Vector Connection

Download a picture of a magazine article (german)
(1300x800) Magazine Article

A multi-I/O, Zorro-II board for Amiga 2000/3000/4000. It features up to 2 parallel and up to 4 serial ports at speeds up to 115200 baud. The older software will run under AmigaDOS 1.3. However, the latest software version requires at least AmigaDOS 2.0. The back cover of the board has two serial ports. Two additional slot covers, each with a serial and a parallel port can be installed if more ports are required. Their cables will then have to be plugged onto the board.


REVIEW

        Due to the quite different concepts in the old 1.3 compatible
software version and the newer one which requires AmigaDOS 2.0, I will
describe them separately.

        The old software requires some changes in your user-startup file.
Most utilities are CLI based programs. Only the preferences editor is GUI
based.  The installation was incomplete when I did it (approximately 2 years
ago).  The software is stable.

        The latest software gets installed very nicely.  There is no more
need to know and use CLI utilities.  The software fits perfectly in the OS.
Earlier versions of the new software were buggy, but the last version I have
tested (2.2) did not produce any problems any more.

        Of those two serial ports on the slot cover at the back of the card,
one is standard 25-pin, and all additional ones are small 9-pin ports. No
adapters are supplied with the card.

        After installing the software, your application programs can use the
new ports by using the new devices and units.  The standard serial and/or
parallel port can also be redirected to the Vector board, so even software
that was not designed to use custom devices can now use the ports of the
board.

        The current version of the Vector Connections allows for up to
115200 baud on each port.  Each two ports share one serial driver and one
4-byte FIFO buffer.  As a result, the baud rates have some restrictions.
According to the manual, two ports using the same driver cannot be used at
115200 and 76800 at the same time.  They can, however, be used at the same
speed without problems.  I myself found that they _would_ work at 115200 and
76800 at the same time, but not at 115200 and 57600.

        I used up to three serial ports for about two years, and I never had
any problems that were due to the board.  I had tried a parallel port once,
and it worked, but I did not use it regularly.


DOCUMENTATION

        The board comes with a printed manual.  It contains all you need to
know to install and operate the board, and I suppose that it is
understandable for beginners and experts alike.

        Since I bought my board in Germany, I got a German manual. I do not
know if an English version is available.


LIKES

        I like the easy way to get four additional serial high speed ports. I
like the new software, too.


DISLIKES AND SUGGESTIONS

        I do not like the fact that they do not include adapters for the
small 9-pin ports. They should be included.  A 16-byte FIFO buffer would be
very nice, and after all, this is the reason why, after having used the
board for two years, I sold it and bought myself a GVP I/O Extender.


COMPARISON TO OTHER SIMILAR PRODUCTS

        I had used a A2232 serial card before.  Even though the Vector
Connection does not provide seven additional ports, like the A2232 does, I
would always prefer a board that allows up to 115200 baud on four ports.  If
you want to use V.34 modems, you have no choice anyway.

        I am currently using a GVP I/O Extender.  I decided to get one
because I hoped that the 16-byte FIFO buffer would leave more CPU available
during downloads.  Even though it really does, the difference is smaller
than I had thought.  IMHO, you get more for your money if you buy a Vector
Connection, as one GVP I/O Extender will give you only two additional serial
ports.


BUGS

        I didn't find any.


VENDOR SUPPORT

        I did not have any problems with my installation, so I did not
contact the vendor directly.  I called the BBS once, asking for help, but
never got an answer.  Fortunately, I was able to solve the problem on my own.

        The new software, however, is freely distributable, which IMO is a
good product support.


WARRANTY

        In Germany, there is a minimum warranty of six months required by
the law.  HK-Computer does not extend this warranty.


CONCLUSIONS

        I found the Vector Connection to be a good product, sold at a fair
price.  I would recommend it if I was asked by a friend.

        I rate this product 4 stars out of 5.


COPYRIGHT NOTICE

        Copyright 1995 Metin Savignano.
        This review is freely distributable, but not for profit.

Here is an FAQ about the UARTs used on this card

Motorola MC68681: MC68HC681/2681 - DUART
========================================
 

The MC68HC681 dual universal asynchronous receiver/transmitter
(DUART) is part of the M68000 Family of peripherals and directly
interfaces to the MC68000 processor via an asynchronous bus
structure. The MC68HC681 consists of these major sections:
internal control logic, timing logic, interrupt control logic,
bidirectional 24-bit data bus buffer, two independent communication
channels (A and B), 6-bit parallel input port, 8-bit parallel
output port.

