Early PDP-11 Peripherals


Depending on options, this system could provide up to 32 channels of A/D input, with a 12 bit bipolar converter. The subsystem was implemented in a dedicated backplane and power supply that was bolted inside a cabinet. The system used DECs FLIP CHIP series of logic cards. The system was connected via a unibus cable.


A one hex module real-time analogue subsystem, consisting of :-
  • 16 channel, 10 bit A/D converter with sample-hold.
  • 2 channel, 10 bit D/A with display control
  • Programmable crystal clock with external schmitt trigger input


    The famous teletype. It was almost completely mechanical, and had a continuously running motor. There was a mechanical reader and punch on the side and it operated at 10 cps (characters per second). By using two stop bits, and ensuring that a line feed immediately followed carriage returns, there was just enough time for the carriage to make it to column 1 from the far margin. Even the parallel output from the keyboard (mechanically encoded of course) was serialised by a distributor driven by the motor via a clutch.

    It was possible to read and punch binary tapes, but the mechanics would quickly wear out at tape, initially elongating the sprocket holes.


    A replacement for the KL11. It was a single quad card, and used the new fangled UARTS. RS-232 was an option, and it used a crystal oscillator in four speed groups.


    A 16 bit general purpose DMA interface. It was a complete system unit with four quad and one single width cards. One ugly feature was that the 16 bit address counter didn't overflow into the extended address bits, so that DMA transfers couldn't cross 64Kb boundaries. It was replaced by the DR11W, which was a single hex board and became almost an industry standard for digial I/O interfaces.


    The original digital I/O interface (latter replaced by a single card). It consisted of a separate interrupt and address decode cards (single width) and the I/O interface with side sockets for PC board cable connectors.


    An eight port serial multiplexer with input silos and interrupt driven output. Available as either RS232 or 20mA loop.

    KE11 (EAE)

    The Extented Arithmetic Element was a unibus peripheral that implemented hardware multiply, divides, multiple shifts and fraction normalisation. It was implemented as a number of registers in the I/O page; you would write the operands into a set of eight registers. For a multiply, simply loading the multiply register would initiate the calculation. This device was mainly used on the PDP11/20 which lacked a multiply and divide instructions (as did the /04, /05 and /10.


    The original serial interface for ASR-33 teletypes. It consisted of three cards, a address decoder, interrupt controller and serial controller card. The interface was set up for a 20mA loop, 110 baud with two stops bits. Other speeds were available as options and the timing was by a RC oscillator.


    A general purpose peripherial processor using bit slice ALU (74181). It could do DMA cycles to main memory, access the I/O page and generate interrupts. It was used with the synchronous line unit (DMC11) to implement DDCMP protocols. Also used with DZ-11 to ease the burden of interrupt per output character by implementing pseudo DMA in software. One board could run 5 DZ-11's.

    LA-30 DECwriter

    The first in a long series of dot matrix printers from DEC and could muster 30 characters per second. It required fill characters after carriage returns.


    This data acquisition system was designed to minimise the cost of building realtime systems. It used a common address and interrupt logic to control several internal devices. The following options were available :-
    • 12 bit, 8 channel A/D converter (40KHz throughput)
    • DMA interface for A/D
    • Additional sample/hold amplifier for dual, simultaneous sampling
    • Additional 8 channel A/D multiplexer
    • Differential input amplifiers
    • Programmable gain amplifiers
    • Realtime programmable crystal clock with two Schmitt triggers
    • Display control with two 12 bit D/A converters
    • 16 bit buffered digital input / output with two relays
    • Expansion box for an additional 48 analogue inputs and 8 outputs
    • 6 digit programmable LED display (standard)


    A high speed reader and punch. The reader used a large stepper motor and optical sensing to attain 300 cps reading. It used unoiled fanfold tape and would ramp up and down the motor speed for start/stops. Early models strobed the data from the optical sensors from a delay after the stepper motor, but were hard to adjust for correct data recovery. Latter models had a optical sensor under the sprocket hole, which being smaller than data punch holes, neatly strobed the data in the middle of the holes. The punch ran at 50 cps.


    The RF11 disk controller can access up to eight RS11 fixed head disk drives. Each drive stored 524Kb of data, addressable on 16 bit word boundaries (most other DEC disk drives had 512 byte blocks). The drive used 128 fixed read/write heads and recorded data on a single nickel-cobalt plated surface. The data transfer rate was 125K bytes per second. The controller occupied the top of a cabinet and had a wire wrapped backplane.

    RK02 RK03 RK05

    These used a single platter, removable pack for 1.25Mb of data (RK02) or 2.5 Mb of data (RK03/05). The initial controller was the RK11C which used 'flip chip' modules to build the controller on a large wire-wrapped backplane that mounted in the top of a cabinet. The latter RK11D controller consisted of 4 quad boards in a custom system unit, mounted within a CPU or expansion box. It was possible to have up to eight drives per controller. Only the RK05 drive was manufactured by DEC, the others being rebadged Pertecs. The RK05 drives did not use embedded servo tracks as latter ones did, but used an optical encoder to sense the position of the voice coil head positioner. A special alignment disk pack was required to calibrate the heads (which was a yearly event). It had specially recorded, off axis data bursts and a CRO was used to match the amplitudes to center the heads. They could only be adjusted out, so if you went too far, the drive had to be run down and the heads pushed back in. The drive geometry was :-
    • 203 tracks
    • 12 sectors
    • 2 heads
    • 512 byte data blocks

    TU56 DECtape

    This was dual tape transport that used small 3.9 inch spools of 3/4 inch tape. The system used 10 tracks, 4 for timing and mark, and the other tracks as a redundant 3+3 Manchester encoded data. The tape paths were short and air buffered. It was possible to read/write data in either direction, and the system was unusual in that each block was addressable and the system could be treated like a slow disk drive. Some early operating systems for the PDP-11 (and others like the PDP-8) could run the entire operating system from these units.


    This is the display processor that was combined with a PDP11/05 in a machine called a GT-40. It was a separate processor that drove a 1024x1024 vector graphic display. It could generate characters, points and lines in hardware (using binary rate multipliers), eight levels of display intensity and had a light pen. The processor shared the same memory as the PDP11/05 which would set up a 'display list' for the system to output. The same processor was latter combined with the faster PDP11/40 to produce GT-44.