1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

PCMCIA or USB to DB9 serial port adapter

Discussion in 'IBM Thinkpad' started by none, Jun 8, 2007.

  1. none

    none Guest

    The ThinkPad here has Win98SE. I need to have at least one, preferably
    two conventional DB9 (COM1 & COM2) serial ports for certain software to
    run. The adapter can be either USB or PCMCIA. Would like input on what
    others have found to be to be a good adapter, compatible with 98.

    Thanks
    Vic
     
    none, Jun 8, 2007
    #1
    1. Advertisements

  2. none

    Mike S. Guest

    It depends, in part on the software - so it would help if you told us
    exactly what you are trying to do. If the software is old and requires
    register-level access to a REAL hardware serial port, then the PCMCIA
    route is more likely to work as it provides exactly that. Socket
    Communications manufactures the gold standard here, though there are many
    cheaper alternatives.

    If the software is of recent vintage and will work with an emulated port,
    then a USB dongle will probably work, is much much cheaper and much easier
    to find. I've not seen any 2-port devices of this kind, though. For older
    software it's a crap shoot with about a 50% success rate according to some
    postings I've seen.

    I've had success with the Belkin dongle, available almost everywhere, that
    is sold for connection to PDA's. Just be sure that you are using Windows
    98SE as there is very little driver support for USB devices under Windows
    98 Gold.
     
    Mike S., Jun 8, 2007
    #2
    1. Advertisements

  3. none

    none Guest

    Very informative post Mike. To answer some of your questions:

    The software is all radio communications related. If you are a amateur
    radio operator then you'll understand most of this. Software is
    everything from old DOS PaKet which interfaces with a TNC (terminal node
    connector) for packet operation, to seven year old WinAPRS for running
    APRS (automatic position reporting system). For those unfamiliar with it
    here is a web link which explain what it is:

    http://web.usna.navy.mil/~bruninga/index.html

    Here's another which will show you APRS stations operating in south FL.
    Scroll down to the java applet and wait. Clicking on a WX icon should
    display that weather stations information at the bottom of your browser:

    http://www.aprs.net/

    Another application requires a serial port to control the radio (CI-V
    interface). It too is many years old (five I think).

    Does any of the software access serial ports at the register level?
    Don't know but would guess NO. My concern of course is spending money
    and time needlessly. Of those little USB to DB9 adapters everywhere
    (I've been looking on ebay) and found the following:

    http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=150130034451

    which said the following:

    Various USB Serial adapters have different incompatibilities. The most
    compatible are IOGEAR cables that you can get for $40-50 from large
    stores. In particular, this cable is incompatible with GARMIN GPS, as so
    are many others except for IOGEAR.

    So, am I playing russian roulette buying one of these? That's why I
    asked for input from those with actual experience. Don't want to spend
    $50, especially if a $6 or $8 unit will work.

    Vic
    ___
     
    none, Jun 8, 2007
    #3
  4. none

    CBFalconer Guest

    none wrote: *** and op-posed - fixed ***
    Please do not top-post. Your answer belongs after (or intermixed
    with) the quoted material to which you reply, after snipping all
    irrelevant material. See the following links:

    --
    <http://www.catb.org/~esr/faqs/smart-questions.html>
    <http://www.caliburn.nl/topposting.html>
    <http://www.netmeister.org/news/learn2quote.html>
    <http://cfaj.freeshell.org/google/> (taming google)
    <http://members.fortunecity.com/nnqweb/> (newusers)
     
    CBFalconer, Jun 9, 2007
    #4
  5. none

    Ben Myers Guest

    You do not know when to give up, do you? ... Ben Myers
     
    Ben Myers, Jun 9, 2007
    #5
  6. none

    PJ Guest


    Roger, Wilco. I understand the 'evils' of
    the dreaded but quickly readable top post!
    *
    Bottom post down below:
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    *
    Here's one I bought for use with W98SE on an
    R31 Tpad that lacks a RS-232 port:

    Manufacturer: USBGEAR
    Barcode: 879676773444431
    Part Number: U232-P9AP

    at www.usbgear.com

    Works OK for bidirectional OBDI and OBDII
    (GM) data at 160 baud and 8192 bps.

    Comes with drivers for Windows -- I think
    they have a Mac version as well.

    My experience with Win98SE is that it takes a
    different driver for each USB device one uses
    make sure you get a driver with the adapter.
     