MC68681 Features:
  o M68000 Bus Compatible
  o 2 Independent Full-Duplex Asynchronous Receiver/Transmitter Channels
  o On-Chip Crystal Oscillator
  o TTL Compatible
  o Single +5V Power Supply




MC68681/2681 Frequently Asked Questions
---------------------------------------

-Question: 
On the MC2681 or MC68681, assuming that the write cycle is
chip-select controlled, what is the minimum data hold time after
chip select is negated on a write cycle? 

Answer: 
The answer is derived from a combination of two specifications.
The first specification defines that there must be a setup time
of the W signal relative to CS (Tch). The second specification
defines the data hold time relative to W going high (Tdh).
Since Tch = 0, and Tdh = 10 ns, adding them yields 10 ns for a
data hold time relative to the CS.


-Question: 
The specification for the MC2681/MC68681 specifies a crystal series
resistance less than 180 ohms. Does that still hold? Can a crystal
of series resistance of 200 ohms suffice? 

Answer: 
The specification that defines that a crystal must have a series
resistance of 180 ohms still applies. It is recommended that any
crystal with a larger series resistance not be used. Below is a
list of manufacturers of crystal manufacturers that meet our
specifications 
Saronix, Palo Alto, CA (800) 227-8974
inside Ca. (800) 422-3355, part #NMP037 

Bomar Crystal, Middlesex, NJ (201) 356-7787 

Crystek Corporation, Fort Myers, FL (813) 936-2109 

Electro-Dynamics Corp., Shawnee Mission, KS (913) 262-2500 

Allied Electronics, Fort Worth, TX (214) 265-9341 part # 994-0345 

US Crystals, Fort Worth, TX (800) 433-7140


-Question: 
What value of shunt resistors must be used with the MC2681/MC68681? 

Answer: 
The MC2681/MC68681 manual is incorrect. A shunt resistor must be
added across the X1/CLK and X2 pins. This shunt resistor is needed
to help provide a 50% duty cycle. The recommended value of a shunt
resistor is 10-20 Mohm. 


-Question: 
Which MC68681 signals require a pull-up resistor? 

Answer: 
The following signals of the MC68681 require pull-up resistors:
DTACK, IRQ. The OP7, OP6, OP5, OP4, OP3 signals may require pull-up
resistors if they are used as open-drain, active-low outputs.


-Question: 
What happens when I read a reserved register on the MC68681/MC2681? 

Answer: 
When a read of a reserved register (locations $02 or $0A) is
attempted, the DUART is forced into a diagnostic mode. This mode
is used to test the baud rate generator circuitry. When the DUART
enters this mode, it outputs baud frequencies on the general purpose
output lines which are multiples of the frequencies listed in the
baud rate table. If the transmitter is enabled, the DUART may also
transmit at these frequencies, regardless of the value selected in
the clock select register (CSR). A read to the reserved registers
may occur accidentally by a monitor program that performs a
read/verify cycle after write cycles, by negating the write strobe
prior to the chip select signal at the end of a write bus cycle,
or possibly by asserting the write strobe after asserting CS* at
the beginning of a write bus cycle. To avoid the last two
situations, the chip select signal to the DUART should be qualified
with DS* from the processor.


-Question: 
Describe the receiver FIFO of the MC68681/MC2681. 

Answer: 
The DUART has a three-byte receiver FIFO that acts more like a
circular queue. It has both a head pointer and a tail pointer.
The head pointer is controlled by a read operation and is
incremented to the next buffer location whenever a read of the
receiver occurs. The tail pointer is incremented whenever a new
character that has been assembled in the shift register is
transferred to the receive holding register. After an external
reset or the issuance of a reset receiver command, the head and
tail pointers point to the same location in the FIFO. The contents
of the FIFO are not flushed when a reset receiver command is issued.
It takes three consecutive reads of the FIFO to move the head
pointer in a circle until it gets back to its original location.
Note that the head pointer can inadvertently be incremented if a
monitor program that performs a read/verify cycle after a write
cycle is in use. 

