Serial Ports as Slots
Volume Number: 1
Issue Number: 8
Column Tag: The Electrical Mac
"Macintosh Serial Ports" 
By Jeff Mitchell, President, Digital Solutions, MacTutor Contributing
The Macintosh has frequently been criticized for its lack of hardware expansion
capability. Apple's response has been that the two high-speed serial ports act as
'virtual slots' which give the user practically unlimited expansion capability.
Although the merits of this approach may be debatable, it is nevertheless our only
point of access to the internals of the Macintosh.
In this article, I will describe both the RS-232-C and the RS-422-A standards
and examine their compatibility. I will then examine in detail Apple's implementation
of RS-422 in the Macintosh serial ports and describe how to interface them with
standard RS-232 peripherals, along with precautions to be taken in order to maintain
data integrity and protect your equipment from damage.
RS-232-C
Electronic Industries Association (EIA) Standard RS-232-C, Interface Between
Data Terminal Equipment and Data Communication Equipment Employing Serial Binary
Data Interchange, released in August 1969, defines how data terminal equipment
(DTE) and data communication equipment (DCE) exchange information serially.
Originally, DTE was intended to be a dumb terminal, and DCE was intended to be a
modem. The large number of handshake lines were included to provide control over
what were entirely hardware driven devices, with no local intelligence. As
microprocessors became popular and devices became capable of implementing
handshaking protocols in software, many of the handshake lines became superfluous.
This, combined with the popularity of using RS-232 to connect all types of different
devices not even imagined at the time the standard was written, makes signal definition
somewhat subjective and is what causes us so much grief when trying to make
interconnecting cables.
EIA Standard RS-232-C defines electrical, mechanical, and signal
characteristics. Electrically, a mark, or binary one, is defined as less than 3 volts
with respect to signal ground, while a space, or binary zero, is greater than 3 volts.
The region between +3 volts is defined as the transition region, where the signal state
is not uniquely defined. The maximum open-circuit voltage is not to exceed +25 volts.
In addition, there are some restrictions on source and load impedances, load
capacitance, and slew rates.
Mechanically, the standard assigns signal definitions to 25 pins (including 3
unassigned pins), although no particular connector type is specified. However, the 25
pin subminiature-D type connector (DB-25) has emerged as a de-facto standard.
Cable length is recommended to be kept to less than 50 feet, although the load
capacitance of 2500 picofarads is really the limiting factor.
Signal characteristics are defined in some detail in the standard, although if not
applied to a strict terminal/modem configuration, they tend to become open to
interpretation. Data rates are restricted to a nominal 20,000 bits per second.
Standards notwithstanding, how is RS-232 generally implemented?
Electrically, most equipment uses MC1488 (or equivalent) drivers operating from
+12 volt supplies, and MC1489 (or equivalent) receivers operating from +5 volts.
The drivers accept a standard TTL level in and convert it to about +10 volts out,
inverting it in the process to meet the RS-232 mark/space requirements.
The receivers do not operate in strict accordance with RS-232, since their input
switching threshold is around +1 volt. Therefore these receivers will correctly
respond not only to RS-232 input levels, but TTL level inputs also. Like the drivers,
the output is inverted from the input.
RS-422-A, RS-449
EIA Standard RS-422-A, Electrical Characteristics of Balanced Voltage Digital
Interface Circuits, released in December 1978, defines the interchange of serial
binary signals between DTE and DCE, or in any point-to-point interconnection of
serial binary signals between digital equipment.
There are several significant differences between RS-232 and RS-422. Most
importantly, RS-422 is a balanced interface. Secondly, note that this standard defines
electrical characteristics only. EIA Standard RS-449 defines the mechanical and
signal characteristics and will be included in this discussion. A final, more subtle
point: notice that the description of the standard has been expanded to include
connection of any digital equipment, not only DCE and DTE.
The RS-422 specification defines a balanced interface. This means that instead of
a voltage on one signal line being referenced to a common ground, as in RS-232, the
signal is carried on two lines which are always in complementary states (see Fig.1).
The threshold criteria in a balanced system is the voltage difference between the two
signals, not the absolute value of either with respect to ground. The open circuit
voltage between the two outputs is not to exceed 6.0 volts, and cannot be less than 2.0
volts when driving a load of 100Ω + 1%. In addition, there are restrictions on offset
voltages, short circuit output current, and receiver specifications, among others.
Mechanically, RS-449 defines a 37 pin and a 9 pin interface, where the 9 pin
interface is defined as a secondary channel for the 37 pin interface. As with RS-232,
signals are given pin assignments. Unlike RS-232, standard connectors are specified.
In recognition of the de-facto DB-25 standard for RS-232, RS-449 defines DS-37
and DE-9 connectors (The D indicates subminiature-D type, the B, S, and E indicate
shell size, and the 25, 37, and 9 are number of pins. Note that a DB-9 would be a
DB-25 with 16 pins missing.). Maximum cable length using 24 AWG twisted pair
cable varies according to signalling rate, from 50 feet at 10 Mbits/sec to 4000 feet at
100 Kbits/sec or less.
Signal characteristics are described in RS-449, again with the emphasis on a
terminal/modem configuration. A table showing equivalent RS-232 signals is
provided in the standard.
