lm-723 min voltage

[p.larner]

Is it poss to make a lm-723 based psu output form 1v to say 12.or does the 2.5v internal zenner limit it to 2.5v up?.

[PCB.Wiz]

Is it poss to make a lm-723 based psu output form 1v to say 12.or does the 2.5v internal zenner limit it to 2.5v up?.

The zener is not 2.5V ?
Google quickly finds :
"The input voltage ranges from 9.5 to 40V and it can regulate voltage from 2V to 37V "
"If you need to regulate lower voltages, the LM723 chip itself needs a separate power input at 10V minimum.
Minimum regulated output voltage is ca. 2.5V (minimum allowed voltage for the error amp input). With less accuracy, you may go below that, but not by much."

So the error amplifier common mode range is > 2.5V, however if you inject a precise current from another higher reference voltage, you can keep the common limit while the output goes to 0V.

[Terry Bites]

Yep https://320volt.com/en/0-50volt-0-5amper-laboratuar-tip-guc-kaynagi-ua723/
Google huh. Who would have thought.

[coromonadalix]

do take note this schematic  will not goe down to zero volt output, 723 by default is around 1.2volt

[Zero999]

1.2V is also marginal.

The reason for the minimum output voltage of 2V for the LM723, has nothing to do with the internal voltage reference. It's because the common mode range of error amplifier only extends down to 2V above the negative supply. In other words, the chip will only work properly, when pins 3 & 4 are at least 2V above v- (pin number 7 of the DIP-14 package).

A negative, as well as 0V and positive supply rail is required to reliably get output voltages below 2V.


https://www.electroniq.net/power-supply/lm723-0-30v-adjustable-power-supply.html

[PCB.Wiz]

A negative, as well as 0V and positive supply rail is required to reliably get output voltages below 2V.

Not quite.

The circuit given above has a flaw in the negative rail - using a zener degrades the PSRR and tempco as VREF+Zener drive the divider chain.
If you must go to the trouble of a negative rail a better/proper  negative regulator should be used.
A 2.5V shunt regulator may be 'good enough'.

Note that a negative rail is not essential to keep the Amplifier common mode above 2V. See this for a 4mV to 25V example, from a 0-5V control voltage.



There is good resource here on the venerable LM723/uA723, which does not have much life left.
https://electronicprojectsforfun.wordpress.com/power-supplies/a-collection-of-proper-design-practices-using-the-lm723-ic-regulator/
The poor 5% reference tolerance consigns it to trimmed/variable designs.

TI suggest the TPS7B7701 as an appx alternative. Better current sense/control, 40V, 300mA, but not a drop in replacement.
The Infineon TLF4277 family is similar.

[p.larner]

shame the schematics dont have pcb artwork to match.

[Zero999]

A negative, as well as 0V and positive supply rail is required to reliably get output voltages below 2V.

Not quite.

The circuit given above has a flaw in the negative rail - using a zener degrades the PSRR and tempco as VREF+Zener drive the divider chain.
If you must go to the trouble of a negative rail a better/proper  negative regulator should be used.
A 2.5V shunt regulator may be 'good enough'.

Note that a negative rail is not essential to keep the Amplifier common mode above 2V. See this for a 4mV to 25V example, from a 0-5V control voltage.
(Attachment Link)


There is good resource here on the venerable LM723/uA723, which does not have much life left.
https://electronicprojectsforfun.wordpress.com/power-supplies/a-collection-of-proper-design-practices-using-the-lm723-ic-regulator/
The poor 5% reference tolerance consigns it to trimmed/variable designs.

TI suggest the TPS7B7701 as an appx alternative. Better current sense/control, 40V, 300mA, but not a drop in replacement.
The Infineon TLF4277 family is similar.

It's not my schematic. I found it though a quick Google search. You're right. It's not very good.

I don't quite follow your LTSpice simulation. It would make it a little clearer and more accurate too, if you used a uA723 model.

Here's a Fairchild model I found awhile ago. I made my own symbol, using the generic 10 pin package, so it doesn't need separate files.

I don't have time at the moment to play with it, but will do later, if there's still interest.

[PCB.Wiz]

I don't quite follow your LTSpice simulation. It would make it a little clearer and more accurate too, if you used a uA723 model.

It's focused on the resistors (R1 to R5) and common mode offsets, so I dropped in a quick opamp + NPN model.

Here's a Fairchild model I found awhile ago. I made my own symbol, using the generic 10 pin package, so it doesn't need separate files.

Thanks.
The 'real' model drops VO for 5V in, by 1.5%, partly due to a lower VREF (by -0.31%) and partly due to the real bias currents & lower gain.
Still well inside the 10.3% error band of the uA723 on VREF alone.

[Zero999]

I've had a bit of time.

I changed the resistor values so standard E24 values can be used. In real life R1 can be two 30k in series.

Use generic models, rather than specific ones. UniversalOpAmp2 is a basic op-amp model, which includes the effect of the supply rails. The diode just stops it from sinking current. There's no point in a BJT at this stage. The current sink can just be a BJT and resistor. I know it's not ideal, but the current isn't critical. It doesn't matter whether it's 100µA or 1mA.


Here's the circuit, with the real uA723 model.