The receiver ready bit (RxRDY) in either the interrupt status
register (ISR) or the status register (SRx) should be polled before
reading the receiver. If the bit is set, the receiver should be read.
The status register should again be read to determine the state of
the RxRDY bit. If it is set again, the receiver should be read again.
This should continue until the RxRDY bit is clear. If a read of the
receiver is performed when the RxRDY bit is clear, the pointers will
be incremented beyond the current valid data. 


-Question: 
How can I detect the end of break"? 

Answer: 
In order to detect a break condition, the DUART receiver
continuously samples the receive data input (RxD). When it
senses a low for the start bit and the number of programmed
bit times, it loads a zero character into the receive FIFO.
If no stop bit is detected, the receiver samples beyond the
character frame for one more bit time. If this bit is low, a
framing error has occurred. Having the framing error bit in
the status register set and a zero character in the receive
FIFO forces the received break bit to be set in the ISR and
the SRx. 

The delta break bits in the ISR must be monitored on order to
detect an end of break. The processor should not read the zero
character in the receive FIFO nor clear the receive break bit
in the SRx until the break is completely over. Once a break
has been detected, the processor should issue a reset break
change interrupt command, which resets the delta break bit in
the ISR. The delta break bit will be set again whenever RxD
transitions, indicating that the break condition is over. When
this happens, the processor should issue a break change 
nterrupt command, a reset error status command, and then read
and discard the zero character in the receive FIFO.


-Question: 
What are the consequences of changing the DUART configuration
without first disabling the receiver and transmitter? 

Answer: 
Whenever you decide to write to the mode registers (MR1x and
MR2x), the clock select registers (CSRx), or the auxiliary
control register (ACR), the receiver and transmitter must
first be disabled. If the mode registers change while
serialization is still active, the transmission may be
restarted under the new configuration. If serialization
is complete and the transmit buffer is empty when the mode
registers change, one of two things can happen; either TxD
can go into the space condition for a short period of time
or the transmitter ready bit (TxRDY) in the ISR may set and
then reset. The above events are independent of the values
written to the mode registers. In fact, rewriting the current
values to the mode registers can produce the same results. 

A more serious consequence may occur if you write to the CSRx
or bit 7 of the ACR while the transmitter and receiver clocks
are running. If the clocks are changed without first disabling
the receiver and transmitter, clipped (shortened) clock pulses
may appear during the change from one frequency to the next.
These pulses may cause the receiver and transmitter to lock up. 

The recommended (and best) way to disable the transmitter and
receiver whenever you plan to change the configuration is to
issue a software reset command ($20 and $30 to the CRx). Not
only does this disable the receiver and transmitter, it also
places the DUART in a known state.


-Question: 
What programming sequence is necessary to operate in the
Multidrop/Wake-up mode? 

Answer: 
When the DUART is operating in the Multidrop mode, the
transmitter sends data with the last bit of each character
flagged as either address or data. Mode register 1, bit 2,
determines whether the character being sent is an address or
a data character. Thus, if you want to send an address
character followed immediately by data characters, you
must write to the mode register. Writing to the mode
register without first resetting the receiver and
transmitter could result in the incorrect transmission
of data. Therefore, the following programming sequence
should be observed when operating in the Multidrop/Wake-up mode: 

  1. Verify that the TxEMT bit is set to guarantee that
     the transmitter is not currently sending a character. 
  2. Reset the MR pointers, the receiver and the transmitter
     via the command register. 
  3. Load MR1 with the previously written data, insuring
     that MR1[2] = 1. Then, enable the receiver and transmitter. 
  4. Load the address character into the transmitter holding
     register (THR). 
  5. Wait until the TxEMT bit is set (character sent). 
  6. Reset the MR pointers, the receiver and the transmitter
     via the Command Register. 
  7. Load MR1 with the previous data, this time insuring that
     MR1[2] = 0. Then, enable the receiver and transmitter. 
  8. Load the data characters into the THR, until the message
     is complete. To send data to a different address, repeat
     steps 1-8. 

Whenever a DUART in a secondary station recognizes an address
character, it will set its RxRDY bit. The CPU must then
immediately read the receiver to see if the accumulated
address matches the station's address. If it does, the CPU
must set RxEN in either CRA or CRB so that the message can
be received. At higher baud rate frequencies, it might be
difficult to perform the compare and the setting of the RxEN
bit before the data begins to arrive. To alleviate this
problem, you could enable the RxEN bit as soon as you receive
the address character and then either reset the receiver if a
match has not occurred or read the data characters from the
receive FIFO if the address matches.


