[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Arrrrrrrgggg!
I can look into it. I'd really be afraid of not being able to
keep all nine bits in the cable in sync. 4Mhz is a fast rate
to push down a long parallel cable. My guess is that going over
1Mhz will not be easy
Your interface sounds like the spec for the standard printer port
maybe only an inverter or driver is required? There are
control lines in the port. It would be easy to hook up your
Block Done to the port's Device Ready bit. If you do this
then no software needs to be written as there are already
general purpose bi directional parallel port drivers.
The only question is if 4Mhz is to fast. It is is
then you'd have to go with some other interface.
Tom Droege wrote:
>
> Chris and all,
>
> I am not planning on building a "new" card. But there may be something
> even easier than what I am planning to do, so if there is I am interested.
>
> When you say "start scan" over the serial IF to the Mark IV, it shortly
> starts sending parallel bytes at an average rate of .4 MBytes per
> second. There are 8 data lines and a strobe. The strobe goes from high to
> low. There are also a couple of control lines. The Mark IV always puts
> out 16 MBytes.
>
> Seems to me that any fast parallel port could handle this.
>
> The control lines are not really needed. Once the computer starts to see
> data coming in on the parallel port, it knows to expect exactly 16
> MBytes. So the "Block Done" signal is not really needed. Since the
> computer sends the "start Scan" it knows when to look for data on the
> parallel port.
>
> So here would be my preferred way to go:
>
> 1) Use an existing parallel port that is byte wide and needs a negative
> strobe.
>
> 2) Like one only with inverters, level shifters, line drivers etc.. to
> meet an existing standard. Note that I have no way to send control
> characters from the Mark IV. So the interface needs to be able to just
> field a string of bytes.
>
> 3) If I have to go to a serial link, steal an existing design. This is the
> present plan. My neighbor has developed such a design and can modify it to
> my needs relatively easily. This not only give me a design but a neighbor
> who is an expert to make it work.
>
> Note that I did not consider 1) from the beginning because I did not want
> to be involved in system software. Once I give up on that, 1) should be
> attempted.
>
> I notice that there is a parallel port on all present computers. How fast
> is it. Can you just drive bytes into it when the computer is expecting them?
>
> Hmmmm! Here is a nice project for you Chris. Just make an adaptor that
> connects the byte output DB-25 cable of the Mark IV into a computer
> parallel port. You have the hardware and the drawings. Is this
> possible? It just might be the simplest possible solution. As I say, I
> did not want to do this before since computers were a lot slower (20x) when
> I started this project and I did not want to get involved in system software.
>
> Tom Droege
>
> At 11:23 AM 8/14/01 -0700, you wrote:
> >Tom Droege wrote:
> > >
> >
> > > Meanwhile, I will hope that the memory cards that are now working will keep
> > > working until we can get a new system in place.
> > >
> > > The PCI card would have a limited FIFO and some control program that would
> > > write what it receives directly to memory. Bill has already done a similar
> > > project at 1000 times the bandwidth we need so it should be possible.
> > >
> > > Software writers might consider how such a card affects their designs.
> >
> >Tom,
> >
> >Please _don't build_ a new card. There are plenty of standard interfaces
> >that meet your needs. You just need to pick one that can handle your pixel
> >rate at 16 bits per pixel.
> >
> >I'd pick an older standard that has been around for years. I think the best
> >thing to use is bit synchronous serial over a differential cable.
> >For example RS-422 or RS-530. Data rates can go up to 10Mbps with very
> >long cables. Also there is software to drive these cards for most OSes.
> >I have even seen RS-422 PC cards that fit in notebook computers. (Using
> >a notebook to control a Mark IV camera would be great.)
> >
> >On the Mark IV you'd need to add a high speed synchronous chip.
> >You could look at using the 16850 UART which has a 128 byte FIFO
> >built in. There are also PCI cards with 16850 chips in them.
> >These will do 1.5 Mbps.
> >
> >Zilog makes some nicer parts like the Z16C32 which will do
> >up to 20Mbps and take care of CRC codes in the hardware.
> >http://www.zilog.com/products/partdetails.asp?id=Z16C32
> >
> >Here is a serial PCI that uses this Zilog chip.
> >http://www.ieci.com.au/products/comm/pci_route56.shtml
> >I thhink this is exactly what is needed but 16850 based
> >products will cost less. For example
> >http://www.avlab.com.tw/product/p001.htm
> >
> >There are also multi-port rs-422 cards with 2, 4 or more
> >serial ports per card. This would allow one PC to control
> >multiple Mark IV cameras
> >
> >THe other way to go would be to put a small computer between the
> >Mark IV and the PC. By small I mean a PC104 sized or even one
> >built on a SIMM form factor. This small computer would have both
> >a fast parallel port and some standard link such as Ethernet.
> >The control PC would then see Ethernet.
> >
> >As for the effect of any of this on the PC's control software
> >I don't see a problem. If the software runs on Linux or Windows,
> >either way the I/O is done by operating system level device drivers
> >which "hide" the hardware from the control software.
> >
> >--
> > Chris Albertson
> > chrisalbertson90278@yahoo.com
> > Redondo Beach, California
> > home: 310-376-1029
> > cell: 310-990-7550
--
--
Chris Albertson
chrisalbertson90278@yahoo.com
Redondo Beach, California
home: 310-376-1029
cell: 310-990-7550