LPTCAP

Parallel print capture system

By Kris Heidenstrom (k@heidenstrom.gen.nz)

Release 1c, 02 April 2000

Introduction

LPTCAP is a hardware and software solution which allows an IBM-compatible PC running MS-DOS to capture print data sent by another PC, or any device with a Centronics parallel printer interface.

It is mainly intended to allow PC-based capturing of data from "closed" systems such as obsolete word processors which use incompatible disks or disk formats, so that the old equipment can be thrown out, or microcontroller-based embedded systems, where the only way to access the data externally is via the unit's printer port, to eliminate paper printouts and/or add processing capabilities.

The LPTCAP system is not a product you can buy. The software is provided in this package, and may be used in any personal or commercial application. The hardware (the LPTCAP adapter) can be constructed by anyone with basic practical electronic construction skills, and is documented in full in this package. Parts cost is about US $20 excluding the case and cables. The author does not sell the adapter as a kit or product.

If you are browsing this document online, you may have noticed that it has no links to other parts of the package. This is because I would prefer you to get the full LPTCAP archive, which is available as http://home.clear.net.nz/pages/kheidens/lptcap/lptcap01.zip.

The LPTCAP system consists of a hardware adapter box called the LPTCAP adapter, and associated support software. The LPTCAP adapter box has two connectors:

CN1 is physically and electrically similar to the Centronics connector on a parallel printer, and connects to the device which is sending printer data (known as the Sender) via the cable that would normally be used to connect the Sender to a Centronics parallel printer.

Data from the Sender is latched in the adapter, and can be captured by a PC (called the Capture PC) which is connected to CN2 (male 25-pin D-sub). The connection between CN2 and the Capture PC is through a short pin-to-pin cable which connects CN2 to a parallel printer port on the Capture PC. The Capture PC controls and interrogates the LPTCAP adapter through this cable.

The basic functions performed by the LPTCAP adapter are:

The support software runs on the Capture PC. It consists of the LPTCAP software module which is included in the LPTCAP package in source (assembly language) and object form, to be linked into other programs. This module provides LPTCAP auto-detection, testing and capturing functions which are easy to use.

A small exerciser program called LPTEST (which uses the LPTCAP software module) is also included. This program performs tests and simple capturing functions. This program is included in source (Borland C) and executable forms, and can be used as a starting point for other programs, or for basic capturing functions.

Capturing can be done in polled mode, or interrupt-driven mode if the Capture PC's parallel port has interrupt capability. Interrupt-driven mode may be more convenient and/or efficient than polled mode in some applications, but is not required to prevent data overrun (as with serial communication, for example), because the Centronics electrical protocol requires an explicit handshake on each character. The LPTCAP adapter provides the handshaking signals and thus prevents data overrun.

Boring Legal Stuff

The LPTCAP hardware and software was designed and developed by the author, Kris Heidenstrom. The LPTCAP package, including the hardware design, support software and documentation, is Copyright © 1996-2000 by K. Heidenstrom.

My philosophy is that this package should be freely available to anyone who wants to use it, and that no-one should deny this to anyone else. By "freely available" I mean easily available and with no specific charge apart from any normal data or disk charges.

You can use the hardware design and the support software freely in any personal or commercial application. You can distribute the software, in its complete and unmodified form only, as long as the author is identified, the package is identified as free and no specific charge (other than data charges or disk duplication) is made for it.

You are welcome to build LPTCAP adapters (the hardware part of the package) and sell them, but you must clearly identify the designer, and state that the hardware design and support software are freely available and that the potential buyer could get the design details for free and build the device him/herself.

You may not distribute the LPTCAP package in an altered or incomplete form. Please distribute only the original archive file.

If you generate a freely distributable derived version of the design or code such as a DOS protected-mode version or a version for a different operating system, please include a note that your code was derived from mine, and please consider sending me a copy of your package.

I provide no warranty that the LPTCAP code and the LPTCAP design will perform to any standard or at all. This is because I cannot control the environment in which you use the software and the hardware design. Any use of the LPTCAP system is entirely at your own risk. If this is not acceptable to you, then do not use the LPTCAP package.

You should also check the Caveats section for notes on the limitations of the design.

I have tested the design thoroughly and I believe that it is sound. If you have a problem with the LPTCAP design, please let me know and I will try to help. You can contact me at:

    Internet:     k@heidenstrom.gen.nz(preferred)
Snail:Kris Heidenstrom
c/- P.O Box 27-103
Wellington
New Zealand
Phone:Work: +64 4 385-6611
Home: +64 4 475-7437
(We are 12 hours ahead of GMT.)

Please feel free to contact me via email, or to send a postcard to the snail mail address above if you find the LPTCAP system useful.

Versions

Version 1 - released 20 December 1997

Original release

Version 1a - released 28 March 1998

Correction to setup procedure - in the second stage when measuring across R1 and adjusting for half resistance, I had said to measure between U2 pin 8 and U1 pin 11 - the correct pins are U2 pin 8 and U1 pin 5. Thanks to Jay Pennell (jpennell@seanet.com) of Penntronics for finding this error.

Version 1b - released 28 July 1998

Correction to layout illustration to add connection points to CN1 pin 10 and CN2 pin 11, and change to the construction procedure to state that there are thirteen wires from the board to each connector including grounds. Thanks to Peter Scales (peter@sleepy.demon.co.uk) for reporting these errors.

Version 1c - released 02 April 2000

Changed email address from kheidens@clear.net.nz to k@heidenstrom.gen.nz.

LPTCAP Adapter Hardware

The hardware design is provided in this package as a schematic diagram (circuit diagram) and a stripboard layout diagram.

LPTCAP Schematic diagram, LPTCAP-S.GIF, 33K

The schematic diagram (above) is contained in the file LPTCAP-S.GIF. As you can see, it is quite large, and should be printed if possible. Note that on the schematic, U1 is shown as six discrete components, named U1A to U1F.

The stripboard layout is given in the file LPTCAP-L.GIF. This view shows the adapter board looking down from above. See the Construction section for more information. Important note: There were two errors in the layout drawing which are fixed in release 1b - connections to CN1 pin 10 and CN2 pin 11 were not marked on previous versions.