-Question: 
What are my options for driving the X1/CLK and X3 pins? 

Answer: 
The DUART is very sensitive to its clock. The clock circuitry
can be driven by either a crystal or a TTL-level signal. When
using a TTL-level clock, the clock signal should be connected
to X1/CLK and X2 should be grounded. If you do use a TTL-level
clock to drive X1/CLK, you must guarantee a minimum high
voltage of 4 Volts and a minimum high and low clock pulse
width of 100 ns. The X1/CLK driver does not have to be an
open collector. 

The areas of most concern when using a crystal to drive the
clock circuitry are the capacitance of C1 and C2, the duty cycle,
and the rise and fall times of the clock signal. Signetics
recommends that the values of both capacitors be around 5 pF
to insure proper charging during the power-on cycle. Our data
sheet says that C1 should be between 10 and 15 pF and C2 should
be between 0 and 5 pF. There are no known problems using these
values. 

Ideally, the duty cycle of the clock signal should be as close
to 50% as possible. However, many DUARTs show a 60-40 duty
cycle when hooked up to the crystal. To force a 50-50 duty
cycle, add a 100 Kohm or greater resistor across X1/CLK and X2. 

The rise and fall time problem may never occur assuming you
have a good crystal to start with. Below is a list of vendors
for 3.6864 MHz crystals that meet our specs. There are
probably many more vendors; these are just the ones we know of. 

Saronix, Palo Alto, CA (800) 227-8974
inside Ca. (800) 422-3355, part #NMP037 

Bomar Crystal, Middlesex, NJ (201) 356-7787 

Crystek Corporation, Fort Myers, FL (813) 936-2109 

Electro-Dynamics Corp., Shawnee Mission, KS (913) 262-2500 

Allied Electronics, Fort Worth, TX (214) 265-9341 part # 994-0345 

US Crystals, Fort Worth, TX (800) 433-7140 

Multiple DUARTs can be driven from the same crystal.
Tap off the clock from X1/CLK, buffer it through an inverter,
add a 1K pull-up resistor, and connect it to the other X1/CLK
inputs (remember to ground X2 on these devices).


-Question: 
What is the potential problem with receiver-controller
RTS* negation? 

Answer: 
When the receiver controls the negation of the RTS* output,
a one should be written to the appropriate output port pin
immediately after enabling the receiver. The receiver will
negate RTS* whenever the FIFO is full and the start bit of
a fourth character has been detected. Because of this, care
must be taken in choosing the transmitter at the other end.
Some transmitters, depending on the manufacturer, will stop
transmission after sending out the character in the shift
register while others stop after sending out the characters
in both the holding register and the shift register. The
latter transmitters will cause an overrun to occur in the
DUART receiver.


-Question: 
How long must the pulse be on the input port pins before it
is recognized internally? 

Answer: 
The state change detection circuitry requires two successive
samples of the new state before the delta change bits are set
in the input port change register (IPCR). This circuitry uses
a 38.4 KHz sampling clock (generated from the baud rate
generator). If the crystal frequency is 3.6864 MHz, then the
sampling period of the internal state machine will be 25
microseconds. If the transition of the input pin occurs at
the same time as the first sample pulse, then the new level
must be present for 25 microseconds. If the level change
occurs just after the sampling edge, the new level must be
present for at least 50 microseconds to guarantee recognition.


-Question: 
Does an interrupt acknowledge cycle clear the bits in the
interrupt status register (ISR) or do I have to clear them
in software? 

Answer: 
None of the bits in the ISR are cleared by an interrupt
acknowledge cycle. The input port change status bit (ISR_7)
is cleared when the processor reads the input port change
register (IPCR), the channel A and channel B change in break
bits (ISR_2 and _6) when the CPU issues a reset break change
interrupt command for the corresponding channel, the channel
A and B receiver ready or FIFO full bits (ISR_1 and _5) when
the CPU reads the receiver buffer for the associated channel,
the channel A and B transmitter ready bits (ISR_0 and _4)
whenever the processor loads a character into the appropriate
transmitter holding register, and the counter/timer ready bit
(ISR_3) by a stop counter command in both the counter and
timer modes.


-Question: 
When do the output ports go high after RESET is asserted? 

Answer: 
The output ports (OP0-OP7) are placed in the logic high state
about 80 to 90 ns after RESET asserts (asynchronous to the
X1/CLK input).

Thanks to Ber