    PJ, Jun 9, 2007
    #6
  7. none

    Lars Guest

    Plonk!

    Lars
    Stockholm
    Visit the big ACF freeware wiki:
    http://freeware.wikia.com/wiki/Main_Page
     
    Lars, Jun 9, 2007
    #7
  8. <snip>
    Greetings.

    If I were you, I'd look for newer software. I'm afraid that the APRS and
    packet software you're dealing with is probably going to go straight for
    COM: ports with standard IRQs and base addresses, just like so much DOS
    software of the past. I'm looking at a Yaesu FT-736r that a friend has,
    saving up my pennies to buy it off him, not only for voice, but also data,
    and since I'm a COM: port short on both TP R51 and desktop, I'm going to
    have the same issues you're having if I decide to get back into packet
    radio. (My PK-88 quit, and I never could figure out how to fix it, nor sent
    it in to get fixed.)

    It's going to take some major rewriting of the serial port drivers in that
    old DOS software to get them to *find* USB ports, much less send and
    receinve data through them. There's just too much RS-232 specific stuff
    that's sent and looked for. Sorry.
     
    Tom Rutherford, Jun 10, 2007
    #8
  9. Just a personal question: Do you EVER post anything useful to others -
    rather than your continued useless rants about top-posters (and I agree, top
    posting does not conform to the original intent - however - the top-posters
    really don't care whether or not you like it)??????????

    I've been about this forum for months, and I don't ever recall seeing
    anything other than a "correction" of other peoples posts.

    Oh, well - at least it is an unusual hobby.

    And yes - I am an owner of a ThinkPad (who also notes that the number of
    posting on ibm.ibmpc.thinkpad consistently is higher than those on the "big
    8" newsgroup that we are reminded of weekly).

    Jim
     
    Ballroom Dancer, Jun 10, 2007
    #9
  10. none

    none Guest

    Hi Tom

    So you are a 'ham' too! Read your post with interest. If (IF) it's true
    the software I have will look for 'old fashion' COM ports then perhaps
    we will have to go the way of PCMCIA to COM (not USB to COM. According
    to Mike S. (previous post), that COM solution (PCMCIA) will more likely
    work. A consideration he mentioned was PCMCIA is being phased out for
    newer technology so that might effect future issues. Also PCMCIA
    adapters cost more. Hmmm, decisions, decisions, which one will work, and
    I've not had 'the luck of the draw' in my favor!

    BTW, OT, did you take a look at the JAVAmap link in the original post?
    In case you did not, check it out and notice there is a menu window at
    the bottom of the application. It allows the user to select what section
    of the USA to view. You can even view APRS stations across the USA or
    even the world! And in case you were not aware, APRS activity has picked
    up quite a bit in the last few years. Don't know about packet but APRS
    is doing good. Perhaps a lot of it has to do with the numerous natural
    disasters and people simply gearing up for emergency communications! I
    went through three direct hit hurricanes in FL and found our hobby to be
    a tremendous asset! In fact, the EOC (Emergency Operation Centers) and
    hurricane shelters even use packet! IN FACT, a modern EOC in FL now
    includes MULTIPLE HAM radios with capable operators and about every
    medium of communication we know (SSTV, PACKET, VOICE, VHF, HF, SSB, FM,
    AM, analog, digital, etc, etc.). The city administrators don't fool
    around, they try to appropriate every dollar they can towards radio
    equipment because they KNOW the ham operators come through, quickly,
    efficiently!

    Hope your interest has been tweaked a little further to look into that
    aspect of the hobby. It was too bad to hear the PK-88 breathed its last
    but you can pick up a TNC on ebay or a swap net rather inexpensively.
    I'd suggest the KPC-3+ with v8.3 EPROM (that's what I have). It is a
    very nice TNC

    Thanks for your input
    73 de Vic
     
    none, Jun 11, 2007
    #10
  11. none

    Mike Y Guest

    I'm assuming your ThinkPad is Cardbus, and not PCMCIA. With PCMCIA
    you only have one interrupt, so even with one of the old dual-port serial
    PCMCIA cards you're a bit hampered for full functionality. Maybe it
    won't matter though.

    However, I've not seen dual-port serial PCMCIA cards at a reasonable
    price for some time, and I don't recall seeing PCCard (Cardbus) dual-serial
    cards at all.

    However, I HAVE seen PCCard USB cards at a number of places...

    Mike
     
    Mike Y, Jun 11, 2007
    #11
  12. none

    Mike Y Guest

    Good luck! I think you'll need it.

