Re: Horrid Serial Dacs

From: John Larkin (jjlarkin_at_highSNIPlandTHIStechPLEASEnology.com)
Date: 09/07/04 

Date: Tue, 07 Sep 2004 15:03:51 -0700

On Tue, 07 Sep 2004 22:50:00 +0200, Andreas Hadler
<Andreas.Hadler@t-online.de> wrote:

>John Larkin <jjlarkin@highlandSNIPtechTHISnologyPLEASE.com> wrote:
>
>>Oh, I'm just coding up an embedded processor for an instrument that
>>has three different kinds of serial DACs (Burr PCM56 and Maxim 5742s
>>and 5205s) and they are all their own sort of mess. A few weeks ago I
>>had to persuade some Analog Devices parts to work. One Max chip uses
>>rising edge clock, the other is falling-edge. The 5742 has 35
>>different serial commands.
>>
>>Bizarre.
>
>I once had a project that cost me some days to figure out how to
>connect some '165/'594, 2 DACs, an RTC, a serial EEProm and an ADC to
>just one Motorola QSPI with automatic SPI scanning, using as less as
>select pins as possible and allowing a fixed QSPI command and data
>structure, only varying the QSPI start- and endpointer.
>A real complex interaction of hard- and software, as different data
>lengths, delays between commands and data answers, different edge
>sensitivity, dead ends and feed-throughs etc. had to be considered.
>

Unless I'm seriously short on CPU power, I'd much rather software
bit-bang than try to use the Moto QSPI. Like the TPU, it's a wonderful
gadget once it's working, but a nightmare to code.

I have a couple of products that use an 8-char, 16-seg alphanumeric VF
display. The QSPI scans the grids and updates the anodes all by
itself, through some Allegro serial drivers, so the display keeps
scanning even if you stop the code. That's pretty impressive.

John

Relevant Pages

  • Re: Horrid Serial Dacs
    ... >different serial commands. ... just one Motorola QSPI with automatic SPI scanning, ... Today, I consider it a lesson, what kind of design should be avoided ... Have fun! ...
    (sci.electronics.design)
  • Re: Horrid Serial Dacs
    ... >>just one Motorola QSPI with automatic SPI scanning, ... >>select pins as possible and allowing a fixed QSPI command and data ... For the QSPI and coding the TPU I agree. ...
    (sci.electronics.design)
Quantcast