Author Topic: LM723 die pictures  (Read 26214 times)

0 Members and 1 Guest are viewing this topic.

Offline TOTALCHIPS

  • Newbie
  • Posts: 5
  • Country: us
Re: LM723 die pictures
« Reply #100 on: June 21, 2021, 12:20:10 pm »
Hi Noopy, Yes, you can use the photos I posted on your website.

Best regards.
 
The following users thanked this post: Noopy

Offline chicken

  • Frequent Contributor
  • **
  • Posts: 257
  • Country: us
  • Rusty Coder
Re: LM723 die pictures
« Reply #101 on: January 01, 2022, 10:24:24 pm »
Zeptobars today posted a beautiful die shot of a Motorola uA723PC:
https://zeptobars.com/en/read/Motorola-uA723PC-uA723-precision-voltage-regulator

 

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: LM723 die pictures
« Reply #102 on: January 20, 2022, 12:13:25 pm »
Zeptobars today posted a beautiful die shot of a Motorola uA723PC:
https://zeptobars.com/en/read/Motorola-uA723PC-uA723-precision-voltage-regulator

Day 12775. They still suspect nothing.
 :-DD



Okay, somebody tell me what the blue (?) structure in the feedback network is?
Either something that has lots of accurate resistance (I doubt it) or drops about 6V (not sure) and then the chip is a drop-in replacement for µA723.
Or something with low resistance or barely any voltage drop, then the chip has 1.8~2V reference output. What IC could that be?
Or something in between?
 :-//
« Last Edit: January 20, 2022, 08:05:56 pm by magic »
 
The following users thanked this post: SeanB, Noopy

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: LM723 die pictures
« Reply #103 on: January 20, 2022, 08:21:10 pm »
Well, okay. It is definitely a Brokaw type bandgap reference, but it's not such blatant cheating as National did, simply scaling the bandgap reference to 7V :)

I am convinced at this point that the mystery element in the feedback divider is indeed a zener diode, by similarity to another zener diode used for biasing (marked in green; also one ordinary forward diode is shown for comparison).

So what we have here is:
lower red: an absolutely classic Brokaw bandgap cell - it couldn't be more obvious if they tried
orange: the cell's PNP current mirror load; the leftmost PNP is a buffer which drives the output stage - very similar to AD1403
upper red: the output stage - a weird bootstrapped cascode level shifter and NPN darlington
blue: the feedback divider: resistor, zener, input of the bandgap cell, two resistors to ground

Assuming that the bandgap reference is tuned "normally" for 1.25V output with minimum thermal drift, the resistors scale its voltage to ~1.9V. Then the zener must add ~5.2V and as it happens, such zeners have minimum thermal drift too. So it checks out - instead of a high voltage zener with positive TC and a transistor with negative TC we got a lower voltage zener with low TC and a bandgap reference with low TC. (The exact tuning need not be zero TC - perhaps they have slight opposite TCs which cancel out).

Why bother? :-//
« Last Edit: January 21, 2022, 09:58:18 am by magic »
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #104 on: January 20, 2022, 08:24:04 pm »

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: LM723 die pictures
« Reply #105 on: January 20, 2022, 08:32:52 pm »
Well, except for one thing :D

Why bother? :-//

I don't get it, but I can't explain this IC in any other way.
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #106 on: January 21, 2022, 09:38:55 am »
Well not everything is explainable and consistent.  ;D
 
The following users thanked this post: Kartika

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #107 on: April 09, 2022, 03:24:01 pm »


Again a 723, this one is built by Mikroelektronika Botevgrad.




There is some potting on the die like we have seen in other chips built by Mikroelektronika Botevgrad like the SM631 (https://www.richis-lab.de/prawez03.htm).




Hey, it´s exactly the same die as in the IL72723 built by Radioindustrie Zagreb (https://www.richis-lab.de/LM723_09.htm)! Either they shared the masks or complete wafers.

The numbers 2723 at the right edge of the die let me assume that the circuit was designed by Radioindustrie Zagreb.


https://www.richis-lab.de/LM723_10.htm

 :-/O
 
The following users thanked this post: SeanB

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #108 on: June 22, 2022, 04:26:24 am »


We had that counterfeit STMicroelectronics LM723CN: https://www.richis-lab.de/LM723_00.htm
Now let´s take a look into a genuine one!




The die shows the typical structures that can be found on most other 723 variants too. The arrangement is very similar, but it is a separate design.

An interesting point is that it seems like at least parts of this design found their way into the RGW (council for mutual economic assistance). You can find the "Double-Output-Transistor" in the Radio Industry Zagreb IL72723 (https://www.richis-lab.de/LM723_09.htm), in the Mikroelektronika Botevgrad 723P (https://www.richis-lab.de/LM723_10.htm) and in the Tesla MAA723 (https://www.richis-lab.de/LM723_04.htm).