    Most of the HAM software I've dealt with seems to be carryovers from old
    DOS stuff, or upgrades that keep trying to be 'backwards compatible' to
    hardware and requires IO ports and access to IRQs. Right there that will
    TOTALLY rule out a dual-serial PCMCIA card as the PCMCIA slot only
    has ONE interrupt line for the whole socket.

    I've seen USB-serial cables blasted in HAM circles, and it really isn't the
    fault of the cables, it's the fault of the OLD software that doesn't know
    how
    to use the cables.

    Mike
    N1DVJ
     
    Mike Y, Jun 11, 2007
    #12
  13. none

    Mike Y Guest

    Hi PJ!

    I'm interested in what you're using for OBD type interfacing. I was going
    to
    buy some software for that, and the Belkin USB-Serial cable was recommended
    so I have that. (The software is 'protected' in that it picks up the serial
    number
    from the Belkin cables and now is 'transportable' from machine to machine
    with no copy protection, as long as it always finds the same Belkin cable)

    Mike Yetsko


     
    Mike Y, Jun 11, 2007
    #13
  14. Hi, Vic.
    I think I missed that part of the post, but yes, I'd go with the PCMCIA
    solution, too, unless you can pick up a cheap docking station or port
    replicator. Man, my Mini-Doc has so dang many ports on it, and some I've
    never seen before. :) But, there is a COM port there. I picked it up for
    ~$100 from Newegg, including shipping. Not sure if they still have the deal
    going, but it's normally a $190 item. Best to look to see what it's
    compatible with, though, before you lay out the greenstamps.
    Yep. We may be amateurs, but especially if we've earned our RACES cards,
    we've been trained to operate professionally. As for the proliferation of
    APRS, it might also have something to do with low-cost and very accurate GPS
    units. The APRS program I played with back in the '90s was free, except for
    the ability to interface with GPS. But, GPS wasn't nearly as accurate then
    as it is, now. It's all well and good to let the APRS program run and
    assume speed and direction are going to be constant, but without an accurate
    input, you can get outta whack in a big hurry.
    Well, the interest is there, but the money isn't quite yet. And, I'm
    looking for something a bit more on the order of the PK-232. 1200 baud was
    fun, back in the day, but I'd go crazy at that speed, now. Even 9600 or
    19k2 is going to be a challenge if I can get that fast. Calculating
    modulation index and so on, 19k6 is the limit per FCC regs for 70cm, I thin,
    because you're only allowed a bandwidth of 20kHz for packet. At any rate, I
    might play with a KPC-3+ if I can get one really, really cheap, like $20 or
    so.
    Take care.
     
    Tom Rutherford, Jun 13, 2007
    #14
  15. Since he's talking about packet radio, and with the TNC he's running,
    9600bps is about all he's going to need, and only that fast if he wants to
    take advantage of whatever data compression is built into the unit. I
    remember running my PK-88 at 9600 for 1200 out, and that was nice because of
    the data compression, plus it got me into command mode, into the mailbox,
    etc., a lot quicker.
     
    Tom Rutherford, Jun 13, 2007
    #15
  16. none

    Mike Y Guest

    It's not a speed issue.

    PCMCIA sockets only have ONE interrupt pin. As a result, it doesn't
    matter WHAT is on the card, you only have ONE interrupt at the
    connector. I suppose you could write a custom driver that uses that
    pin as a third interrupt and then virtualizes two that you need from it, but
    then you still run into the program limitation of hardcoded com ports.

    The hardcoded port limitation is because early software assumes
    COM1 is com1, but what really is com1? Is if the first port address
    in the BIOS Data Area list? (The BDA) USUALLY that will be
    at port 3F8. Usually. But NOT always. However, even if a program
    looks to see, EVERY software package I've seen will ASSUME that
    if the IO port is at 3F8 (regardless of how it figures that part out) the
    IRQ will be IRQ4. If the software sees an IO address of 2F8, then
    it assumes IRQ3. After that it gets tricky, as the BDA allows for
    COM3 and COM4, usually at IO of 3E8 and 2E8, but where the
    heck do you find the IRQ? There are some 'trends' as to what
    manufacturers did, and yes, there were ISA cards that had up to
    COM8, but NOWHERE is there any kind of consensus as to what
    IRQ to use for com ports after the second one.

