At lunch today Jack regal
ed us with a tale about a state-of-art instrument he worked on during his tenure at
Tektronix. The machine used a convoluted method to measure very short transient signals, that is signals that only appear occasionally, and only for a brief period of time.
The machine used a CRT from an oscilloscope and a primitive digital camera to record very fast transient signals. The CRT was buried inside the machine, it was not visible to the operator. The signal from the probe was used to draw the pulse on the screen of the CRT. The digital camera faced the CRT and digitized the projected image. This was back in the early 1970's, and the digital camera had a very low resolution, about one-quarter of a mega-pixel, to use today's parlance.
The big customer were the guys doing underground nuclear bomb tests. They would bury a bunch of these instruments in the ground along with the bomb. When the bomb was detonated, the R7912 digitizers would send out digital information until the detonation destroyed them.
This gadget is so weird I thought I better include another description of it. From
Tekwiki:
The Tektronix 7912 is a series of high-speed digitizers that take one 7000-series vertical plug-in
and one 7000-series horizontal plug-in.
All 7912 models use the same internal CRT-based, digitizing scan converter tube (T7912) that is not visible from the outside.
The signal from the vertical plug-in deflects a writing beam through distributed deflection plates.
The electrons hit a small flat rectangular solid state target,
conceptually similar to the image sensor in a digital camera.
The resolution of the target is 512×512, giving 512 points in the time
domain and 9-bit linear quantization of the input voltage.
With a 7B92 sweeping the whole X-axis in 5 ns, and the 7912 capturing 512 samples in that sweep,
the 7912 performs the function of a 100 GSample/s A/D converter.
5 nanoseconds divided by 512 samples, means we are taking 100 measurements every nanosecond, which means our sample rate is 100 billion measurements every second, i.e. 100 GSamples/s.
1 comment:
I was a programmer on the R7912 project at Tektronix in the early 1970s. I wrote the first device drivers to remotely control the device and get the video and digital data out of the device as quickly as possible. I went to Los Alamos to demonstrate an array of R7912s controlled by a single RK05-based PDP-11.
Thank you for sharing this information!
Lynn Robert Carter
Professor of Practice
School of Computing and Augmented Intelligence
Arizona State University
Post a Comment