TL16C750: TL16C750 BAUDOUT Problem after writing divisor value 0x000C

This sounds as if only writing to DLM latches the entire 16-bit value. This would contradict the datasheet:

When either of the divisor latches is loaded, a 16-bit baud counter is also loaded to prevent long counts on initial load.

Ladisch-san,

Thank you for your prompt reply.
And sorry for my misleading expression.

I have rewritten the title and content, so please read the following rewritten content.


My customer asks us whether TL16C750 has writing sequence regarding to DLM and DLL.
Please let me know it.

The customer's evaluation results are bellow:

When the customer writes first 0x00 to DLM then 0x0C to DLL as divisor value 0x000C,
output from BAUDOUT sometimes does not come out.
When the customer writes first 0x0C to DLL then 0x00 to DLM as divisor value 0x000C,
corret output from BAUDOUT always comes out.

①first 0x00 to DLM, then 0x0C to DLL → BAUDOUT sometimes NG

  The writing sequence is bellow:

    LCR b(1)=1, b(0)=1
    LCR b(7)=1

    DLM 0x00
    DLL 0x0c

    FCR b(2)=1, b(1)=1, b(0)=1
    LCR b(7)=0
    MCR b(1)=1
    MCR b(2)=1
    IER b(2)=1, b(0)=1

    note:  When BAOUOUT does not come out, the customer confirms that the correct value
               0x000C is written in the register

②first 0x0C to DLL, then 0x00 to DLM → BAUDOUT always OK

  The writing sequence is bellow:

    LCR b(1)=1, b(0)=1
    LCR b(7)=1

    DLL 0x0c
    DLM 0x00

    FCR b(2)=1, b(1)=1, b(0)=1
    LCR b(7)=0
    MCR b(1)=1
    MCR b(2)=1
    IER b(2)=1, b(0)=1


Best regards,
Fumio Nakano

Ladisch-san,

Thank you for your cooperation.

I got the following information from the customer.

・There are many samples that have the same phenomena.
・There is no sequence problem for divisor value 0x0003.

So, I think TL16C750 has DLM/DLL writing sequence as to divisor value 0x000C.

The customer must report measures to end users.
So, we hope to receive your answer as soon as possible.

Best regards,
Fumio Nakano