System Bus


System Bus

When I was starting with the bus I was looking for an existing solution, no need to invent the wheel. Soon I found the following web site of the RetroBrew Computers which is a site about hobbyists computers based on the ECB bus. The bus seemed to fit nicely for the Q-Bus.

ECB Bus

So I mapped all the Q-Bus signals to appropriate signals of the ECB bus and used it as the base for my system and named the system bus the QBUS64. This allowed me to use one of the standard bus boards available to begin with the system setup. Later I designed my own PCB. The following shows the schematic

QBUS64

and the finished Bus of my own design.

System Bus

As the real Q-Bus it is fully asynchronous. The speed is about the same as the real Q-Bus. But as with any DCJ11 based system the performance depends primarely on the memory speed. In the case of the PDP‐11/Hack the memory is located on the CPU board and operates at the speed of a cache similar to the PDP-11/93.

Please note that I often use slightly different names for the signals than are used on the Q-Bus. E.g. I did not prefix all signals with B for bus and spelt the signals differently, e.g. IACK instead of IAK, A instead of DAL, etc. This was very thoughtless. I’m continuously changing this whenever I make a new revision.

Address/Data Bus Signals

The Address and Data are multiplexed like in a real Q-Bus. However unlike the real Q-Bus the Address/Data signals use positive logic and tri-state buffers. The main reason is because the first trials on the bread-boards did not use bus transceivers. Also later the UART and the CPLD talked directly to the system bus. The UART has also positive logic. Later I tried to use an inverted bus with the CPLD but it always caused problems with WinCUPL. Might also be caused by my limited experience then with CPLD programming at that time. For a short time I was thinking about a redesign to better match the Q-Bus but then I decided it is not worth the effort and it is not necessary in case I want to translate a design to Q-Bus.

Control Bus Signals

The rest of the signals however uses open collector drivers and inverted logic. It seems natural for an asynchronous bus with a lot of bidirectional control signals and the option of multiple masters. At least it is very easy to implement the Q-Bus protocol using open collector outputs. QBUS64 uses the following Q-Bus signals:

QBUS64Q-BusRemark
SYNCBSYNCAddress Latch
DINBDINMaster read data from slave
DOUTBDOUTMaster write data to slave
WTBTBWTBTMultiplexed Write/Byte
BS7BBS7Bank Select 7 aka IO Page on a PDP-11
RPLYBRPLYReply from slave, acknowledge
IRQ0BIRQ4Interrupt request level 4, currently all boards use this level
IRQ1BIRQ5Interrupt request level 5
IRQ2BIRQ6Interrupt request level 6
IACKIBIAKIInterrupt Acknowledge In Daisy-Chain
IACKOBIAKOInterrupt Acknowledge Out Daisy-Chain
DMRBDMRDMA Service Request
SACKBSACKDMA Service Request Acknowledge
DMGIBDMGIDMA Grant In Daisy-Chain
DMGOBDMGODMA Grant Out Daisy-Chain
EVENTBEVNTCyclical Event, System Tick
INITBINITInitialise
HALTBHALTCPU Halt

The QBUS64 does not implement or does not make use of the following signals

  • BPOK
  • BDCOK
  • BREF
  • BIRQ7

The open collector signals are only terminated on the CPU board. The bus is too short to require termination on both sides. Also I do use a 3.3V voltage and a single pull-up resister, this is the electrical equivalent to the voltage divider setup of the Q-Bus and saves a lot of power. Modern SCSI terminators for example used the same trick.

Errata

Initially I thought it would be a good idea to power the system via an USB power supply and therefore added a Type-B connector to the PCB, however I swapped GND and VCC which makes this option unusable. Don’t solder the connector, just use the screw heads as you can see on the picture.