Parts List

The following parts list for the LPTCAP adapter excludes parallel cables.

   Quantity     Reference(s)     Description
3C1-3 0.1µF 50V ceramic or monolithic capacitor
1C4 10µF 10V tantalum capacitor
2C5,6 470pF 50V ceramic capacitor
3C7-9 100pF 50V ceramic capacitor
1CN1 Female 36-pin Centronics connector, chassis mount
1CN2 Male 25-pin D-sub connector, chassis mount
2R1,2 50 kilohm preset potentiometer (trimpot)
2RSIL1,2Single-in-line resistor network, 9-pin, 8-element, 100 kilohm
1JP1 3-pin jumper block or single-pole changeover switch (see below)
1U1 74HC14 integrated circuit - hex Schmitt trigger inverter
1U2 74HC74 integrated circuit - dual D-type flip-flop
1U3 74HC165 integrated circuit - serial-output shift register
1 Jumper shunting block for JP1
2 14-pin DIP IC socket, dual-wipe (for U1 and U2)
1 16-pin DIP IC socket, dual-wipe (for U3)
1 Piece of stripboard - 25 tracks by 20 holes
1 Plastic or metal case
2 Screws, nuts and lockwashers for mounting CN1
2 Hexagonal tapped-head screws with nuts and lockwashers for CN2
5m Hookup wire (stranded, insulated)
2m Solder

   Notes:    
  • All of the above parts should be readily available from electronic hobbyist shops such as Radio Shack or the local equivalent.
  • R1 and R2 are 50 kilohm preset potentiometers (trimpots). These are screwdriver-settable devices which are adjusted in the adjustment procedure. Single-turn or ten-turn types are both suitable.
  • JP1 is listed as a three-pin jumper strip with a jumper shunt. This jumper selects the interrupt polarity, which may be different for different Capture PCs. If the LPTCAP adapter will regularly be used with several different Capture PCs, it may be worthwhile connecting a single-pole changeover (SPCO) switch such as a small slide switch or toggle switch, instead of the jumper, so that the interrupt polarity can be selected without opening the adapter case.
  • U1-3 must be 74HCnn types. 74nn (TTL), 74LSnn, 74ACTnn and 74HCTnn types are NOT suitable. Typical manufacturers are Motorola, Philips, National Semiconductor, Harris and Texas Instruments. Use DIP (dual in-line package) devices, not surface-mount (SOIC) devices (surface mount devices do not go well with stripboard).
  • U1-3 are static sensitive devices. Keep them in their anti-static tubes until you are ready to use them.

Construction

The LPTCAP adapter circuit is constructed on a rectangular piece of stripboard. A printed circuit board could be designed, but these are uneconomical except in large quantities, and would still require wire links (unless a double-sided PCB was used, but these are even more expensive). The assembled board is connected to connectors CN1 and CN2 (and a switch connected to the JP1 position, if a switch is used) with hookup wire, and the unit is mounted in a plastic or metal case. Cutouts and holes for the connectors must be made in the case. This is easier with a plastic case than with a metal case.

In addition to files and a drill, you will need pliers, side cutters, soldering iron (or a "soddering iron" if you're American :-) and a small screwdriver. During testing and adjustment, a multimeter (analogue or digital) is required.

LPTCAP Layout diagram, LPTCAP-L.GIF, 21K

The stripboard layout drawing (above) is contained in the file LPTCAP-L.GIF. This view is drawn looking down onto the adapter board from above. The copper strips are shown in grey, with gaps indicating cuts where appropriate. Coloured numbers refer to pins on CN1 (red numbers) and CN2 (blue numbers). Important note: There were two errors in the layout drawing which are fixed in release 1b - connections to CN1 pin 10 and CN2 pin 11 were not marked on previous versions.

Here is the full detailed construction procedure.

  1. Drill and file holes in the case, to suit CN1 and CN2 (and the switch, if you are using one). If possible, arrange things so that you will be able to wire the connectors to the stripboard before mounting them on the case. If you mount the connectors against the outside of the case, you won't be able to push them through the apertures from the inside after they have been wired up, so you would have to bring the wires through the connector apertures when wiring the connectors up to the stripboard. If the case is in two halves, you can make the apertures open to the edge of one half, to avoid this problem.
  2. Cut the stripboard to size and make the 31 cuts in the copper strips, using a quarter-inch drill bit or a proper stripboard cutting tool. Just cut deep enough to remove the copper, don't drill through the board. The cut positions are indicated by gaps in the grey strips in the drawing.
  3. Install the three IC sockets - these will act as position references. Don't insert the ICs yet.
  4. Install the link wires. These are shown as black lines joining two points. Many of these are straight horizontal links, and uninsulated wire can be used for these (if you use stranded wire, twist it tight). For the other links, use insulated stranded hookup wire.
  5. Install capacitors C1-9. C1-3 are not shown in the drawing - solder these on the underside of the board, under U1, U2 and U3 respectively, across the VCC and GND pins (the corner pins - pins 14 and 7 for U1 and U2, and pins 16 and 8 for U3). Use lengths of insulation (taken from hookup wire) on the leads, so they cannot touch other pins or tracks.
    C4 is polarised. If the polarity is not marked on the body, the longer lead is normally positive. C5-9 are not polarised.
  6. Install R1, R2 and JP1. If you are using a switch instead of a jumper to set the IRQ polarity, solder three lengths of hookup wire instead of JP1, and take these to the IRQ polarity switch. Make sure that the middle pin of the switch connects to the middle pin of the JP1 marking on the drawing.
  7. Solder resistor networks RSIL1 and RSIL2 directly to connectors CN1 and CN2 respectively. The resistor networks will have a dot or similar mark at the pin 1 end. Pin 1 does not connect to the connector, but all other pins do (they connect to pins 2 to 9 of the connector).
  8. Make the connections from CN1 and CN2 to the stripboard, using lengths of insulated stranded hookup wire. Connection points for these wires are indicated by red and blue numbers on the drawing. Red indicates a wire to a pin on CN1, and blue indicates a wire to a pin on CN2 (as mentioned in the drawing).
    Ground (common) on CN1 and CN2 must be connected to the stripboard positions marked "GND" in red and blue. For CN1, connect together pins 16, 17, 19-30 and 33 and the shield, and wire to the stripboard. For CN2, connect together pins 18-25 and the shield, and wire to the stripboard.
    There are a total of 13 wires from CN1 to the stripboard, and 13 wires from CN2 to the stripboard, including the "GND" wires.
  9. Using hookup wire, connect pin 1 of RSIL1 and pin 1 of RSIL2 to the stripboard positions marked "RSIL1" and "RSIL2".
  10. Connect the following pins of CN1 and CN2 together using hookup wire. These signals go directly between the two connectors, not via the stripboard.
    • CN1 pin 13 to CN2 pin 6
    • CN1 pin 14 to CN2 pin 14
    • CN1 pin 31 to CN2 pin 16
    • CN1 pin 32 to CN2 pin 5
    • CN1 pin 36 to CN2 pin 17
  11. Insert the ICs.
  12. Place the stripboard in the case (if the case is metal, use corrugated cardboard or similar, glued to the case, to insulate it from the bottom of the board) and mount the connectors.