STMicroelectronics worked together with some RGW semiconductor manufacturers at least with Mikroelektronika Botevgrad which you can see with the 7812: https://www.richis-lab.de/voltageregulator13.htm.




With this background it´s possible that the counterfeit STMicroelectronics LM723CN came from the RGW area. I have seen this structures at the right edge of the die but I can´t remember which manufacturer used them. Can someone help me out?




That are the structures we know from ST. 8015 seems to be a process name and L9232 seems to be the project name.






Thanks to TOTALCHIPS we have a second picture of a LM723CN.

There is no difference in the integrated circuit.




The etch markers are missing on this die. With the numbers 800A we can guess that they changed the process.
And we have L923. Perhaps that´s the first revision of the design and the upper 723 called L9232 is a second revision.

NL923A6  :-//


https://www.richis-lab.de/LM723_11.htm

 :-/O
 
The following users thanked this post: SeanB, RoGeorge

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
 
The following users thanked this post: Noopy

Offline RoGeorge

  • Super Contributor
  • ***
  • Posts: 6145
  • Country: ro
Re: LM723 die pictures
« Reply #110 on: June 22, 2022, 07:15:47 am »
Thanks for all the pics and comments!  :-+
 
The following users thanked this post: Noopy

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #111 on: June 22, 2022, 07:43:05 am »
The fake one is Texas Instruments.
https://www.eevblog.com/forum/projects/ua723-voltage-regulator-viable-in-21st-century/msg1932811/#msg1932811

Texas Instruments? Interesting...
I will check that. I don´t think I have ever seen a Ti-Chip with these test structures.
And there is no Ti logo. Well it´s not necessary to have a Ti logo but you see it quite often.


Thanks for all the pics and comments!  :-+

 :-+
I still have a lot to publish...  :)

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #112 on: August 16, 2022, 08:08:08 pm »




NTE still serves the aftermarket with 723 voltage regulators. These are sold under the name NTE923. On the back of the package it says that the NTE923 is the successor of the ECG923. The company ECG was taken over by NTE.






Nothing special on/in the package.




Here we see that NTE uses the bandgap version of the LM723 from National Semiconductor.
(LM723CH: https://www.richis-lab.de/LM723_07.htm // LM723J: https://www.richis-lab.de/LM723_02.htm)

The auxiliary structures are completely the same. Most likely NTE (or ECG) bought up the last batches from National Semiconductor. These could have been finished devices or uncut wafers.




Now let´s take a look at the schematic. In my view the bandgap LM723 does its job like this:

The power section (red) is easy to recognize. Q14 is the driver transistor for the power transistor Q15. D3 is the Z-diode, which in some package is not connected to a pin. Q16 can sink the base current of the driver transistor and thus makes it possible to implement current limiting.

Also easy to recognize is the differential amplifier (gray), which is usually used to compare the actual value of the controlled system with a setpoint. The two input transistors Q11 and Q12 work with the current sink Q13 for this purpose. Q12 sinks the base current of the driver transistor if required.

The current sources and sinks of the device (blue) are current mirrors that use transistor Q2 as a reference. This transistor generates a relatively constant current using the Z-diode D1.
The self-supply is interesting. Via Q7/Q9/Q10/Q25 the transistor Q2 itself sets the current through the Z-diode that is necessary for its operation. To ensure that the circuit starts up safely, there is a small current sink (cyan) in the lower left area.

The reference voltage of the bandgap LM723 is generated with a Brokaw bandgap reference. The core of this reference is formed with the transistors Q21/Q22 with their resistors R14/R15 and the current mirror Q19/Q20 (light green). The current mirror ensures that an equal current flows into both branches. As described with the TL7705 (https://www.richis-lab.de/TL7705.htm), a very temperature-stable voltage is established at the base of Q21 and Q22 if the current densities in the transistors and the resistor values are suitably designed.
The purpose of the diode D4 remains unclear.  :-// It should not become conductive during normal operation. A protection of the base-emitter path against negative voltages does not seem to be necessary due to the high resistance values.

If the desired operating point in the bandgap reference has not yet been reached, the path Q6/D5/Q4 becomes conductive and reduces the current flow through Q5 (orange). Less current flows into the current mirror and into the voltage divider (yellow) of the bandgap reference. As a result, the operating point shifts until the same current flows through the transistors Q21/Q22.

The yellow voltage divider not only controls the transistors Q21/Q22, it also ensures that the desired reference voltage of typically 7,15V is set at the top resistor. The purpose of the transistors Q23/Q24 (purple) remains unclear. It seems that the circuit adds an additional temperature drift. For a Brokaw bandgap reference such an auxiliary circuit should not be necessary.

