Prev: Smart wristband
Next: No multicore grahics from Intel, Intel withdraws Larrabee!
From: david on 25 Aug 2009 05:43
On Tue, 25 Aug 2009 00:37:41 -0700, Dan Lenski rearranged some electrons
to say:


> It's one of these awful custom cards, seemingly unique to the instrument
> manufacturer. It is basically a glorified GPIO port, as far as I can
> tell... but it's *their* glorified GPIO port, and no one else knows
> exactly how it works. They want $10,000 to replace the ISA card with a
> newer PCI version. Yes, $10,000. Scientific equipment manufacturers
> are ridiculous, especially when they choose to use custom cards.
> Thankfully, that's becoming less prevalent as standard
> serial/USB/Firewire interfaces are increasingly provided. Either that
> or GPIB... which is basically a glorified serial port with an extremely
> bulky cable, costing "only" hundreds of dollars for the interface card.
>
> </rant>
>

Actually, GPIB is parallel, not serial.
From: david on 25 Aug 2009 05:47
On Tue, 25 Aug 2009 04:26:41 -0400, Paul rearranged some electrons to say:


> Obviously, if the bridge is from USB to ISA, the environment it is used
> in, is going to have to understand USB. I don't know what shape official
> USB support is in for Win98 or ME.
>

If the OPs' instrument software works by writing directly to a memory or
I/O mapped location then the USB to ISA converter is likely not going to
work.
From: Dan Lenski on 25 Aug 2009 14:16
On Aug 25, 5:43 am, david <n...(a)nospam.com> wrote:
> Actually, GPIB is parallel, not serial.

Right, but it basically functions as a glorified serial port: you send
a stream of bits/bytes down it to your instrument... the instrument
sends a stream back. About the only interesting features is daisy-
chaining of devices, which is sometimes handy but requires very bulky
and expensive cables.

I wish we just used RS-232 interfaces for all of our instruments
instead of GPIB!

