Technical Info about the Senster
The Computer Hardware and Software
This photograph (taken in 1974) shows the control panel of the Senster.
There are four racks of printed circuit boards in the top of the right-hand cabinet. Below them is an array of lights used to display information for diagnostic and error investigation. Next to them is a column of switches used to manually enter some settings. Below them is the computer front panel and below that is a number of electrical meters.
The left-hand cabinet contains two pressure gauges for the hydraulic system (top left). Below them is a set of headphones (used to debug the sound localisation circuits). Below and to the right of that is an oscilloscope. The two knobs on the left of the 'scope allowed the user to choose various signals to display on the 'scope. Below that is a panel of knobs (function unknown) and what appears to be a panel of indicator lights. At the bottom of the cabinet is a number of racks of printed circuit boards (maybe the 'guts' of the computer ?).
There were several racks of custom made circuits for interfacing the computer with the mechanical structure of the Senster. They are described in further detail in the Electronic Hardware page.
(click to enlarge)
The computer used to control The Senster was a Philips P9201 (evidence from an original shipping receipt). It had 8K of core memory, a punched paper tape unit and a teletype. From the insurance value on the shipping note, it was worth more than my parent's three bedroom house in London.
There is very little on the web about this series of computers, just a page here, whence this photo on the right. You can see that the control panel in this photo is very similiar to the one in the photo of Senster's front panel at the top of this page, but not quite the same.
It turns out that this computer was clone of one of the more common Honeywell 16 series. Adrian Wise runs a website about the Honeywell 16 series where there are details of the history, programming and even an emulator. He told me by email:
"I'm still sure that a Philips P9202 is a DDP-516, but you mention that the Senster used a P9201 and I think this is the key. I believe that the 9201 was a DDP-416, a cost-reduced version of the 516, with a very limited instruction set. The code supports this, since there isn't a single use of the index register or the B register and the interrupt service routine doesn't save anything other than the A register."
The Honeywell DDP-416 was introduced in April 1967. Relative to the H316 and DDP-516 machines, the DDP-416 lacked:
- The index register, X
- The auxilliary register, B
- The carry bit
- The Interchange Memory and A (IMA) instruction
- The Compare and Skip (CAS) instruction
- Almost all arithmetic instructions
- The sense switches
- Input/Output keys
- The 8-bit character manipulation instructions
- The instructions concerned with double precision operation
The machine had the same cycle time as the DDP-516 (0.96 microseconds).
It was intended for control applications that didn't require the more
advanced arithmetic.
The Philips P9202 computer, from the same series as the P9201, which the Senster used.
Senster's program was written by Ihnatowicz himself. You can tell from the style of the code that he was not a professional programmer (I wish he'd put in more comments !) but the code is remarkably sophisticated. He got some help from Peter Lundahl, the Evoluon sysadmin, who also added refinements to the software after Ihnatowicz left at the end of 1970.
I have an almost complete listing of Senster's code from 1970. Nick Lambert scanned it for me, and after a bit of processing, I passed it on to Adrian Wise. He ran it through an OCR program and generated an html listing. You can find the complete code listing here, as a set of html files:
Alternatively, I have assembled it into one pdf file (9.15MB). Here are the first three pages to give you an idea of what it looks like:
Adrian and I are gradually working our way through the code, trying to extract the algorithms Ihnatowicz used. We will post the results here when they make sense !
If you want to study the code, it'll help if you read the description of the Senster's electronic system, because there are non-standard input/output registers. Also, there are full scans of the manual for the 16 series here.