    It's not all the fault of the early software writers... The early machines
    were kind of crude and kludgey, and the com routines left a lot to
    be desired, so the developers did their own.

    Nowadays though, it makes sense to use the resources of Windows
    to access com ports. If done right, you can have hundreds of ports
    on a machine. And it doesn't even have to physically be on the
    machine, they can be anywhere. Have virtual com97 out on the
    web, and it works the same as com5 directly on the laptop. Well,
    almost, there's always the potential for some lag, but for 99% of
    the applications it doesn't matter.

    Mike
    N1DVJ
     
    Mike Y, Jun 14, 2007
    #16
  17. Isn't the usual practice to share IRQs between COM1 and COM3, and between
    COM2 and COM4? That's what I always assume is happening, so if I'm using a
    rat on COM1, I won't be using a TNC on COM3. Thankfully, most pointing
    devices nowadays don't require a COM port, but are usually either USB or
    PS/2.
    Now, that would be strange. :) I do remember, back when I was playing
    with OS/2 Warp 3, I used a comm. driver called SIO.SYS, from the same guy
    who wrote the X00.SYS FOSSIL driver for FidoNet and other BBS programs at
    the time. They both acted to interface the software with the hardware to
    make everything run smoother. IBM's original comm. driver for OS/2 was a
    mess, but I can't recall now why. I just knew that every time I reinstalled
    OS/2, I'd better install SIO. At any rate, I was sort of wondering if the
    PCMCIA card could be made to do a little of its own interrupt hanling, but I
    guess that wouldn't solve anything.
    Ah! Another ham, and from near ARRL Country. :)
     
    Tom Rutherford, Jun 14, 2007
    #17
  18. none

    Mike Y Guest

    Yes. And no. Yes in the sense that a lot of setups did that. No in the
    sense
    that you can't run both com ports 1-3 or 2-4 of the pairs at the same time
    with interrupt support UNLESS you write a custom SINGLE driver that
    hooks into the interrupt chain, figures out what to do, and then dispatches
    the
    interrupt to the right routine. The other option is to run one of the ports
    as
    polled only.

    Now, you can KLUDGE it so that it works... But I won't go there for now.
    In any case, the kludge requires custom software as well.

    Interrupt sharing was not an idea favorable to legacy PC designs. That's
    not
    to say it wasn't done. The IBM ThinkPads, like the 700 series, would
    allow it with Win95 but only IF you had a special driver installed. Then
    they would allow interrupts to be shared but only interrupts from 'swapable'
    devices. However, if you went into the registry and found the mask for
    sharable interrupts (I can't remember if it was in the CardServices entry
    or in a separate area) you could change the mask so that ANY interrupt
    could be shared between ANY device. Or at least attempted. Software
    implementations could bite you if you weren't careful. This went away
    with Win98, as it allowed the sharing anyway, but not between all devices.
    Well, remember that the PCMCIA card, at least for serial ports, does NOT
    have any code on it. The most it has is tuples. As far as I know, there
    were
    no XIP (execute in place) implementations that included serial cards.

    Also, the PCMCIA card has a generic 'int' pin. It doesn't even know what
    the interrupt will be. Now, there were a few brain dead serial cards that
    only put 'legacy' configurations into the tuple descriptors, but thankfully
    they died quick deaths. I tried making some kludgey configurations for
    those cards and could get them to work on a 'one on one' basis, but not
    in a generic sense. I finally figured what the heck, they're brain dead
    anyway and said tough. The best way was for the tuple descriptors to
    specify what the card was and that it used x number of sequential registers
    and an interrupt and then let the configuration software try to match it to
    a
    legacy address with an iterrupt if possible.

    By the way, this is why some manufacturers, like Compaq, made the
    default condition on some laptops be COM1 and COM3 for the internal
    serial stuff, and leave COM2 legacy address open for PCMCIA use.
    The internal stuff would share a common interrupt for 1-3 (irq4) and then
    leave COM2 open for a PCMCIA modem to be inserted and use
    irq3. Yes, making the internal ports non-sequential technically violated
    the spirit of the specs, but it allowed everything to work. Without that
    trick, it meant that any inserted PCMCIA modem card would have to
    use a 'unique' and non-shared interrupt like 5 or 9 and neither was a
    standard and a lot of terminal programs wouldn't work. I took flack for
    making that suggestion when the LTE5000 was in developement, but
    that scheme was eventually adopted by HP, DEC, GateWay and others
    as the past of 'best' compatibility.