Adapter Adjustment And Testing Procedure

The adjustment procedure sets the timing of the Acknowledge pulse and the BUSY latch reset using trimpots R1 and R2 on the LPTCAP adapter. You will need a digital or analogue multimeter.

The LPTEST program (included in the LPTCAP package) is required, to generate timed signals during adjustment. Use the "generate adjustment signal" option and enter a blank test duration (the adjustment signal will stop when you press the Ctrl key).

Disconnect the Sender cable from CN1. Connect the multimeter negative lead to the common rail of the LPTCAP adapter (the connector shields are connected to the common rail). Start the adjustment signal using the LPTEST program. With the adjustment signal running, measure the voltage on VCC (pin 14 of U1). This voltage should be stable, and between 3V and 5V. Note the actual voltage measured.

With the adjustment test still running, measure the voltage on pin 9 of U2 and adjust trimpot R2 for a voltage of 14% of the VCC voltage measured earlier. For example if VCC is 4.0V, adjust R2 until you measure 0.56V on U2 pin 9.

With the adjustment test still running, measure the voltage on pin 3 of U2. Rotate trimpot R1 to each end of its travel. At one end, the voltage should be nearly zero, and at the other end the voltage should be about 0.5V (the exact value is not important). The end where voltage is present is the minimum resistance end. Turn R1 to this end, and rotate R1 slowly towards the other end. The voltage should drop steadily, until a point is reached where the voltage suddenly jumps to zero (or nearly zero). Leave R1 at this position.

Unplug CN2 to power down the adapter, and remove U1 from its socket. Connect the multimeter between pin 5 of U1's socket and pin 8 of U2 (i.e. across R1) and note the reading, then rotate R1 towards the minimum resistance end until its resistance is half of this value.

Replace U1, reconnect CN2, run the LPTEST diagnostic test and check that the tests pass.

Optionally, temporarily connect a 68 ohm resistor from VCC to ground, measure the VCC rail (should be around 2-3 volts) and repeat the diagnostic test - it should pass.

A data transfer test requires a second PC, to act as the Sender. Run the PRNFILE.COM program (included in the package) on both machines to generate a one megabyte file called PRNFILE.OUT containing pseudo-random data. Run the capture software and start capturing to a file, then on the Sender, enter the command

COPY /B PRNFILE.OUT PRN

You can measure the time taken to transfer the file and calculate the throughput if you wish. When the file has been transferred, use a file-compare program such as DOS's COMP to compare the captured file with PRNFILE.OUT on the Capture PC. If the files do not match, see the Caveats section for suggestions.

Signal Naming Conventions

LPTCAP documentation and comments in source code use signal names as marked on the LPTCAP schematic. The Capture PC's parallel port is not used for printing, so most of its standard signal names are not meaningful. Therefore, the signal lines on the Capture PC's parallel port are assigned names (such as -DACK and SDI) according to their function when used with the LPTCAP adapter. Signals from the Sender are named using the standard convention, since these signals are used for their normal purposes.

The four control signals (-INIT, -SELECT, -AUTOFEED and -STROBE) are connected directly from the Sender's port to the same signals on the Capture PC, so they are named identically on both CN1 and CN2.

LPTCAP Adapter Technical Description

The adapter is controlled and powered by signals from the Capture PC. When data lines from the Capture PC (pins 2-9 of CN2) are high, input protection diodes in U1, U2 and U3 conduct and provide operating voltage for the circuit. This is why the default state for the Capture PC's data register is 11111111b. C4 acts as a reservoir for the supply rail and provides current for U1-3 when they switch (change state).

The LPTCAP adapter contains one 74HC14 hex Schmitt inverter chip (U1) which is used to invert various signals and clean up slow signal transitions from the parallel ports, one 74HC74 dual flip-flop (U2) which detects and latches data bytes from the Sender and generates Acknowledge pulses, and one 74HC165 parallel/serial-input, serial-output shift register (U3) which loads and holds each captured data byte and passes it serially to the Capture PC.

These devices must be 74HC types. Other types with compatible pinouts, such as 74LS and 74HCT, should not be used.

The default state of the adapter is with U2A in the reset state, i.e. U2 pin 6 high and the +-DRDY signal line in the inactive state (whether this is high or low depends on the setting of the IRQ Polarity switch or jumper). In this state, assuming that -RESETB is inactive (high), the adapter is ready to accept a data byte from the Sender machine.

When the Sender wishes to send a byte, it asserts the data value on D0-7 and then briefly pulses -STROBE active (low). This causes the BUSY latch (U2A) to flip to the set state, giving a BUSY indication back to the Sender, and switches the shift register (U3) from transparent load mode to shift mode, latching the data byte into the shift register in the process. The +-DRDY line goes active, indicating to the Capture PC that a data byte has been captured and generating a parallel port interrupt on the Capture PC if the interrupt is enabled and the port is interrupt capable.