Capacitors C1 and C2 (dark green) stabilize the bandgap reference. It is interesting to note that the Z-diode D2 in combination with the current source Q3 generates an auxiliary reference for the capacitors. As will be shown this is necessary because the capacitors are not very voltage resistant.




The individual elements can be found on the die as shown in the circuit diagram.




The different sizes of the transistors Q21/Q22 can be seen clearly. The ratio of the emitter areas is 10:1.

As shown in the schematic R15 can be varied to optimize the temperature drift of the bandgap reference. However, it is not just a single resistor that can be added or excluded. The layout would allow many small areas to be bridged with the metal layer. Towards the bottom end two zener fuses are integrated, which allow two resistors to be bridged after fabrication. A thin line can be seen in the left fuse while the right fuse still looks intact.




The capacitors are comb-shaped. They are emitter areas in areas with base doping. The special shape ensures higher capacitance because the emitter doping has a certain depth and the comb structure thus increases the surface area.
However, the base-emitter junction has a low breakdown voltage. For this reason, it was necessary to generate a higher auxiliary potential as reference potential for the capacitors.  :-+


https://www.richis-lab.de/LM723_12.htm

 :-/O
 
The following users thanked this post: SeanB, RoGeorge, mawyatt, EJE

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: LM723 die pictures
« Reply #113 on: August 17, 2022, 05:41:42 am »
Pretty much my own conclusion last year - a zener reference is only used for internal biasing. The pink stuff could be some curvature correction or whatnot.

Nice box :wtf:
TI apparently still offers National's version for a little premium, see this part and its datasheet. Dunno if they make new dice or sell remaining stock.
https://www.ti.com/product/LM723QML
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #114 on: August 17, 2022, 06:10:51 am »
Pretty much my own conclusion last year - a zener reference is only used for internal biasing. The pink stuff could be some curvature correction or whatnot.

That's how it is.  :-+

A zener to reduce the voltage at the capacitors...  ;D


Nice box :wtf:
TI apparently still offers National's version for a little premium, see this part and its datasheet. Dunno if they make new dice or sell remaining stock.
https://www.ti.com/product/LM723QML

I will have to take a closer look at these...  ;)

Offline magic

  • Super Contributor
  • ***
  • Posts: 6733
  • Country: pl
Re: LM723 die pictures
« Reply #115 on: August 17, 2022, 06:53:02 am »
I'm sure there is nothing interesting in there.

BTW, the link above was some military(?) qualified version or whatnot, here's the ordinary LM723. Still metal cans only.
https://www.ti.com/product/LM723

And TI's own version, in plastic packages.
https://www.ti.com/product/UA723
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #116 on: August 17, 2022, 06:56:30 am »
You never know what you will find inside...  ;)

Offline iMo

  • Super Contributor
  • ***
  • Posts: 4675
  • Country: nr
  • It's important to try new things..
Re: LM723 die pictures
« Reply #117 on: August 17, 2022, 08:03:28 am »
From an artistic point of view I like the picture above, a kind of "gothic architecture" style.. The newer designs are pure modernism like "art deco"  or "bauhaus" :)
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #118 on: August 17, 2022, 08:22:00 am »
I agree with you.  :-+ :D

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 3193
  • Country: us
Re: LM723 die pictures
« Reply #119 on: August 17, 2022, 01:01:02 pm »
From an artistic point of view I like the picture above, a kind of "gothic architecture" style.. The newer designs are pure modernism like "art deco"  or "bauhaus" :)

Agree, the image noise also adds to the artistic effect. This image could be printed and wall mounted for display!!

Well done and rendered by Noopy as usual :-+

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: Noopy

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #120 on: August 22, 2022, 11:03:42 am »
At first I thought the 923 is just a special naming NTE had chosen for its 723 but I have found old magazines with prices for a LM923 and on the die of the ST LM723 there are also the numbers 923.

Does anyone know what this 923 is? A 723 with a higher temperature rating?

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #121 on: October 13, 2022, 06:08:04 pm »


Harris Semiconductor produced a variant of the LM723 under the designation CA723. The C marks a slightly worse specified variant. These are approved for a temperature range from 0°C to +70°C (compared to -55°C to +125°C for the variant without C). Nevertheless, the accuracy and stability is slightly worse. The T stands for the metal housing.

Besides the CA723 from Harris Semiconductor, voltage regulators with the same designation can be found from Intersil and RCA. Harris Semiconductor controlled both Intersil and RCA for some time. The earliest references to the CA723 are found in 1978 and are attributed to RCA. Intersil advertises an LM723 as an alternative in 1979. It seems very likely that the CA723 is an RCA development that was adopted by Harris. With the integration of Intersil, the CA723 probably then displaced the Intersil variant of the LM723 as well.




