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.

ADS8598H: ADS8598H Readout frequency (serial and parallel)

Part Number: ADS8598H

Hello,

I have two questions about the ADS8598H data readout.
The data sheet shows some contradictions in this topic.

(1)
For parallel readout, the minimum RD# low and high times are 15ns both, what leads to a max. parallel readout frequency of 33.33MHz.
For serial readout, the minimum SCLK time period is 50ns, which leads to a max. serial readout frequency of 20MHz.
Is it correct that the serial readout frequency is lower than the parallel readout frequency?

(2)
In section 7.4.2.3.3 "Serial Data Read", I find the following phrase:
"The primary disadvantage of using just one data line for reading conversion data is that the throughput is reduced if a data read operation is performed after conversion."
This implies that even with "one lane serial data readout", maximum throughput (500kSps) can be achieved when using "read during conversion".
When using both lanes (DOUTA and DOUTB), maximum throughput (500kSps) can also be achieved when using "read after conversion"?

Unfortunately, the maximum SCLK frequency of 20MHz (50ns) does not fit to this. Serial readout needs a minimum of 72 clock cycles. 72*50ns = 3.6us.
The conversion time is 1.19us to 1.29us.
When using "read after conversion", the conversion time adds to the readout time. 1.29us + 3.6us = 4.89us This is a bit tool ong for 500kSps.
When using "read during conversion", the minimum conversion time of 1.19us is a bit too short to finish the 3.6us serial readout prior to the falling edge of BUSY.

I am looking forward to your answers.

Regards, Niels

  • Niels,


    Unfortunately, your question requires a bit more information than I have for a good answer. A lot of the applications people are out, so it may take extra time this holiday season to get you a good answer.

    I'll forward on this question, but again it may take a few extra days.


    Joseph Wu
  • Niels,


    I made a quick search and did come up with this post about the limitations of serial data read.

    e2e.ti.com/.../2418341

    It looks like the serial data read isn't capable of a 500kSPS readout. Let me know if that answers your basic question.


    Joseph Wu
  • Hello Joseph,

    first of all, I would like to wish you a happy new year.

    for me, there is still a contradiction in the ADS8598H data sheet:

    According to the technical data, it is not possible to achieve 500ksps with serial readout, even when using both data lines. But accoring to the text in section 7.4.2.3.3 "Serial Data Read", the maximum throughput (500kSps) can be achieved even with "one lane serial data readout" when using "read during conversion".

    To keep it simple, you could reduce it to the following questions:

    1. Is the minimum SCLK time period of 50ns from the data sheet correct?
    2. Is it correct that the max. SCLK frequency is lower than the max. parallel readout frequency?
    3. What is the maximum achievable sample rate when using serial readout with both data lines?

    I am looking forward to your answers.
    Regards, Niels

  • Niels,


    Thank you. Happy new year to you too.

    Let me start with the questions directly:

    1. I believe it is correct that the minimum SCLK period is 50ns. The timing information in the datasheet is usually based simulation and then characterization of the process used for the design.

    2. I'm not sure I understand this question. Are you asking if max SCLK frequency should be faster to get out the 500kSPS data rate that can be clocked out using the parallel readout method?

    3. I believe that the max sample rate with serial readout is 278kSPS. I still think that it is not possible to achieve 500kSPS with serial readout and this is stated incorrectly in the datasheet. This is was was commented on in the post that I provided a link to.

    I will admit that I'm not the normal support for this product (I was filling in during the holiday break). I'll check in with the normal support and see if he has any alternate answers or comments.


    Joseph Wu
  • Joseph,

    even though you are only the vacation replacment, your answers are more helpful than some of the expert answers I got in the past.
    Thank you for your help so far.

    Let me give you some explanaition to question 2:
    My experience with other devices is, that serial interfaces can typically clocked faster than parallel interfaces.
    With the ADS8598H, it seems to be the other way round:
    For serial readout, the minimum SCLK time period is 50ns, which leads to a max. serial readout frequency of 20MHz.
    For parallel readout, it is the RD# signal that times the length of the data access, thus acting as a kind of clock signal. The minimum RD# low and high times are 15ns both, summing in 30ns for one "clock" cycle. This leads to a max. parallel readout frequency of 33.33MHz.
    As you can see, the ADS8598H can output its data at a rate of 33.33MHz when using its parallel interface. But when using the serial interface, the rate is reduced to 20MHz.

    Until now, nearly a whole year passed by since the first question to this topic came up: click here
    But the ADS8598H data sheet is still unchanged. I wonder why.

    It would be nice if someone from the expert group could give his comments to my questions after they returned from their vacation.

    Regards, Niels

  • Niels,

    I did hear back from Dale. He had some communication with the India team that designed this device (the systems engineer is also remote and is at that location).

    The 500kSPS sampling rate is only for the parallel interface. Additionally, he added some extra comment to the serial interface. The serial interface with DOUTA and DOUTB requires 18x4=72 SCLK cycles, and the total cycle time = 72x50ns+ tCONV+ tD_CNBSY =3.6us+1.24us+15ns =4.855us, so the max sampling rate for serial interface is 206kSPS, which is a little slower than I had mentioned in my previous post.

    Dale sent an email to the systems engineer and we'll see if we can get the datasheet changed soon. In the mean time, at the bottom of every other page, starting on page 2, there is a clickable link for Submit Documentation Feedback:

    Can you please click on the link and submit a comment? It may help get the ball rolling faster (and you may get more direct comment from the systems and design team).

    Joseph Wu

  • Hello Joseph,
    206kSps is not the answer I hope for. But anyway, thank you for your help.
    I will use the button you suggested.

    Regards, Niels