Re: Kinect tear down
- --- In SeattleRobotics@yahoogroups.com, "Peter Balch" <peterbalch@...> wrote:
>I'm a little remiss to post this link, as last time I did so to one
> > > they're using a PIC16F690 to subsample the camera's
> > > 260line composite video to give an 18×16 image at 4 grey levels
> > >
> > However, undersampling works fine with a PIC as long as you have
> > a stable trigger, and they derive that from the analog video
> > signal using an LM1881, which has been around for decades.
> Yes-ish. I designed a product last year that put a white square on the > top-left of a video signal - to synchronise the recorded video with another > signal. The question was analogue or digital? An analogue circuit of > monostables has many more components but the timing is "clean".
of my pages, I got a lot of criticism that everything on the page
was wrong [LOL]. However, this does show that 100-Khz waveform
sampling is possible on a PIC, using equivalent-time sampling
techniques. The trigger ckt was just an LM393 fed by an opAmp
amplifier. I won't say how I did it, as a lot of the current
interest involves commercialization, so you can figure it out for
yourself. :) Circa 2001 ...
The 50- and 100-Khz signals have pretty good fidelity. My guess
is, it still wouldn't be near fast enough for video, except for
lower-resolutions, up to maybe 32x32 or 64x64 with the faster
PICs. That still might not be too bad, for a simple electronics
> The client eventually asked for a simpler digital circuit: LM1881, an 8-pin > PIC and a couple of resistors. The problem is that the PIC can only > start/stop the white signal on its internal (1Mips) clock edge and the > internal clock isn't related to the video timing. So there's 1uS jitter on > the horizontal edges on the square - they're really ragged. It looks horrid > but is prefectly functional.
> So for a sub-sampled camera, I'd want to sample the video at a more
> precisely controlled "correct" time. Either have fixed sample windows > controlled by analogue monostables or maybe the PIC clocked by a PLL locked > to the video syncs.
- You mean the SPO256 by GI, 28pin DIL ?They also had a 40 pin DIL CTS256A-AL2 text to speech for spo256I have use the SPO256 back in the 80s and also last month but never got around to using the latter, I do have the Tandy data that came with them (somewhere I hope).David - robots42----- Original Message -----From: Rich ChandlerSent: Sunday, December 12, 2010 11:26 PMSubject: Re: [SeattleRobotics] Re: Speech and frequency vs time
A long time ago, when Radio Shack still sold IC's, I got a speech
synthesizer set, but I never did anything with it. Has anyone used them?