• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 747 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TLC2543: Outputs only 8 bits of precision when 16 bit interface selected.

user3319081
Prodigy 20 points
Part Number: TLC2543

We have been using the TLC2543 for many years with the same circuit and software, which respects all setup, hold and conversion times with a margin. We now find that a number of recently manufactured units have the same fault - the TLC2543 outputs only 8 bits of precision followed by 8 zeros, not 12 bits of precision followed by 4 zeros, despite 16 bit interface being selected in the command byte. We can see that the clock line has no glitches. We have looked at every angle on the CRO and we can find no fault, and most units correctly output 12 bits of precision. What can cause this? We thought that we might be reading a new channel too soon after the last, but we padded our software to wait 12us from EOC high to /CS low and thisachieved nothing. Everything looks normal except the output bits and previous batches worked. We have also tried replacing the A/D with one from an older batch to no avail.

  • 0
    TI__Genius 16890 points

    Hello,

    Please try the following.  Take two boards, one that is working correctly and one that is not working correctly.  Take the part from the good board and put it on the bad board.  Take the part from the bad board and put it on the good board.  Does the problem follow the part or the board?

    Do you have scope wave forms of a bad part showing CS, clock, and data out that you can post?  What is the value of the Input Data Byte that you are writing?

    MIke

  • 0 in reply to Michael Stout
    Prodigy 20 points

    Thanks Mike. We swapped A/D chips and the problem stays with the board. The circuit is as per the TLC2543 data sheet with all the bypass caps and is proven in previous products. On the scope the signals appear correct as per the data sheet page 11 figure 13: Timing for 16-Clock Transfer Using /CS with MSB First, with plenty of time for setup, hold and conversion. SPI clock is 3.7MHz and there is no sign of glitches. The rising edge of the SPI clock is in the middle of the A/D data in. On the faulty units we can clearly see that the data out has only 8 bits of precision in 16 bits out as we vary any analog input. This isn't intermittent and changing A/D chips from another batch doesn't fix it. No inputs are floating. At half scale in the output is 0x8200 on every board with this fault. This is a weird one.

    We send it command 0x0C00 to read input AIN0. We discard the data in and send 0x1C00 to read AIN1, with data in now for AIN0. It reads as only 8 bits of precision, as we see on the scope. We aren't losing the other 4 bits in buggy software, the A/D really is only giving 8 bits of precision. I even tried the command 0x0C0C to see if somehow it was going into 8 bit interface mode but the result wasn't 0x8282, it was still 0x8200.

    We can't see any difference on the scope between the signals on the units that output 16 bits with 12 data and the units that outputs 16 bits with only 8 data.

    VCC is 4.960V. VREF+ is 4.995V. VREF- is 1.000V. Analog inputs have 250R to ground for current loop, then 1k protection to the pin.

    Tomorrow I shall try lowering the SPI clock frequency but the high & low times are well within spec so I'm not hopeful. Regards, Daniel.

  • 0 in reply to user3319081
    TI__Genius 16890 points
    Hello,

    You mentioned that the problem stays with the board. That would seem to indicate that there is a problem with the board and not the ADC. If parts work on some boards and then don't work on another board, there is some type of problem with the board, or something that is causing the part to act incorrectly on that board.

    If you send a continuous stream of command 0x0C00 to the part so that you are only reading AIN0, does the part continue to act incorrectly?

    You mentioned that VREF+ is at 4.995V and VCC is at 4.960V. You need to decrease VREF so that it is at VCC or below.

    That would be a good thing to try lowering the SPI clock rate. The terminal pin description of the inputs on page 3 notes that the analong input driving source impedance should be less than 50 ohms for a fast clock of 4.1MHz.

    Is your EOC waveform also as shown in figure 13?

    Mike