A long time ago in a galaxy far, far away I worked for a telecom equipment vendor and one of the significant projects I was involved in was a major upgrade of the frequency source & distribution function within a narrowband PSTN switch. In a recent clearout I found a set of boards (actually “SIUs” (for ‘Slide-in Units’) as they were officially known, probably called “blades” nowadays) and began to think about resurrecting them. They were designed to work as a set of 3, but without an exchange shelf/backplane to fit them in I will need to wire them up by hand… a challenge indeed – there are a lot of inter-plane sync, status & comms signals between them – but I do have copies of all the engineering documentation so it should be quite a straightforward proposition 😉
The upgrade project and the design of these boards was started around 25 years ago, so they contain a mix of technology from the late 1980s & early 1990s. The same functions could be achieved with today’s technology in about 10% of the size/weight/power or even less.
The design of PSTN switches of the time was such that to achieve the “five nines” of availability (i.e. 99.999% of the time) almost all of the hardware functions were duplicated or triplicated over different physical iterations. The philosophy was that a single hardware failure shouldn’t bring down the whole system. The exchange clock signals that this set of boards provided were themselves triplicated so that other hardware subsystems that depended on the clocks could perform a “two out of three” majority decision – a common scheme in electronics to allow a receiver to detect if one of the clocks was not as it should be. 2048kHz clock signals were supplied for the main exchange processor and 500Hz/8kHz clocks were delivered locally within the backplane of the main narrowband switch function.
The functional entity the boards provided was known as a “Waveform Generator” or WFG – the enhanced versions shown here added some improvements in reliability, performance and standards compliance (G.703 interfaces) and so were known as an “Enhanced Waveform Generator” or EWFG.
After powering up each board individually a gentle exploration via the good old debug port shows all the cards are still working yay!
Backplane wired… power on…
Testing the inter-plane comms and oscillator status/selection wiring gradually, starting with only 2 out of 3 due to power supply limitations…
And why not add a Raspberry Pi for good measure? As these 3 boards are without control and status reporting from/to the exchange subsystem processor I’m planning on emulating that with some bit-bashed Pi output (another option would be to add 3 x serial-to-USB converters hooked up to the debug ports) and get some sort of web-based control/status dashboard up & running.