• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 34 subscribers
  • Views 138 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AM26LS32AC: abnormal absorption on new supply of devices

Marinella Carboni
Prodigy 10 points
Part Number: AM26LS32AC
Other Parts Discussed in Thread: AM26C32,

Tool/software:

We have been producing boards with AM26LS32ACD for many years and during testing we check the current absorption on 5V. Normally we have an absorption of about 40mA per chip.
In the latest production, however, the absorption has become practically zero.
For this reason we have studied the issue in depth by creating two small wired boards where we have placed a previous chip (which we indicate with A) that consumes 40mA and a recently purchased one (which we indicate with B) that consumes 0mA.
In the boards created for the test we have connected only the +5V power supply on pins 16 and 4 while we connected the GND to pins 8 and 12. In this way the integrated circuit is powered and enabled (G signals activated).
Device A absorbs 40mA while device B absorbs almost zero mA.
Device A has the Texas logo printed on it and was purchased on DigiKey while device B does not have the Texas logo but the "TI" text and was purchased on a reel directly from the Texas Instruments e-store.
The printing on the device is identical except for the logo (symbol and text) and the supplier (DigiKey and Texas).
The datasheet indicates a consumption of about 50mA but we do not explain this difference. Such a high difference should be documented and the chip should have a different code.
Finally, both boards produced work with both devices apart from the different consumption.
But, since there are three devices per board, the total consumption is considerable and this is a problem.

Thanks

  • 0
    Guru 317480 points

    Marking: PCN#20211123004.0
    New design: PCN#20230814006.1

    The new design is CMOS; it is probably identical with the AM26C32.

  • 0
    Prodigy 10 points

    thanks for the reply.


    The first PCN explains the issue of the different marking from the logo to the "TI" text. I had also noticed this on other Texas chips.
    However, the second PCN indicates changes but which do not refer to the change in CMOS technology. In the PCN the link to the new datasheet is that of the AM26LS32AC where the typical consumption is 52mA. Apart from the datasheet, which may not yet be updated, I cannot understand how a technology that involves such a significant change in a very important characteristic could have changed without the name of the device being changed at all, not even for a final letter.
    Such a thing would create a dangerous precedent capable of calling into question both a project and a production. In this case the card still works but it may happen that a current consumption of 0mA is not always equivalent to a consumption of 52mA, resulting in a different sizing of the power supply especially if there are several chips per card.

  • 0
    TI__Guru 74801 points

    Hi Marinella,

    As Clemens pointed out. The new device you have is likely the newly designed die we've made. You can see the PCN for details.

    To explain the current difference. The old design of the die was designed in the late 1970s. During this time, the semiconductor industry mainly designed using BJTs as this was a mature technology back then. BJTs are controlled by pushing current through the base, by using many of them the overall current consumption would be very large. I believe in the 90's CMOS technology had matured to the point where they were now more cost efficient than BJTs which shifted designers to use CMOS over BJTs. The biggest advantage (aside from cost) is that CMOS are voltage controlled on the gate. This results in very low current consumption since these gates are high impedance. The new dies for these devices were redesigned in the early 2020s which mainly consist of CMOS technology. This is why you see such a large difference between the old die and the new die.

    BJTs are still used today, but mainly for applications where bandwidth/frequency needs to be very fast (CMOS gates are capacitive and it takes more time to turn them on and off compared to BJTs). 

    Marinella Carboni said:
    However, the second PCN indicates changes but which do not refer to the change in CMOS technology. In the PCN the link to the new datasheet is that of the AM26LS32AC where the typical consumption is 52mA. Apart from the datasheet, which may not yet be updated, I cannot understand how a technology that involves such a significant change in a very important characteristic could have changed without the name of the device being changed at all, not even for a final letter.

    TI's policy with electrical characteristics is to only change min/max values if they differ and issue a PCN when those values are changed. With typical values, these are not guaranteed. Customers are expected to design around min/max values so if a new design's characteristics are between the min/max then the application the previous die was designed in would still work with the newer die. Consumption current is one of those specs which lower is better and is not expected to cause a failure in new designs. 

    We issue PCNs to customers with the expectation that they evaluate the new dies in their system and validate the new dies work. 

    TI's move to replace the die is the alternative to EOL'ing the device since we are shutting down our fabs that still use the old process technology in favor or 300mm wafer technology. 

    -Bobby

  • 0 in reply to BOBBY
    Prodigy 10 points

    ok, the policy adopted by Texas to guarantee only min/max is correct, this allows the update of the technology without creating new part number. Now it is clear to me that this new CMOS device is a direct replacement for the old one, a lower consumption generally does not cause problems.
    I say generally because for example, in our case, the problem arose when the automatic test system of the board verified that the consumption, which was typically 150mA (with three chips per board), had now become almost zero. It is true that the characteristic is still within the minimum and maximum but usually a test system is calibrated starting from a series of sample boards and considering a certain margin. For years we were used to about 150mA and now seeing zero we were alarmed. We did not know about this change in technology and, since the manufacturer code remained identical, we could not distinguish them from the old chips.
    Anyway, thank you very much for the exhaustive and rapid response.