The LPTCAP adapter will remain in the "busy" state until the Capture PC detects and acknowledges the new data. In this state, the Sender cannot send a new data byte, because BUSY is active and no -ACK pulse has been sent by the adapter.

When the Capture PC detects that new data is ready, via the +-DRDY line, either by polling the line or through the parallel port interrupt, it can read the data serially via the SDI and -SDI signals (-SDI is always the complement of SDI), using the -SCL signal to clock the data through U3. Initially when a data byte is captured, bit 0 is present on SDI. On every falling edge of -SCL from the Capture PC, the data shifts through the shift register by one bit, shifting new data in from the SDO line (which is used in the diagnostic test but not during normal operation) and presenting the next data bit of the captured byte on SDI and -SDI.

Once the Capture PC has read the data byte, it brings -DACK low, clocking a "1" into U2B and starting a pulse on the -ACK line back to the Sender (and to the Capture PC on the -ACKP line). Approximately 5 microseconds (set by R1 and C5) after the start of this pulse, U2A's clock goes high and U2A is reset (assuming that -STROBE from the sender is high, which it should be). Approximately 10 microseconds (set by R2 and C6) after the start of the -ACK pulse, U2B resets, ending the -ACK pulse.

The -RESETB signal is provided to allow the Capture PC to reset U2A without generating an -ACK pulse, and is used only during initialisation of the LPTCAP adapter. The 74HC74 device responds in a defined way when its set and reset inputs are both active (low) - both the Q and Q-bar outputs are forced high. With U2A, only the Q-bar output is used, so the reset input (from -RESETB) has priority over the set input (from -STROBE). Therefore, if -RESETB is active (low), U2A will be in a known state regardless of the state of -STROBE. This is important because the Capture PC cannot guarantee that -STROBE will be high (inactive) - it may be being driven active by the Sender. By forcing -STROBE low itself, the Capture PC can control U2A's Q-bar output via -RESETB.

The DDATA signal feeds the Data input of the acknowledge pulse flip-flop and should be held at logic "1" at all times. It has no practical purpose except to provide another source of supply current for the adapter.

The following diagram shows a typical data byte capture.

LPTCAP capture timing diagram, LPTCAP-T.GIF, 4K

Parallel Port Registers On The Capture PC

The tables in the following sections give the I/O assignment for the Capture PC's parallel port, with standard parallel port functions given in parentheses.

In these tables, the "!" symbol indicates that the PC's parallel port circuitry inverts the data between the register and the hardware signal, i.e. a bit value of "1" in the register corresponds to an electrically low signal and vice versa. The "-" symbol indicates that the signal is electrically active-low, i.e. it is active or "true" when it is electrically low.

Therefore a signal name which starts with "-" or "!" (but not both) is in its active state when the register bit is 0, and a signal name with both "!" and "-", or neither, is in its active state when the register bit is 1. The polarity of +-DRDY depends on the setting of the IRQ Polarity jumper (JP1) or switch on the LPTCAP adapter. See the Status register section for details.

Data Register
Data register: LPTBase+0, read/write, driven by software
    76543210
*.......   (D7)   DDATA   Should always be held at "1"
.*......   (D6)   -RESETB   Set to 0 to hold BUSY latch clear
..*.....   (D5)   -NOPAPER   Set to 0 to signal PAPER OUT to sender
...*....   (D4)   SELCTD   Set to 1 to signal SELECTED to sender
....*...   (D3)   -ERROR   Set to 0 to signal ERROR to sender
.....*..   (D2)   -DACK   Normally 1, 1-to-0 edge acknowledges data byte
......*.   (D1)   -SCL   Normally 1, 1-to-0 edge clocks shift register
.......*   (D0)   SDO   Drives serial data input of shift register
Status Register
Status register: LPTBase+1, read-only, driven by hardware, read by software
    76543210
*.......   (!BUSY)   !-ACKP   Pulses to 1 when we acknowledge data
.*......   (-ACK)   +-DRDY   New data ready (see explanatory, below)
..*.....   (NOPAPER)   SDI   Carries serial data from shift register
...*....   (SELCTD)   -SDI   Carries inverted shift register data
....*...   (-ERROR)   LOOPIN   Driven from bit 0 of data register
.....***       Undefined
The +-DRDY signal on status register bit 6 indicates whether new data is ready. Status register bit 6 (the -ACK input of the parallel port) is the input which is fed to the IRQ line when the buffer is enabled. On some older machines, the IRQ is triggered on the falling edge of this signal line, but on most machines, the IRQ is triggered on the rising edge. To support both port types, the LPTCAP adapter has an IRQ Polarity switch or jumper (JP1), which feeds either a true signal or an inverted signal onto this line. The meaning of +-DRDY for each position of this switch or jumper is:

    JP1 positionMeaning of bit 6 of status register
    Negative IRQ     Normally 1; 0 means new data is ready;
    Positive IRQ     Normally 0; 1 means new data is ready.

Control Register
Control register: LPTBase+2, read/write, driven by hardware and software
    76543210
***.....     Not used in this application
...*....     Interrupt Enable (1 = enable)
....*...   !-SELECT   Will be 1 if Sender is asserting -SELECT
.....*..   -INITIALIZE   Will be 0 if Sender is asserting -INITIALIZE
......*.   !-AUTOFEED   Will be 1 if Sender is asserting -AUTOFEED
.......*   !-STROBE   Will be 1 if Sender is asserting -STROBE
Can also be driven to 1 to trigger a Strobe.

The control register is used in a slightly unusual way. The electrical signal lines driven by this port are all open-collector with pullup resistors. This allows these signal lines to be used as inputs if the drivers are set to the electrical high state. In this state, the pullup resistors pull the lines up to electrical high, but the lines can be pulled to electrical low by the Sender. Reading the control register yields the actual line states, so this port can be used as an input port in this application.

Because control register bits 0, 1 and 3 have inversion between the register bits and the signal lines, a value of xxxx0100 binary must be written to the control register to set the drivers to electrical high. This will appear as xxxx0100 binary when reading the control register. If the Sender drives any of those signals to electrical low, the corresponding bit(s) in the control register will change to the opposite state.

This software can cause a data strobe to occur explicitly by setting and then clearing bit 0 of the control register. When this bit is set, the port output forces the -STROBE signal line to electrical low, causing a data strobe on the adapter even if the Sender is not driving -STROBE low.

LPTCAP Software Module

The LPTCAP software module is included in this package in
source (LPTCAP.ASM) and object forms. It provides a set of functions which can be called from a C or assembly-language program, which interface with the LPTCAP adapter and provide standard auto-detection, diagnostic and data capture capabilities.

Assembly Time Settings

There are two assembly-time settings for the LPTCAP software module - memory model and slow-timing.These are controlled by constants which are defined on the assembler command line. These constants are:

    TINY     Selects tiny memory model
SMALL     Selects small memory model
COMPACT     Selects compact memory model
MEDIUM     Selects medium memory model
LARGE     Selects large memory model
HUGE     Selects huge memory model
SLOW     Selects slow timing
VERYSLOW     Selects very slow timing

These constants are defined on the assembler command-line with the "/dname" option. For example, to specify large model and slow timing, the command line to the assembler (TASM or MASM) should include:

/dLARGE /dSLOW

A memory model should be specified. If no memory model is specified, during assembly the module issues a warning message and the module defaults to SMALL model.

The SLOW and VERYSLOW options may be useful during development. The VERYSLOW setting should not be used when generating production code. If neither SLOW nor VERYSLOW are specified, the module will be assembled for standard timing.

Refer to the comments in LPTCAP.ASM for more detailed information on the memory model support and the slow timing settings.

Functions Provided

The following functions are provided in the LPTCAP software module:

    lptcap_version()     Returns version numbers and assembly settings
lptcap_port() **     Detects LPTCAP adapter(s) on parallel ports
lptcap_test() **     Diagnostics - logic and data loopthrough tests
lptcap_test_fail_data()     Returns details of a diagnostic check failure
lptcap_adjust()     Generates timed signals for adapter adjustment
lptcap_get_cont()     Returns states of control signals from Sender
lptcap_set_stat()     Sets states of printer status signals to Sender
lptcap_send_ack()     Sends an acknowledgement to the Sender
lptcap_poll_new()     Tests whether a new character has been captured
lptcap_next_char()     Returns the next captured character
lptcap_wait_char()     Waits (with or without timeout) for a character
lptcap_autodetect_irq()     Auto-detects the IRQ number of the LPTCAP port
lptcap_intmode_install()     Installs interrupt-driven capture mode
lptcap_intmode_uninstall()     Uninstalls interrupt-driven capture mode

** The functions lptcap_port() and lptcap_test() set up global variables which are used by other functions, and must be called and return a successful result before other functions are used.

The code in the LPTCAP module manipulates the interrupt flag extensively. All functions return with the interrupt flag unmodified. Functions do not lock interrupts out for more than about 50 microseconds at one time (any operations which take longer than this are split into blocks, with the interrupt flag restored to its initial between the blocks) except for lptcap_adjust() which locks interrupts out for about 5 milliseconds at a time. Some functions explicitly enable interrupts temporarily during their execution.

Individual Function Descriptions

lptcap_version()
unsigned int lptcap_version(int request)

This function returns version-related information, according to the "request" parameter. Values for the "request" parameter are:

    0     LPTCAP_REQ_MAJVER     Returns major version number
1     LPTCAP_REQ_MINVER     Returns minor or sub-version number
2     LPTCAP_REQ_MODVER     Returns modification or sub-sub-version number
3     LPTCAP_REQ_FILREV     Returns file revision number
4     LPTCAP_REQ_DATEL     Returns low word of YYYYMMDD value
5     LPTCAP_REQ_DATEH     Returns high word of YYYYMMDD value
6     LPTCAP_REQ_CONDS     Returns a bitstring indicating conditional assembly-time parameters (see below)

For a request parameter of LPTCAP_REQ_CONDS, the function returns an unsigned integer which is bit-allocated as follows:

    FEDCBA9876543210
************....Reserved for future use
............*...Module was made with VERYSLOW option
.............*..Module was made with SLOW option
..............*.Module made for far data (0 = near)
...............*Module made for far code (0 = near)

lptcap_port()
int lptcap_port(int findmode)

This function detects and returns the LPT port number of the LPTCAP port. It accepts a "findmode" parameter which specifies the search type desired:

    0     LPTCAP_FIND_SAME     Return previously-found LPTCAP port
1     LPTCAP_FIND_FIRST     Look for LPTCAP port starting from LPT1
2     LPTCAP_FIND_NEXT     Find next LPTCAP port after last found port

This function is able to find more than one LPTCAP port, though no other functions support capturing from more than one LPTCAP adapter simultaneously.

This function returns the LPTx number of the port where the LPTCAP adapter was found, in the range 1 to 4, or 0 if it could not find any port (or any more ports, if findmode was LPTCAP_FIND_NEXT).

Once a parallel port with an LPTCAP adapter is found, this function initialises the parallel port's control register and outgoing status lines, and the port becomes the LPTCAP port used by the other functions in this module.

This function sets up several port-related global variables within this module which are used by other functions. IT MUST BE CALLED AND RETURN A SUCCESSFUL RESULT BEFORE OTHER FUNCTIONS IN THIS MODULE WILL WORK.

This function should not be called while LPTCAP is capturing data as it will disturb the capturing process.

This function detects the LPTCAP adapter via the LOOPOUT and LOOPIN signals. It does not perform any diagnostic check on the port, so it should not disturb any printers or other standard parallel peripherals that may be connected to ports as the ports are checked.

Ports are located through the BIOS parallel port base address table in low memory. This means that LPTCAP will respect any parallel port reassignments that have been made during or after boot-up via the BIOS parallel port table (this does not include reassignments from parallel to serial ports made via MODE), and will support parallel ports at non-standard addresses provided that they have been detected by the BIOS or by program executed during or after boot-up, and that they are compatible or backward-compatible with standard PC parallel ports in hardware and software.

lptcap_test()
int lptcap_test(unsigned int ntests)

This function performs diagnostic tests on the LPTCAP adapter. IT MUST BE CALLED AND RETURN A SUCCESSFUL RESULT BEFORE OTHER FUNCTIONS IN THIS MODULE WILL WORK. The test consists of a basic presence detection, followed by a test loop consisting of a logic test and a data loopthrough test, which is repeated according to the ntests parameter.

The basic presence detection uses the loopback signal to check that the LPTCAP adapter is present, and determines the setting of the IRQ Polarity switch or jumper (JP1) on the LPTCAP adapter.

The logic test checks the operation of the data capture flip-flop and the data acknowledge flip-flop. The data loopthrough test checks the shift register by shifting data values of 0, 0xFF, 0x55 and 0xAA through it.

The test takes approximately 0.4 milliseconds to 2 milliseconds (depending on the speed of the Capture PC) multiplied by the ntests value (longer if the SLOW or VERYSLOW parameters were used when LPTCAP was assembled).

If a test fails, the hardware is set to its default state and lptcap_test() stops testing and returns the error number.

If this function is called while interrupt-driven receive mode is enabled (via lptcap_intmode_install()) it will return LPTCAP_ADAPTER_INTMODE and will not perform any logic or data tests.

This function should not be called while LPTCAP is capturing data as it will disturb the capturing process. It should not be called while the Sender is trying to send data. This means that any capture program should perform its initial setup and tests BEFORE the Sender is instructed to send data.

Return values are:

    0     LPTCAP_ADAPTER_PRESENT     Adapter present and working normally
1     LPTCAP_ADAPTER_NO_PORT     No parallel port has been selected
2     LPTCAP_ADAPTER_MISSING     Adapter is not attached
3     LPTCAP_ADAPTER_INTMODE     System is running in interrupt-driven mode
4     LPTCAP_ADAPTER_NO_DRDY     Adapter failed to activate +-DRDY signal
5     LPTCAP_ADAPTER_ACKSTUK     Adapter -ACKP is stuck in active state
6     LPTCAP_ADAPTER_NO_DACK     Adapter failed to activate -ACKP signal
7     LPTCAP_ADAPTER_ACKLONG     Adapter -ACKP pulse too long
8     LPTCAP_ADAPTER_BSYSTUK     Adapter indicated BUSY active after -ACK pulse
9     LPTCAP_ADAPTER_STRSTUK     -STROBE line stuck low (see notes below)
10     LPTCAP_ADAPTER_SDISERR     Adapter SDI and -SDI states inconsistent
11     LPTCAP_ADAPTER_DATAERR     Data loopthrough error (see notes below)

All non-zero values indicate a problem with the LPTCAP adapter or the parallel port of the Capture PC, except return codes LPTCAP_ADAPTER_BSYSTUK and LPTCAP_ADAPTER_STRSTUK, which may indicate that the Sender is already trying to send data when lptcap_test() is called, or possibly that the Sender is faulty or is connected but powered off.

For a return value of LPTCAP_DATAERR, the byte values that were sent and received in the data loopthrough test that failed are available through the lptcap_test_fail_data() function and may be displayed in the error message to possibly help a user to diagnose the problem.

This function temporarily uses timer channel 2 for timeout checking. This may interfere with any background sound generation or timing that is being done by other parts of the program using timer channel 2.

lptcap_test_fail_data()
unsigned int lptcap_test_fail_data(void)

This function returns an unsigned integer which is comprised of the "sent" value (in the low byte) and the "received" value (in the high byte) from the data loopthrough test that failed, when lptcap_test() returns an error code of LPTCAP_ADAPTER_DATAERR. This may be displayed as part of the failure report message if desired.

lptcap_adjust()
void lptcap_adjust(unsigned int ticks)

This function generates a pattern of accurately timed signals which are used during adjustment of the LPTCAP adapter. See the Adjustment section for details.

The "ticks" parameter specifies the duration of the test. One timer tick is approximately 55 milliseconds, so one second is about 18 ticks and one minute is about 1092 ticks. If a zero "ticks" parameter is supplied, lptcap_adjust() will generate the test signal until the user presses the Ctrl key.

This function may be used even if lptcap_test() has not been called, or when lptcap_test() returned an error code. This is because it may be necessary to adjust the LPTCAP adapter before lptcap_test() will report success.

This function should not be called while LPTCAP is capturing data as it will disturb the capturing process. If it is called when the LPTCAP system is operating in interrupt-driven receive mode (via lptcap_intmode_install()) it will return without generating any adjustment signal.

This function temporarily uses timer channel 2 for timekeeping. This may interfere with any background sound generation or timing that is being done by other parts of the program using timer channel 2.

lptcap_get_cont()
unsigned int lptcap_get_cont(void)

This function returns the states of the control lines provided by the Sender at the time the function is called. The return value is bit-allocated:

    76543210
****...*     Reserved for future use
....*...   SELECT   signal from Sender; 1 = active
.....*..   INITIALIZE   signal from Sender; 1 = active
......*.   AUTOFEED   signal from Sender; 1 = active

lptcap_set_stat()
void lptcap_set_stat(unsigned int stat_word)

This function sets the printer status signals which are returned to the Sender, according to the stat_word parameter which is bit-allocated:

    76543210
**...***   Reserved for future use; should be 0
..*.....   PAPER OUT   signal to Sender; 1 = paper out
...*....   NOTSELCTD   signal to Sender; 1 = not selected
....*...   ERROR   signal to Sender; 1 = error

The default and initial values for these signals are the "0" values, which are "normal" (i.e. "no-error") values. This function need not be used in normal circumstances, since the outgoing signals are set to their default values when lptcap_port() is called.

lptcap_send_ack()
void lptcap_send_ack(void)

This function sends an acknowledgement signal to the Sender, consisting of -ACK going low, BUSY going inactive, then -ACK rising again. This may be required if the Sender does not receive an acknowledgement for some reason and gets stuck waiting for it.

lptcap_poll_new()
int lptcap_poll_new(void)

This function tests whether a new character has been captured and is available. It returns 0 if not, or 1 if so. It does not return the character value, and does not remove the character from the buffer.

