Asus SAA7134 'VCR Input'

[Incast]

There is an option on my SAA1734 options marked 'VCR Input' i've ticked this and comparing frame by frame can't see any difference. Does anyone have any idea what this does?

[Arachnotron]

I think this has to do with video sync control. I suspect this setting is the same as the HPLL setting in Dscaler:

HPLL: TV Mode, VCR Mode and Fast Tracking Mode.

My guess would be the drivers only implement Fast Tracking mode and VCR mode and skip TV mode, but I could be wrong here, they might have skipped fast tracking instead.

The SAA7114 datasheet (I never found one for the saa7134 with register descriptions in them :( ) mentions a HTC register with the following description which I think is the same thing:

horizontal time constant selection HTC[1:0]
00 TV mode, recommended for poor quality TV signals only; do not use for new applications
01 VTR mode, recommended if a deflection control circuit is directly connected to the SAA7114H
10 reserved
11 fast locking mode; recommended setting

If there is any difference, I think you should be able to see it in how smooth sharp vertical lines are if you zoom in on them.

But again, I am guessing here and I hope someone can correct me on this.

[edit] and it probably makes a difference in the number of dropped frames with poor tapes or bad TV reception.

[Incast]

As always thanks for your knowledgable reply.

I've ran my own test and can't notice any difference. Below is a png (lossless compression) comparing the two inputs. It would appear VCR input is slightly sharper. This comes from a copy of a tape 8 years old so I hope it would be a good test.
(600kb size, have patience)
http://www.btinternet.com/~jconlin/saacomparison.PNG

If it fails to load click here (http://www.btinternet.com/~jconlin/saacomparison.PNG).

One thing I have noticed with the SAA7134HL is sporadic frame drops which don't follow any specific pattern, it will drop 2 frames then receive the next frame then drop a few more frames. Since this doesn't happen with my Conexant chip and i've only been capturing from VHS i guess it's because I've had VCR input off.

[Arachnotron]

Interesting stuff! :)

So it looks like it has more to do with dropped frames than horizontal jitter. I have been experiencing some framedrops too (about 2 every 1000) but so far attributed that to audio synchronisation. I still have to do some testing there. My VCR has a TBC, so in principle you would expect the VCR input setting not to matter.

I have been trying to track down exactly which registers are set by the driver when you enable the VCR input. Unfortunately, regspy (comes with dscaler) monitors only a few registers when dealing with the saa7134 so i will have to hack something myself. I'll try this weekend, perhaps I can find out more.

So far I have been relying on the dscaler source code for saa7134 info, since the datasheet from philips is completely devoid of any programming info. (there should be a programming guide too, but so far I have been unable to find it)

I wonder how many other phillips chips have this driver option too?

[Incast]

I just did a test to test the theory. I got the same pattern of frame drops, but i got 55 minutes into the capture at 1280x1024 desktop resolution (i don't know if this makes a difference but it appears to be more stable at lower resolutions), whilst word processing on a hard drive which wasn't cleanly formatted. Previously i'd be lucky to manage 45 minutes at 800x600, with the PC only capturing and on a cleanly formatted drive. So it looks like an improvement. However, this is my first capture since a reboot and I'm yet to confirm if this is due to that or due to VCR input. I'm going to do another capture now.

I capture with wdm programs virtualvcr and sometimes iuvcr, so dropped frames with me are not caused by audio synchronisation as i don't allow these programs to alter the frame rate.

My SVHS Panasonic player like yours has a TBC and i came to the same conclusion before as since my VCR is probably better than the ones Philips tested on. I can confirm that frame drops are not coming from the source tape. If i do multiple captures with the same tape frame drops are completely random and irrelevant of the tape source and these are from new rarely watched tapes from 2001.

Edit: I've done another capture and once again reached 55 minutes without dropped frames, so it's certainly an improvement on what I had before.

[Incast]

I did another picture test and after painstaking examination i think we can rule out it being picture related. The numbers indicate the order i captured these, so to take into account possible tape degradation.

http://www.btinternet.com/~jconlin/newcomp.PNG

[Arachnotron]

I did another picture test and after painstaking examination i think we can rule out it being picture related. The numbers indicate the order i captured these, so to take into account possible tape degradation. I agree with you there.

So far I have been unlucky in reading out the register I suspect is involved (Sync control, or SAA7134_SYNC_CTRL in dscaler terms). I did manage to crash my computer a few times, so the time was not wasted :D

I did find this Horizontal time constant on various pages relating to video equipment like line doublers etc.
The quote below is from the website of some sort of video converter box, the Ace.

In general, it seens this HTC determines the time spent in trying to sync with the horizontal sync pulse. The more noise, the more time the signal needs to be monitored to differentiate between noise and the start of the horizontal sync pulse and the more noise (jitter) you get.
So I think you go from a high HTC (slow mode) for TV, medium for VCR and low HTC for fast tracking. Since I assume the price for fast tracking is a low tolerance for bad signal, your findings indicate that the driver alternates between VCR mode and Fast Tracking mode, NOT TV mode. Otherwise disabling VCR mode would increase the HTC and result in better sync, not worse. I can only assume that for a good VCR, the difference between the two modes is to small to really influence jitter much.

I'll try to read out the register settings on another machine in a few days to confirm this.

For Noisy TV Signals, Amateur TV Reception or ATV Repeater Stations often have to cope with very noisy signals. Optimum performance requires a long time constant in the horizontal phase-locked loop so that the ACE can ignore the effects of noise on the sync pulses. On the other hand the standard ACE has a short time constant which was specially chosen for optimum performance with the relatively clean but often jittery video coming off VCRs which must be tracked accurately.

[Incast]

Interesting findings, too bad they can't make this information easily available to us!

Good luck, hopefully we will soon know what VCR input really does!