either
- not enough voltage
- too much noise
- not very well defined current limiting

but I will have a closer look.

[profdc9]

As an alternative to uA723, which is not common anymore, you can use an op-amp based power supply.  I made one based on a two op-amp design I found which has been floating around.

https://github.com/profdc9/LinearPS

A linear regulator is basically a slow unipolar amplifier with feedback, and so I added external inputs so that the voltage and current limits could be regulated by an external analog input (for example using an arduino output PWM).

[David Hess]

Discrete designs using operational amplifiers work but are not very practical now compared to integrated regulators, and operational amplifiers have a performance disadvantage in dual loop designs because of integrator windup.

The reason I mentioned the MC1466L as an improvement is that it has the same operational transconductance amplifier as the 723, but has two of them with parallel outputs so switching between voltage and current limiting is seamless.  There are modern switching regulator controllers with the same arrangement.

The same thing might be accomplished with an externally compensated operational amplifier like the 301A or 308 but that is hardly a modern solution.  I have seen a couple of power supply designs now which used two 723s with the error amplifiers paralleled to do it.  A general purpose single/dual/quad precision OTA as a universal building block would be ideal but the closest thing is the LM13700.  Tektronix used a lot of dual transistors in their regulators to do this.

[Wolfgang]

As an alternative to uA723, which is not common anymore, you can use an op-amp based power supply.  I made one based on a two op-amp design I found which has been floating around.

https://github.com/profdc9/LinearPS

A linear regulator is basically a slow unipolar amplifier with feedback, and so I added external inputs so that the voltage and current limits could be regulated by an external analog input (for example using an arduino output PWM).

I would not say your approach is uncommon. Most lab supplies with adjustable voltage and current work this way, with some op amps and a reference. The weakness of the LM723 is the low precision, temperature dependent current limiting. It can be used to avoid desasters and shorts, but it is not elegantly adjustable and always needs a diode voltage drop to work.

[iMo]

@Wolfgang: I can see a nice picture on your 723 site - "Proper design practices.." - with a 7815 pre-regulator for the 723 itself. Did you make any measurements how the output ripple improved?

[Wolfgang]

@imo:
I dont remember it but the gain in ripple rejection is considerable. My guess is around +30dB.
When you look at the 723 internals the error amp is a single long tailed pair, so this can profit
from a quieter internal supply. 723 collector voltage can be supplied by a less filtered supply, no problem.

[DimitriP]

In short, the viability of the uA723 in the 21st century depends on having hoarded as many as you think you might need.
For some reason there seems to be no "replacement" with "ultra low noise" specs and the voltage range of the uA723;   Progress!