Re: Purpose of txtail ?

From: Joerg Reuter (hzxwcugk@smrtcall.net)
Date: Sat Feb 05 2000 - 02:46:43 EET

  • Next message: Joerg Reuter DL1BKE: "Re: DG2FEF-AX.25 Patch for 2.2.14 available"

    > > If the modem needs a delay it should be responsible for creating the
    > > appropriate delay. The software should not be responsible for operation of
    > > the modem.

    Welcome to the real world. Most don't. MODEM designers say it should
    be done in software because it is more flexible, programmers say it
    is an hardware issue... The MODEM developer wins, though.

    > Thatīs true. It should also be responsible for generating the txdelay
    > via the CTS line.

    I agree: it should, but in the real world how is the MODEM supposed to
    know how long the TRX takes to stabilize on its transmit frequency (and
    the time the remote DXE needs to synchronize on the signal)?

    The Z8530 has a nice auto-CTS/RTS mode I cannot use because the USCC
    uses CTS for timer purposes and nearly no MODEM provides a reasonable
    CTS signal.

    > As another major influence of this parameter is the
    > RF design (PLL lock delay etc.) a modem should be locate inside the
    > transceiver. As a transceiver requires a little controller anyway
    > (PLL divider ratio programming) it could also do transmitter keying
    > sequencing and CTS line control.

    Yes, but a great part of amateur radio includes using off-the-shelf
    components for purposes they weren't intended for. It isn't that much
    different from the commerical world, BTW.

    > But since we must also support old equipment we have to provide
    > software-txdelay and txtail which is - at least for SCC based devices -
    > a little bit problematic due to the fact that the resolution is given
    > by system timer intervals (one jiffy, 10ms).

    Not only that, most cards do not support DMA, and even if they do,
    ISA DMA is a pain in the a.se. With cards like the USCC everything
    above 19200 baud is generating an enormous load, especially because the
    IO port access to the Z8530 is ridiculous slow. Modern variants of
    the chip are much better, but still hard to get. Plus they don't do
    much else (at least for our purposes) than HDLC.

    A cheap, general-purpose DSP card with a PCI DMA busmaster interface
    would be nice. Plus a 70cm TRX that takes a 455 kHz base band signal for
    in- and output, frequency adjusted by two PLLs (one for TX, one for RX)
    and other measurements to keep down switch time... And both together
    shouldn't be more expensive than EUR 300... Yes, I know, I'm dreaming.

    Joerg Reuter http://poboxes.com/jreuter/
    And I make my way to where the warm scent of soil fills the evening air.
    Everything is waiting quietly out there.... (Anne Clark)





    This archive was generated by hypermail 2b29 : Sat Feb 05 2000 - 20:09:17 EET