The schematic shown in the Harris Semiconductor datasheet is very similar to the schematic of Fairchild's µA723 (https://www.richis-lab.de/LM723_06.htm).




The housing is connected to the negative supply.




The characters TA6563 at the top edge of the die are probably an internal project designation. The auxiliary structures in the lower right corner are typical for RCA components (CA555: https://www.richis-lab.de/555_11.htm, CA3161: https://www.richis-lab.de/logic22.htm).

The arrangement of the components corresponds to the arrangement in Fairchild's µA723. Some structures, such as the power transistors look exactly the same.




The circuit of the CA723CT from Harris Semiconductor also corresponds to the µA723 from Fairchild. Two errors were found in the Fairchild schematic. Here you can see the corrected Fairchild schematic. The supply of transistor Q11 is already shown correctly in the Harris schematic. However, the supply of transistor Q9 shows the same error as in the original Fairchild schematic. As shown here, it is supplied by the reference voltage, not by the positive supply voltage.




The Fairchild µA723 uses the base-emitter junction of a normal transistor as a Z-diode for generating the reference voltage (left). In the CA723CT (right) a special structure has been integrated for this purpose in which the anode has a much larger area.


https://www.richis-lab.de/LM723_13.htm

 :-/O
 
The following users thanked this post: SeanB, RoGeorge, floobydust, EJE

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #122 on: March 13, 2023, 08:10:51 pm »


Motorola produced a variant of the LM723 with the name MC1723. The MC1723C is specified for an operating temperature range between 0°C and 70°C. The MC1723 without index allows operation between -55°C and 125°C.






Two schematics can be found for the MC1723. The upper schematic is from the Motorola Linear Integrated Circuits Data Book from 1971. More recent datasheets show the lower schematic. Similar to the µA723 built by Fairchild (https://www.richis-lab.de/LM723_06.htm), the supply of the non-inverting branch of the differential amplifier has been relocated. Here, however, a mistake has crept in quite surely. The branch is not only connected to the reference voltage range, but also to the common base of the current sources. The latter makes no sense.






The design of the MC1723 is similar to the other LM723 variants, but there are some interesting differences. It is very different to the design zeptobars has shown.

In the middle of the die a resistor strip is connected to two metal surfaces. Obviously the structure was used to determine the resistor value after manufacturing.

The angles shown with the different masks and a sequence of three letters seem to be typical features of a Motorola design. The same patterns can also be found in the MC1455 (https://www.richis-lab.de/555_13.htm).




In many areas, the circuit corresponds to the circuit diagram. However, there are also some differences. In the right area there are surprisingly many elements that are not included into the circuit.




If you draw the schematic, you will see that the reference current for the biasing is generated quite differently in the MC1723 than in the µA723, for example.




Motorola has printed the metal layer of the MC1723 in the Linear Integrated Circuits Data Book. The structures do not match the present die at all. Either the design was heavily revised or the MC1723 released for the extended operating temperature range contains a completely different design, which was shown here.


https://www.richis-lab.de/LM723_14.htm

 :-/O
 
The following users thanked this post: SeanB, RoGeorge, ch_scr

Offline mediatechnology

  • Newbie
  • Posts: 1
  • Country: us
    • Pro Audio Design Forum
Re: LM723 die pictures
« Reply #123 on: June 24, 2023, 09:15:57 pm »
New here to this forum and just wanted to point out that the µA723 is going to be 55 years old in 2023.

Here are some pretty ones I found in my parts stash along with links to the September 1968 data sheet and a clean copy of the 1968 Application note I scanned.

µA723 Data sheet, September 1968: https://proaudiodesignforum.com/images/pdf/uA723_Data_Sheet_Fairchild_1968.pdf

µA723 Application Note, 1968: https://proaudiodesignforum.com/images/pdf/uA723_Application_Notes_Fairchild_Semiconductor_1968.pdf



Source: https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1327
 
The following users thanked this post: blackdog, SeanB, Wolfgang, RoGeorge, ch_scr, Noopy

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1707
  • Country: de
    • Richis-Lab
Re: LM723 die pictures
« Reply #124 on: August 13, 2023, 06:19:46 pm »


Taking a look into the DIL variant of the MAA723, the MAA723CN.
(December 1984)




The edge length is 1,3mm. The die is exactly the same as in the TO variants.







https://www.richis-lab.de/LM723_15.htm

 :-/O
 
The following users thanked this post: SeanB, RoGeorge, ch_scr


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf