Tannery Drum Controller

PWP timeline
PWP main control panel

Fully populated 68070 controller in main panel.


BSD cabinet

Main computer panel: Industrial PC housing, 19" rack, swivel for keyboard, trackball mouse, water mixer panel above keyboard location.


BSD screen

X11 window

Introduction

Folklift truck lifting drum

The PWP Tannery was the second set of tannery drums that I automated. I subcontracted to Jendamark, as the project was too big for me to finance. The total job involved water mixing controls (from Honeywell for hot and cold mixing), variable speed drives to rotate or jog the large drums, and overall installation of the automation. My part was to develop the controller that was installed next to each drum, a communications controller to accept commands from the drum operators, interface to the Honeywell mixing valve controller, and communicate with the variable speed drive controllers. A single 68070 board was developed which had seven serial ports, 16 analog to digital converters to interface to weigh scales for drum loading and temperature probes, digital inputs and outputs, HMI interface, recipe storage and event logging.

Prior tannery projects included temperature control in “sweat rooms” via fans and mist spray, monitoring and data logging. The drums are basically large washing machines with round pegs inside, plus obstacles for stirring the sheep skins. A ring gear on the side attached to the six spoked spider driven by a 15kW SEW Movitrac 3000 variable speed drive and a Eurodrive R103 geared motor. The large black boxes on top of the red mounting stands contained load cells for weighing the drums (check loading, water, rinsed etc.)

Recipes were downloaded for drum speeds, rotation forwards, reverse, idle time, operator input, water loading at a particular temperature, plus event logging.

PWP tannery drums

There was a fair amount of time available for development, as the tannery was being constructed at a wool washing warehouse and would only be ready for installation two months later. According to the documentation written in 1993, six 680xx designs preceded this design. The real-time kernel, various libraries and debug functions were tested in factories in the Eastern Cape. The bound software listings for the PWP 68000 code was 40mm thick. The design files, Unix code and MS-DOS code was 30mm thick—fortunately mostly reused code.


PCB Layout

Drum PCB mosaic

My first 6U VME board was for a tannery drum controller. The standard 64 pin DIP 68000 chip certainly occupied a lot of real estate. The Signetics (Philips) 68070 included a 68000 core, a serial port, several peripherals housed in a 84 pin PLCC package. The layout was done in Tango PCB with a photograph taken of the screen. (The layout package still ran in a DOS window). The screen was not a flat panel; introducing a fair amount of distortion as well as a 50mm lens. The colours were inverted to get rid of the black background, then combined with a scan of the top & bottom laser plot, scan of top side of the physical PCB, and a scan of the bottom physical PCB. The outputs were scaled and cut into tiles to produce a mosaic. The bottom PCB layer scan was horizontally routed. The top scan was vertically routed. The bottom was also flipped horizontally as we are now looking from the top. The lineup is not exact, due to the distortion mentioned above of the screen photograph. The scanned PCB tiles were placed along diagonal lines running from bottom left to top right. The bottom layer has four tiles running from the second bottom left tile to the top right hand corner with a heat sink area for the 12V regulator. The top board tiles are two diagonal bottom left to top right with three tiles on the top diagonal, and two along the bottom diagonal.

The first batch of boards did not pick up power or ground as the board manufacturer decided to swell the thermal pads creating little islands. (That is why there was a blank board to scan after all the years after the original installations). Ten boards were made with six installed at the PWP tannery in Port Elizabeth. The tannery ran for several years and then closed down—possibly effluent problems; not sure as it was pretty close to town.

Drum Controller

Partially populated tannery drum controller

The partially populated controller board was used next to each drum. It did not need all the functions that the communications controller required. There were five drums.

Example LCD screens

Collage of LCD screens

The drum controller 4×40 character LCD screens were mirrored onto a X11 window on the BSD Unix based PC. The mixer control panel was installed below the large monitor, with its I/O driven out of the fully populated communications controller board. The letters are a bit small; top left is the overall menu, next across is the “Motor Menu”, followed by the “Valve Menu”. The “Status Screen” in the middle has running information during drum operation. All text based menus were written to the “ncurses” API, which allowed host testing before porting to the target. It also simplified mirroring the drum LCD displays onto the BSD Unix PC’s X11 display.

Automation Specifications

There were four parts to manufacture/program for the automation:

A scanned in copy of some of the original documentation is availaible here ( (PWP Tannery Distributed Control System(119k pdf).

Variable Speed Drives and Geared Motors

SEW Movitrac variable speed drives SEW Eurodrive R103 geared motor

The five 15kW Movitrac variable speed drives from SEW were installed in a ventilated panel away from the tanning drums (due to extremely corrosive environment). The Eurodrive R103 geared motors connected to the ring gear to turn the drum. The drums were filled with water or brine almost up to the axil, which is easier to turn than the sheep skins without any liquid, as the motor then needs to ramp the full “dead weight”.

Mixing Valves

Mixing valves

The hot, cold and brine water mixing valves were pnuematically controlled from the communications controller, as only one drum at a time could be filled, and this saved having to use semaphores across the drum controllers. The pnuematic valves had limiters inline, as the rapid opening or closing of a valve produced such a large “water hammer” that the far side wall had a few bricks knocked off it!