Dan

From: NT on 26 Aug 2009 16:00
On Aug 25, 9:26 am, Paul <nos...(a)needed.com> wrote:
> Dan Lenski wrote:
> > On Aug 25, 1:26 am, Benjamin Gawert <bgaw...(a)gmx.de> wrote:
> >> * Dan Lenski:
> >>  > I'm in charge of a piece of scientific equipment that uses an old
> >>  > Pentium III computer (running Windows 2000) as its controller.  The
> >>  > computer is, needless to say, slow and annoying to use.
>
> >> Have you considered just upgrading it? Parts for a P3 should be
> >> obtainable for almost free today. Or get a better P3 computer.
>
> > Yeah, I mean it's not THAT expensive to do a full upgrade.  We can go
> > from P3 clunker to screaming Core 2 Quad beast with 4gb of DDR3 RAM
> > and dual NICs and 500gb hard drive in about $750.  Also, did I mention
> > the horrifically noisy PSU and ugly case which we'd be replacing as
> > well? :-)
>
> >>  > We'd like to upgrade to a newer computer, but here's the catch: it
> >>  > needs to have 3 PCI slots and 1 ISA slot (yech) to interface with the
> >>  > instrument.  The equipment manufacturer is okay with us replacing the
> >>  > computer, by the way.  (Replacing the ISA interface card would cost
> >>  > $10,000, so that's pretty much out of the question for us.)
>
> >>  > So does anyone know where I can get a modern motherboard with ISA
> >>  > slots?
>
> >> Yes. Kontron (www.kontron.com) does make several ISA-based solutions.
> >> However, they are not cheap.
>
> >> Sadly you didn't provide more details as to what card you want to use.
> >> Maybe it is cheaper (and in the longer term more cost effecive) to just
> >> replace the ISA card and use a standard mainboard without ISA.
>
> > It's one of these awful custom cards, seemingly unique to the
> > instrument manufacturer.  It is basically a glorified GPIO port, as
> > far as I can tell... but it's *their* glorified GPIO port, and no one
> > else knows exactly how it works.  They want $10,000 to replace the ISA
> > card with a newer PCI version.  Yes, $10,000.  Scientific equipment
> > manufacturers are ridiculous, especially when they choose to use
> > custom cards.  Thankfully, that's becoming less prevalent as standard
> > serial/USB/Firewire interfaces are increasingly provided.  Either that
> > or GPIB... which is basically a glorified serial port with an
> > extremely bulky cable, costing "only" hundreds of dollars for the
> > interface card.
>
> > </rant>
>
> > Here's another oddball product: ISA to USB.  http://www.arstech.com/item--usb2isa.html
> > Any guesses about possible compatibility pitfalls with such a beast?
> > Thanks for the advice so far!
>
> > Dan
>
> According to this, the ISA bus can run at 6MHz or 8MHz, and
> at 16 bits wide, that would make the maximum transfer rate about
> 16MB/sec. USB2 can manage about 30MB/sec transfer rate to a
> hard drive, so it sounds like for large block transfers, there
> is enough bandwidth. If you needed to "peek" and "poke"
> individual registers on your ISA card, that might be slower
> with the USB concept, because each peek or poke could take
> separate USB packets.
>
> http://en.wikipedia.org/wiki/Industry_Standard_Architecture
>
> http://upload.wikimedia.org/wikipedia/commons/0/0b/ISA_Bus_pins.png
>
> I notice arstech also mentions powering, and I notice in the pinout
> for the ISA connector, there is a pin available for -5V. On modern ATX
> power supplies, -5V has been removed, and you'd find one pin on
> the ATX connector that is not being used. That would be where the
> -5V used to be. Now, there isn't a strong reason to be using
> that rail. It might be convenient for something like wiring up ECL
> chips. Or perhaps some of the really old DRAM technologies. So that might
> be an item to check as well.
>
> Considering your scientific card costs $10000, you might want to
> start your project, by using a throwaway ISA card for testing, and get
> that running with any adapter first. For example, I have an old Soundblaster
> of some kind, in my first computer, and I think that might be the
> only ISA card I've got. If I could get that working, then I'd
> switch over to the real card and give it a try.
>
> Obviously, if the bridge is from USB to ISA, the environment it is
> used in, is going to have to understand USB. I don't know what
> shape official USB support is in for Win98 or ME.
>
>     "Windows 98, ME, 2000, XP, Server 2003, Vista, Server 2008"
>
> One other thing to consider, is the software on the old machine.
> Can the software that drives the ISA device, operate on a more
> modern operating system ? Can the new computer operate with an
> older OS ? You have a lot of mixing and matching to do.
>
> This is my current motherboard. I run a Core2 dual core processor
> in it (E4700). I've actually installed Win98SE on this motherboard,
> using an IDE hard drive. I used a five year old video card that has
> Win98 drivers. The video card was AGP type, and this board has
> an AGP slot. You'll find more modern boards, where the
> Device Manager would have things in it that might not be
> recognized. When Win98SE runs on my computer, only one core of
> the processor works (since Win98 doesn't support multiple cores),
> but I still found the box to have snappy performance. The install
> was more of a joke than anything, but I was surprised how easy it
> was to do. (I had to stop the install half way through, to make
> a tweak to prevent the 512MB memory limit from getting in the way.)
> A lot of other motherboards, would not have allowed this to work.
> About the fastest processor that runs in this board now, is an
> E7500 (the charts don't mention support for E7600, but logically
> it should work as well). This board does not support FSB1333 processors,
> so an E8600 would be out of the question.
>
> http://www.newegg.com/Product/Product.aspx?Item=N82E16813157115
>
> I suspect you have a lot of mixing and matching to do in any case,
> so there are bound to be some less than ideal choices you have to
> make, to get it all to work.
>
>     Paul

fwiw win98 supports usb perfectly with 3rd party nusb3.3 drivers

More directly, I'm wondering whether the slowness may be due to
something other than hardware. IME it nearly always is.


NT
From: NT on 26 Aug 2009 16:02
On Aug 25, 2:51 pm, daytripper <day_tri...(a)NOSPAMyahoo.com> wrote:
> On Tue, 25 Aug 2009 04:26:41 -0400, Paul <nos...(a)needed.com> wrote:
>
> [...]>I notice arstech also mentions powering, and I notice in the pinout
> >for the ISA connector, there is a pin available for -5V. On modern ATX
> >power supplies, -5V has been removed, and you'd find one pin on
> >the ATX connector that is not being used. That would be where the
> >-5V used to be. Now, there isn't a strong reason to be using
> >that rail. It might be convenient for something like wiring up ECL
> >chips. Or perhaps some of the really old DRAM technologies. So that might
> >be an item to check as well.
>
> [...]
>
> Back in the day, the -5V was often used in UARTs. If the OP's custom card is
> *that* old and relies on serial communication devices that are not single,
> positive input voltage, that could be a problem requiring resolution...


.... not too hard to provide the -5v from a regulated wallwart though,
if necessary


NT
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4
Prev: Smart wristband
Next: No multicore grahics from Intel, Intel withdraws Larrabee!