PAL-Plus never took off in the UK. I didn't realise that it was still in use elsewhere in Europe.
The limit on the capture block size of 674*256*4 is presumably enforced by hardware. It means that continuous waveform sampling can be done for almost 20 ms. which in itself is not bad going. The VDELAY restriction means that the dummy sync generator must generate pulses for at least two lines and I would imagine that a further couple will be needed before the field pulse to allow odd / even field discrimination. I know that we don't really need this but the 848 seems to get confused and refuses to capture if the field ident gets mangled based on problems I've had transcribing video from camcorders. I don't think that the 848 is too picky about the serration pulses so overall it should be possible to capture consecutive 20 ms blocks separated by 360 us holes. For display purposes, the horizontal refresh is unlikely to be an exact multiple of the sample window size so the holes in successive scans should get covered up. The chopping action on most dual trace 'scopes relies on this effect.
The big advantage of using a video capture card for a cheap DSO is the high bandwidth into PC memory. It is not necessary to reduce the AtoD sampling rate when using a slower timebase setting so the aliasing problems which plague most external box solutions (and even some Tek LCD DSOs) simply go away. The display software should translate sample distribution within each time slot into intensity, thereby emulating a conventional 'scope. Modern processors should be fast enough to do this with imperceptible delay.
Is the RAW mode to which you refer the same as VBI frame mode? This would appear to capture blocks of up to 674 * 256 *4 samples but whilst it ignores the horizontal sync, something has to cause a vertical reset to initiate capture. The VSYNC bit appears only to be software resettable only so I assume that some input on the GPIO has to be provided or suitable vertical sync inserted into the incoming analogue signal (a PIC might be fast enough to do this).
The limitations on the capture block side could be a serious problem with slow waveforms unless capture can be restarted quickly. What is your experience with this?
I suspect that the limited signal to noise ratio of many capture cards may be more of a limitation that the 8 bit quantisation but even 36 dB would be useful for a DSO.
Incidentally, many thanks for all your splendid work on bttv.
Slashdot Mirror
PAL-Plus never took off in the UK. I didn't realise that it was still in use elsewhere in Europe.
The limit on the capture block size of 674*256*4 is presumably enforced by hardware. It means that continuous waveform sampling can be done for almost 20 ms. which in itself is not bad going. The VDELAY restriction means that the dummy sync generator must generate pulses for at least two lines and I would imagine that a further couple will be needed before the field pulse to allow odd / even field discrimination. I know that we don't really need this but the 848 seems to get confused and refuses to capture if the field ident gets mangled based on problems I've had transcribing video from camcorders. I don't think that the 848 is too picky about the serration pulses so overall it should be possible to capture consecutive 20 ms blocks separated by 360 us holes. For display purposes, the horizontal refresh is unlikely to be an exact multiple of the sample window size so the holes in successive scans should get covered up. The chopping action on most dual trace 'scopes relies on this effect.
The big advantage of using a video capture card for a cheap DSO is the high bandwidth into PC memory. It is not necessary to reduce the AtoD sampling rate when using a slower timebase setting so the aliasing problems which plague most external box solutions (and even some Tek LCD DSOs) simply go away. The display software should translate sample distribution within each time slot into intensity, thereby emulating a conventional 'scope. Modern processors should be fast enough to do this with imperceptible delay.
Is the RAW mode to which you refer the same as VBI frame mode? This would appear to capture blocks of up to 674 * 256 *4 samples but whilst it ignores the horizontal sync, something has to cause a vertical reset to initiate capture. The VSYNC bit appears only to be software resettable only so I assume that some input on the GPIO has to be provided or suitable vertical sync inserted into the incoming analogue signal (a PIC might be fast enough to do this).
The limitations on the capture block side could be a serious problem with slow waveforms unless capture can be restarted quickly. What is your experience with this?
I suspect that the limited signal to noise ratio of many capture cards may be more of a limitation that the 8 bit quantisation but even 36 dB would be useful for a DSO.
Incidentally, many thanks for all your splendid work on bttv.