The industry standards for RS-422 drivers and receivers are the Am26LS31
(or equivalent) and the Am26LS32 (or equivalent), respectively. Both of these
devices operate off a single +5 volt supply, and meet all the requirements set forth in
the EIA standard.
As defined, how compatible is RS-422 with RS-232? During development of the
RS-422 and RS-449 standards, the EIA gave close attention to an orderly transition
from existing RS-232 equipment. For systems designed strictly to the letter of the
standards, interoperability is assured. Realistically, most systems will be compatible
electrically, but signal definitions and mechanical connections are another matter.
Remember, however, that RS-422 is differential, and the only relation it has to
signal ground is that the midpoint of the output voltage swing must be within +3 volts
of ground and the input to either receiver input terminal cannot exceed +10 volts with
respect to ground. For systems operating off of standard supply voltages and using
industry standard parts, this is not a problem. In addition, RS-422 does not meet the
voltage swing requirement of RS-232, but generally suffices at the short cable lengths
found in RS-232 installations. The specifics of how to interface a single-ended to a
differential signal will be covered below.
Do you need to worry about damaging your Mac or peripherals while trying to
connect a new piece of equipment? RS-232 specifies that drivers be able to withstand
an open circuit, a short to any other signal line, or a short to ground without damage.
Receivers must be able to withstand +25 volts without damage. A short between any
two signal lines should not result in a current of more than 1/2 amp. RS-232,
therefore, is essentially bulletproof.
RS-422 leaves fail-safe operation more to the system designer, but specifies
that an output shorted to ground be limited to 150 mA and that receivers shall operate
correctly over a common mode voltage of +7 volts and with a maximum differential
voltage of 12 volts. These requirements result in a fail-safe system as long as it
remains an RS-422 interface.
For a mixed RS-232/RS-422 system, there is a possibility of device damage
since the maximum RS-232 voltage levels are +25 volts. The 26LS32 receivers have
been designed to withstand this input voltage range, meaning that they can be connected
to RS-232 drivers with relative impunity. RS-422 drivers can be damaged by
shorting them to voltages above 5.5 volts or below ground (26LS31), or to voltages
greater than +15 volts (26LS30). To summarize, RS-422 cannot damage RS-232,
but RS-232 can damage RS-422 under some conditions.
Macintosh Serial Ports
In order for their concept of the 'virtual slot' to be accepted as a viable
alternative to true hardware expandability, Apple had to forego the industry standard
RS-232 in favor of an interface that would support a much higher data rate, thus the
choice of RS-422. They also chose to compromise the standard somewhat in order to
make the output voltages more compatible with RS-232.
For output drivers, the Macintosh uses 26LS30's running from +5 volt
supplies, rather than the single +5 volt supply recommended by the manufacturer for
RS-422 operation. The result is a symmetrical open-circuit voltage swing of about 8
volts centered around ground. This drops to about 4 volts when loaded with 100Ω.
Centering the output signal around ground makes it RS-232 compatible, but the 8 volt
open circuit voltage exceeds the RS-422 restriction of a maximum of 6 volts.
The receivers are 26LS32's and meet all RS-422 requirements. Since the
Macintosh handshake line is a single-ended input, the inverting input on that receiver
is tied to ground, making that input RS-232 compatible. The TXD and RXD lines are
differential but are compatible with RS-232 levels when connected to the inverting
(-) terminals. The 26LS32 is designed so the non-inverting (+) input floats at about
+2.7 volts when open and the inverting input floats at about +1.8 volts. Therefore,
when an RS-232 signal is connected to the inverting input the receiver will switch
around +2.7 volts if the non-inverting input is left open, which is within the +3 volt
RS-232 spec.
Between the drivers/receivers and the connectors, the signal lines go through an
RC filter network. This is not part of the RS-422 specification, but is included to
meet FCC EMI requirements and to provide signal deglitching.
The Macintosh serial connector pinout is as follows:
PIN # FUNCTION
1 GND
2 +5 V
3 GND
4 TXD +
5 TXD -
6 +12 V
7 HSK
8 RXD +
9 RXD -
For RS-232 compatibility, use Gnd, TXD-, and RXD- (remember, RS-232
signals are inverted). The HSK line goes to both the CTS and the TRxC inputs on the
8530 chip in the Macintosh. Depending upon how the chip is configured by the serial
driver, this line may be used as a handshake line or as an external clock in order to
increase data transfer speeds. The +5 volt and +12 volt outputs come directly from
the power supply inside the Macintosh. Apple warns that the +5 volt line may turn
into an output handshake line at some future date and that it should not be used. They
also warn that the +12 volt line is for detecting power only. When connecting
peripherals or making cables, exercise caution, but you'll probably not be able to
damage any of the interface circuitry. Pay special attention to the power pins (2 &
6), however, since these are tied directly to the power supply and are not protected
from accidental shorting.
Electrically, like most of the Macintosh hardware, the serial ports are quite
simple. Some compromises to the RS-422 standard were made in Apple's
implementation, but generally speaking, these are not significant. The end result has
been to provide users with a much more flexible interface, able to communicate with
nearly all RS-232 peripherals in addition to supporting much higher data transfer
rates for applications that require it, such as external disk drives and networking.
Referenced by (2):