Why no built-in TBC in 1980+ VCRs?

[ppppenguin]

The whole history of VTRs is a battle against mechanical imperfections.

Quadruplex, with its small drum, could, under ideal conditions, keep the jitter down to 100ns or so, good enough for synchronous monochrome replay. It wasn't easy to avoid banding and other visible errors. The first TBC was Amtec, an analogue delay line with varicaps instead of fixed capacitors. Correction range less than 1uS (I think) but it was enough. Followed by Colortec to get good enough stability for direct NTSC playback.

The frequency spectrum of timing errors on helical scan VTRs was very different. Much more low frequency stuff due to the big drum. Hence hooking. Still all sorts of other errors caused by tape flutter and heaven knows what else. If the errors are bad you get velocity errors along a line which aren't easy to correct.

Colour under was a brilliant kludge to record colour on low-cost VTRs, at the expense of wrecking the delicate relationship between colour subcarrier and H. Didn't matter in domestic environment.

General purpose TBCs had a tough job following all these errors on all sorts of VTRs and tapes. When memory was expensive enough that you only afford a few lines of storage a few early TBCs had a feedback output to drive the capstan motor of the VTR via a power amplifier in order to get true synchronous replay. No idea if this was used very much in practice. I once did it manually with an audio oscillator, public address amplifer with 100V output and a step-up transformer. I could just about keep an Ampex 7003 1" VTR within the correction window of an early CVS517 TBC. But not for very long.

From the early 1990s I designed framestore TBCs for a small company called G2 Systems in the UK. They also did standards conversion, though not to full professional quality. The first versions used dedicated framestore FIFO chips. Later I used first generation SDRAM with an FPGA to give multiport video framestore memory. There were great "get out of jail" boxes in their time. You could stick just about any SD video signal into them (including S video, SDI and in some cases analogue component) on PAL/SECAM/NTSC (and oddballs like NTSC443, PAL-M, PAL-N) and have the output perfectly timed into your vision mixer.