    (And yes, with Windows 9x you COULD make the shared com ports
    share an INT pin for built in devices on SOME platforms, but
    then NOT with legacy 'DOS compatible' software support.)

    Now with PCCard it's a different story. PCCard is essentially CardBus,
    which is for all intents and purposes a hot-swapable PCI superset. The
    socket controller turns into a PCI bridge to the card, and you will see
    it as such if you walk the PCI bus. There is no int pin as such. A few
    chipsets had what was called a 'side channel' interrupt, but that was for
    legacy PCMCIA support. In CardBus the int pin goes to PCI_IRQx
    on the PCI bus which is one of the four 'rotatable' interrupt lines that
    the host to PCI bridge now manages, and don't really exist as an interrupt
    in the classical sense the way they do as IRQ 0-15.
    Well, not originally. While I'm actually from PA, I got my first call as
    KA5MJQ when I lived in Texas. I changed to a '1' call after I moved
    up here in 1985.
     
    Mike Y, Jun 15, 2007
    #18
  19. And, some interrupt controllers may or may not allow the polling. Back to
    OS/2: I had a motherboard that wouldn't allow polling for the printer. I
    had to hard code into the CONFIG.SYS file that the printer was on IRQ7.
    A hardware kludge? I've seen articles back in the ISA days about a
    multi-I/O card having the IRQ lines for its serial ports tied together via
    diodes, and both sharing one line. But, that was for a very specialized
    purpose, and not recommended for everyday stuff.
    Right. I heard back in those days that NICs didn't like to share with
    anything else.
    Okay. I was just thinking that something could be made that would do it.
    But, with only one interrupt pin, that would've been tough.
    Oh, okay.
    As long as it worked and it was well documented, it's no worse than what
    Tandy did to some of their desktops.
    I've only had two calls (was KA8NMM), and both 8 calls, obviously. Nothing
    exciting like moving all over Creation, except when I was too young to
    remember it. Born in 4-land, lived in 3-land for a while, then been in
    8-land for the past 50 years, first in Ohio, then in Michigan. Was in
    Michigan 13 years before getting licensed, though.
     
    Tom Rutherford, Jun 17, 2007
    #19
  20. none

    Mike Y Guest

    You could poll just fine. IRQ7 had a different issue, as there was a bug
    in the 8259 that there were 'phantoms' and they all generated IRQ7. (We
    found that with the early mouse board for the Tandy 1000. The IRQ would
    get 'lost' but an IRQ7 would get generated. IRQ15 had the same issue
    on the AT)

    While there was a lot of tricks to handle all the things that happend, a LOT
    of the printer stuff was polled. Thing is, you didn't need an IRQ for a
    printer.
    Interactive interrupts on a byte level were overhead that was actually a
    performance hit. The best thing was to poll the 'status' every so often,
    and
    when it was clear, you could just stream bytes as fast as you could talk
    to the port and check status to see if the printer filled up.

    It's the same with the com ports. Interactive byte level interrupts
    actually
    slowed you down. That's why with the 16550 and so on they have
    'pipelines' that when you get an int it means time to fill the pipeline,
    then
    you stream till it's full. Some of the modems even had phantom pipelines
    that let you stream up to 2K!

    Oh, interrupt sharing was easy. IF you had control over the hardware
    path up until the bus. The problem was that the ISA bus in a PC was
    active high... If it were active low then you could have used as many
    devices as practical on a single interrupt. Set your system up for say
    IRQ4 being EVERY com interrupt... Then you just have your software
    register to get a callback.. A single int could have handled a dozen
    com cards. (Hey, sounds like the original TRS-80! Naw... they
    would have never done it right...)
    That was because of poor NIC software, nothing else.
    As I said, it would have been easy to write a routine that would handle
    a dozen cards with one interrupt. If it's written right.

    Tandy had this Sync com card for the T2000 with 4 ports on it. Each
    port could interrupt 3 different ways, rx, tx, or status. That's 12 things.
    Tandy sold that machine with special software with TWO of those
    cards in it all using the same IRQ. No problems. Well, no problems
    with the com...

    But ONLY if you had the right CardSoft package from SystemSoft, and
    ONLY if you knew how to go in and set it up. On all but the IBM machines,
    the code was not configured or configurable for this except by manually
    going into the registry and 'fixing' it. On the IBM platforms it would
    allow the sharing automatically, but only in certain cases.
     
    Mike Y, Jun 17, 2007
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.