This function behaves differently in polled and interrupt-driven modes. In polled mode, it examines the +-DRDY signal from the LPTCAP adapter directly to see whether a new character has been captured. In interrupt-driven mode, it examines the receive buffer pointers to determine whether there is any unread data in the buffer. Also, in interrupt-driven mode, it checks and corrects a condition where a character has been received but the interrupt was lost. This condition could occur when this software is used with badly behaved TSRs. It does not result in loss of data.

lptcap_next_char()
int lptcap_next_char(void)

This function returns the next character which has been captured and is available, or returns an indication that there is no data available.

If no character is available, this function returns -1 immediately.

If a character is available, this function acknowledges the character and removes it from the buffer, and returns the character value, which will be in the range 0 to 255.

In polled mode, this function examines the LPTCAP adapter directly, and in interrupt-driven mode, it examines the buffer pointers and checks for missed interrupts - see notes on lptcap_poll_new() for details.

lptcap_wait_char()
int lptcap_wait_char(unsigned int timeout_ticks)

This function waits for a character to be received, and returns the character value. If the timeout_ticks parameter is non-zero, it will implement a timeout check, and if no character is received within the timeout period, it returns a value of -1. The timeout value is in units of one timer tick, or about 55 milliseconds. A zero timeout value will give an infinite timeout. Because timer ticks are not synchronised to the function call, the actual time before a timeout is reported may be up to one tick longer than the specified value.

In polled mode, this function examines the LPTCAP adapter directly, and in interrupt-driven mode, it examines the buffer pointers and checks for missed interrupts - see notes on lptcap_poll_new() for details.

lptcap_autodetect_irq()
int lptcap_autodetect_irq(unsigned int irqmask)

This function attempts to auto-detect the IRQ (interrupt request) number associated with the LPTCAP port. It returns a value from 2 to 15 indicating the IRQ number, or one of these error indications:

    0     LPTCAP_IRQ_NONE     No associated IRQ detected
-1     LPTCAP_IRQ_INVERTED     IRQ polarity was incorrect
-2     LPTCAP_IRQ_SEVERAL     More than one IRQ detected

If lptcap_autodetect_irq() returns LPTCAP_IRQ_NONE or LPTCAP_IRQ_SEVERAL, interrupt-driven capture (if enabled) will probably fail to work properly. A return value of LPTCAP_IRQ_INVERTED means that the IRQ Polarity switch or jumper on the LPTCAP adapter must be changed to the other position. When this has been done, lptcap_test() must be called to detect the new state of this switch/jumper, then lptcap_autodetect_irq() may be called again.

Any other value indicates a valid and correct IRQ number that is associated with the LPTCAP port. On PC and PC/XT machines the IRQ number will be in the range 2-7. On the AT and later machines the IRQ will be in the range 3-15 excluding 8 and 13.

If interrupt-driven capturing is not required, there is no need to call this function at all.

This function does not enable the interrupt (except briefly during its testing), and does not install the LPTCAP interrupt-driven receiver.

The irqmask parameter is a bitstring which specifies which IRQs are to be tested. Most interrupt-capable parallel ports are wired to IRQ7 or IRQ5, and some may be jumper-selectable to other IRQs. Bits 0 to 15 of the irqmask parameter correspond to IRQs 0 to 15, and a 1-bit enables testing of that IRQ. Therefore a typical value for irqmask is 0x00A0, allowing IRQs 5 and 7 to be tested, though any value, including 0xFFFF, can also be used.

Some IRQs are not available on the slot bus, and these are not tested by this function even if enabled in irqmask. On PC and PC/XT machines, IRQs 2-7 are available on the slot bus (IRQ0 and IRQ1 are used on the motherboard). On AT and later machines, IRQs 3-7, 9-12, 14 and 15 are available on the slot bus (IRQ0, IRQ1, IRQ8 and IRQ13 are used on the motherboard and IRQ2 doesn't exist).

This function will probably fail on MicroChannel machines (IBM PS/2). It will also probably fail if used in a DOS compatibility box within a multi-process operating system such as Windows 3.x, Windows 95, Windows NT, OS/2 or Linux.

Many soundcards use IRQ7. If the sound card has not been initialised, its IRQ line driver may be in the high-impedance state and the parallel port may be able to use the IRQ line. However, software that uses the sound card (such as Windows or game software) may leave the buffer enabled, and this may prevent the parallel port IRQ from working after these programs have been run.

This function should not be called while LPTCAP is capturing data as it will disturb the capturing process.

lptcap_intmode_install()
int lptcap_intmode_install(unsigned int irqnum, lptcap_bufdescriptor * bdp)

This function installs and enables interrupt-driven reception for the LPTCAP system. In interrupt-driven mode, reception of a character by the LPTCAP adapter causes an interrupt, which causes the CPU to read the character and place it in a circular buffer or "queue" for later processing. Interrupt-driven mode is not necessary in order to keep up with the data rate or to avoid data loss, but it may make a program easier to write, and may improve the maximum data transfer speed if the capture software is doing a lot of other things as well as capturing the data.

The function returns 1 if successful, or 0 if there is a problem with the irqnum parameter and interrupt mode has not been installed.

The irqnum parameter specifies the IRQ (interrupt request) number to be used. The range of acceptable values depends on the type of machine on which the software is running. On a PC or PC/XT, allowed irqnum values are 2, 3, 4, 5, 6, 7. On an AT or later machine, allowed irqnum values are 3, 4, 5, 6, 7, 9, 10, 11, 12, 14, 15.

The LPTCAP port IRQ number can be determined by calling lptcap_autodetect_irq() but some parallel ports do not have interrupt capability, in which case interrupt-driven receive mode will not work.

This function does not attempt to confirm that the specified interrupt request line is associated with the LPTCAP port, nor that the port is interrupt capable, nor that the interrupt line is not in use by some other peripheral.

