UA723: "Drift is a one-time thing (“zener walkout”) that stabilizes after time", Output voltage is drifting during test.....!!!!!

Hi Eleazer,

I have some questions to help us to provide the best guidance possible:

• What is your input voltage, output voltage, and load current?
• Is the reference drifting proportionally to the output voltage?
• Are you measuring the voltages with a DMM or looking at it with a scope?
• When you say drift, are you talking about the output changing vs temperature or vs time?
  • Can you provide scope shots which show the phenomenon (include inputs and output on the same plot if possible)
  • How much drift are you seeing?
• Are you seeing this drift for every device or just some devices?

This is a very old device and therefore it has a pin out which we do not have a pin to pin replacement for, so swapping it for a modern device won't be an option. 

Hello Kyle,

What is your input voltage, output voltage, and load current?

The inout voltage is 18.60VDC, the output voltage is 5.XXXVDC and the output load current 12A (but I'm applyin on my experiment 3.25A)

Is the reference drifting proportionally to the output voltage?

Well, the first read on the output was 5.088V (first sample) and the Vref was 7.18V...then after 12 minutes:

Output voltage = 5.196V   and Vref =  7.27V

Are you measuring the voltages with a DMM or looking at it with a scope?

With DMM, is very similar to the method that use Our ATE.

When you say drift, are you talking about the output changing vs temperature or vs time?

Output changing Vs time, check my experiment and data below (it was with a second I.C sample)

            Output Voltage                                                     Time                                                      Readings with DMM Fluke 8845A

Are you seeing this drift for every device or just some devices?

All components have this behavioral (100%) and this is a nightmare for Our test operator (some power supplies have from 1 output to 5 outputs)

This is a very old device and therefore it has a pin out which we do not have a pin to pin replacement for, so swapping it for a modern device won't be an option.

Yes, We know that but unfortunately or fortunately We still manufacturing in one of our production lines:  linear products that are old design. We can't redesign the boards (isn't cheap). The LM723CN is obsolete now on the distributors sites, the only active product is your UA723 I.C and We need to get some fix for this issue. This is my work email: eleazar.galvez@slpower.com I hope You or some of Your technical engineers can help Us.

Thanks in advance....

In fact the Vout was increased by 55mV, from the first meausure and after 1 minute You can see below:

        Output Voltage                                                     Time                                                      Readings with DMM Fluke 8845A

   My experiments confirm the Zener Walkout (one time) issue, all the I.C's are more stable for regulation after the first test (my test bench is around 10 minutes)

materials department found 4k pieces from the LM723C (Texas Instruments I.C, plastic dip or case) but looks like that was the last pieces on the market. Again in the near future We're going to have the same issue to found more I.C's. The only solution that I found until now is test the I.C's in a test fixture before to use on production floor or wait around 1 or 1.5 minutes to warm the I.C and after that continue with the test. I tried some fix with a zener on pin 6 across the resistor but doesn't work. And again the UA723 has 100% of this problem not is an isolate issue......!!!!!!!!

This is my update.....

O.K, Kyle....

    If after many experiments the only option is to test the I.C's, my next question is:

Is necessary to test functionally 100%..?? I mean working the I.C direct on the power supply or Is there another way..? maybe applying only Vcc or direct voltage on the pin 6...?  because if I need to test the I.C direct on the power supply with load, etc....my production output are going to be low, but if there is a test fixture where I can put 20 .... or 50 UA723 "only" applying Vcc or DC voltage on pin 6 my output are going to be more higher....!!!!

Could you please check these options with your technical support engineers or group..??

Thanks in advance......

Hi Eleazar,

Yes doing it one by one will be very time consuming. I would recommend creating a test fixture which has many sockets that you can load many parts into and power them up at the same time.  The goal is to power up the devices while they are at high temperature for at least a few minutes to cause the trapped charge in the Zener to move, you can do that in  a couple different ways.

1. Put the board with the devices into an oven at 125C and power on the devices with minimal load 
   1. Positives: This makes it easier to find a usable input power supply since you don't have a lot of current being drawn from the power supply3
   2. Negatives: Requires an oven, ovens can take a while to get to temp so the more units on the board the more efficient this approach will be. 
2. Increase the power dissipation to self heat the silicon
   1. Positives: Don't need an oven and the silicon will get to temp very quickly 
   2. Negatives: Less exact way of heating the die, you should going in and out of thermal shutdown (with the output turning on and off), and you need to make sure you have a supply or multiple supplies (with the board laid out for multiple supplies) that can handle the potentially large amount of current

Below is an example of a board that is made to go into an oven with ~80 units 

O.K, thanks Kyle....!!!