The bdp parameter must point to an lptcap_bufdescriptor structure (see the typedef in LPTCAP.H), which is formatted as follows:

	volatile unsigned int tail_index; /* Tail (index into buffer) */
	unsigned int head_index;	/* Head (index into buffer) */
	unsigned int buf_size;		/* Size of buffer in bytes */
	unsigned char * buf_base;	/* Pointer to base of buffer */

Memory for the buffer descriptor and the buffer area must be allocated by the calling program. The caller must provide the buf_base and buf_size parameters in the descriptor structure; this function provides initial values for tail_index and head_index (both 0).

The maximum buffer size is limited to 65535 bytes, even in the "huge" memory model, because the buf_size is a word (16-bit). The maximum number of data bytes that can be held in the buffer is one less than the buffer size.

The calling program may remove data from the buffer manually, by manipulating the head_index value, or it may use the lptcap_poll_new(), lptcap_next_char() and lptcap_wait_char() functions.

Enabling interrupt-driven receive mode causes the behaviour of the character input functions (lptcap_poll_new(), lptcap_next_char() and lptcap_wait_char()) to change - they use the interrupt-driven receive buffer, instead of going direct to the LPTCAP adapter hardware.

The interrupt line associated with the LPTCAP parallel port can be found by lptcap_autodetect_irq(). If an incorrect interrupt number is chosen, data capture may appear to work (because software will capture data in polled mode if the interrupt handler is not activated properly), but throughput will be reduced.

Using lptcap_port() to change to a different port while interrupt-driven capturing is installed will cause strange behaviour, so don't do it!

If a program uses interrupt-driven capture mode, steps must be taken to ensure that lptcap_intmode_uninstall() is called before the program terminates. This may involve a critical error handler and/or a Ctrl-Break handler, since either of these conditions can usually terminate a DOS application without allowing it to clean up. If this occurs, the interrupt vector is left pointing to code in the LPTCAP module which could be overwritten when another program is run, which would cause a system crash.

lptcap_intmode_uninstall()
void lptcap_intmode_uninstall(void)

This function uninstalls interrupt-driven receive mode, disables the parallel port interrupt and returns the LPTCAP system to polled mode. Any data in the circular buffer when this function is called is lost.

lptcap_intmode_uninstall() uses DOS function 0x25 (set interrupt vector) and therefore must not be called when DOS is in an unstable state (for example, from within an interrupt handler).

This function must be called prior to termination of the program. If not, the interrupt handler will be left installed and this can cause a system crash.

Application Information

Caveats

The LPTCAP software is written for real-mode DOS, and will not work correctly (probably not at all) under multi-tasking, hardware-abstracting operating systems such as Linux, OS/2 or any variety of Windows. Plain old DOS (version 2.0 or later) is required. (The test program, LPTEST, may require DOS 3.x.)

The LPTCAP software module cannot be used with a DOS extender. It is written for real-mode only. It could be adapted to work in protected mode.

The LPTCAP system will run on any member of the IBM PC family, including the old PC and PC/XT, provided that it has a parallel port which is compatible (in hardware and software) with a standard parallel port or a PS/2 bidirectional parallel port. It may or may not work with ECP and EPP ports. If the parallel port is integrated into the motherboard, the BIOS Setup may provide a facility to set the parallel port's operating mode.

The LPTCAP adapter has been tested with two parallel port dongles (hardware copy-protection devices) from different manufacturers and worked with both, but it is possible that some dongles will interfere with LPTCAP.

Do not use the LPTCAP software on a machine which is running in slow mode (i.e. a machine which has a turbo switch which is set to "slow"). This may cause lptcap_test() to occasionally report bogus LPTCAP_ADAPTER_NO_DACK failures.

The LPTCAP hardware adapter is not heavily protected against electrical interference or damage due to common-mode or differential-mode surges being induced into the cables. It is very important to follow these guidelines:

During testing, the prototype adapter performed well and was able to transfer one-megabyte pseudo-random data files without error. However, the Centronics interface is susceptible to electrical interference, so if data accuracy is important, you should perform some testing by sending a known data file through the adapter and trying to induce interference in the following ways:

If capturing is unreliable with certain Senders or Capture PCs, or certain combinations, try adding a 10 kilohm resistor between pins 4 and 14 of U2.

Performance

The author's testing has used various PCs as the Sender but no other type of device. The design is based on the Centronics interface documentation in the manuals for various parallel printers, so any unit that works with a normal Centronics parallel printer should work with the LPTCAP adapter.

The LPTCAP system is not optimised for high throughput. The hardware constraints of the standard parallel port severely limit the maximum data transfer rate, and software changes would make little difference to the throughput.

Using the SLOW or VERYSLOW options when assembling the LPTCAP module will affect throughput. SLOW will roughly halve throughput, and VERYSLOW will reduce throughput by a much greater amount (depending on the speed of the Capture PC).

Throughput is limited by both the Sender and the Capture PC. The Capture PC generally does more work than the Sender and needs to be fast if throughput is significant. Prototype throughput test results were:

    SenderCapture PCThroughput
100MHz Pentium     100 MHz Pentium     16100 bytes/sec
8 MHz XT75 MHz Pentium6000 bytes/sec
75 MHz Pentium8 MHz XT1400 bytes/sec

These tests were done with standard timing (i.e. SLOW and VERYSLOW not used).

Only the first line, using fast machines for both Sender and Capture PC, has a data transfer rate higher than 115200 bps (the maximum rate available on a standard PC serial port).

Throughput in interrupt mode is generally slightly lower than polled mode, unless the software on the Capture PC is performing other tasks in addition to capturing data.

Summary

The LPTCAP system is an inexpensive and practical way to get data out of closed or proprietary systems, or capture real-time data from embedded controllers.

The hardware can be built by anyone with some practical electronic construction experience. The support software can be linked into user-written programs, or the test program can be used for one-off data capturing.

The design has some limitations (generally these are inherent in the Centronics protocol), which are discussed in the Caveats section.

Anyone is free to use the LPTCAP system for their own personal or business purposes, provided that they don't deny others the same opportunity. The design details and support software are freely available to any user.

I hope you find the LPTCAP system useful.

Kris Heidenstrom

02 April 2000