It is currently 19 Mar 2024, 08:35




 Page 27 of 31 [ 606 posts ]  Go to page Previous  1 ... 24, 25, 26, 27, 28, 29, 30, 31  Next
Author Message
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 20 Oct 2020, 23:58 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
For anyone that is contemplating mod'ing the cxadc card as i have done, this is the LPF i used with the Video Essentials and Beauty and the Beast samples: https://www.minicircuits.com/pdfs/SXLP-13+.pdf I have not done "thorough" testing with this, this is just what i used. I have also this purchased: https://www.minicircuits.com/pdfs/SXLP-10.7+.pdf which will probably work better for PAL LDs. I'm not doing tape captures, so not sure what LPF would work for that, but maybe this one: https://www.minicircuits.com/pdfs/SXLP-5+.pdf
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 26 Oct 2020, 02:17 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
I tested an LD capture with a new stock ATI "Philips" half height CX23883 card. vs. my stock original Asus "HP" CX23880 card. With no modifications sending the signal through the s-video luma pin, the gain is set to 31 on both cards. Before I started comparing the captures though, I ran a control test with no inputs connected to both cards. The noise floor is -85dB for the Asus one and -63dB for the ATI one. The big difference is due to a spike around 5.8MHz on the ATI card, which creates 4 harmonics in the upper and lower sidebands. Without this, the noise floor is -85dB, which is spot on with the Asus card. Actually the 5.8MHz signal could be raising the noise floor just a bit as well, so without it the noise floor could be even lower.

This is a good illustration of differences between cards. My Asus card has quite a number of components, probably adding noise to the capture. The ATI ones I have are much simpler, but have 20dB of some rouge signal polluting the capture.

Asus noise floor:

Image

ATI noise floor:

Image

Comparing two CLV captures of my Sony demo disc...

The Asus card signal level is -52dB.

Image

The ATI card signal level is -62dB.

Image

As for the decodes, the Asus one is much the same as before (utter crap), but now there is a pulsing signal that stomps all over the image. That was probably filtered out by the Radio Shack video amp I was using before. FM audio still decodes fine with this capture.

Image

With the ATI card the decode is, as Jim Carrey in The Grinch would put it, "Horribly mangled!" It's 110% total garbage. Not sure why, other than maybe that 5.8MHz signal completely threw ld-decode for a loop. FM audio is affected too, with lots of noise and dropouts. One of the lower sidebands lives in that neighborhood of the spectrum, so no surprise.

Image

So stock, the ATI card performs magnitudes worse. However, after adding an appropriate low pass filter suitable for LaserDisc and tapping into the vmux 2 input directly, Tony was able to get much better results with a very similar card of his that rival or match the Domesday Duplicator. That method bypasses all the noisy components and exposed traces on the card. That will be my next step using a BNC female header, RG174 cable, 1μf DC removal capacitor in series, 50 ohm resistor to ground for impedence, and 13MHz low pass filter.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 27 Oct 2020, 23:36 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
I was able to trace the signal path from mini DIN connector to CX chip on my ATI cards. It takes quite a trip across the entire card behind the unshielded bottom of the RF tuner, where it can pick up a lot of noise. Mini DIN pin 3 -> 75 ohm terminating resistor on back -> fuse -> long trace across the card -> another 75 ohm terminating resistor -> 25 ohm resistor -> 50 ohm resistor -> through hole pad to trace underneath CX chip on front -> pin 142 of CX chip. The data sheet for this says pin 143 should be vmux 3, which is mapped to vmux 2 in the CXADC driver. It may just be my probe slipping when I got that last photo.

Image
Image
Image
Image
Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 00:24 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
i can confirm that the PCIe to PCI riser card works fine with cxadc. i used this one:

https://www.newegg.com/p/17Z-00PR-00F85 ... KFD9YZ9124
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 01:11 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
There's more than one way to skin a cat! on trying to obtain 40msps from my cx card, i decided to go another route: replace the timing crystal.
root@:/laserdisc# ./mydd 40mspscrystal2.r8
^C1768+0 records in
1767+0 records out
926416896 bytes (926 MB, 884 MiB) copied, 23.1788 s,** 40.0 MB/s**

root@:/laserdisc# ld-decode 40mspscrystal2.r8 out/40mspscrystal2
file frame 1 CAV frame 15345
file frame 2 CAV frame 15346
file frame 3 CAV frame 15347
file frame 4 CAV frame 15348
file frame 5 CAV frame 15349
file frame 6 CAV frame 15350
^CTerminated, saving JSON and exiting
10fsc now= 50msps:
root@:/laserdisc# ./mydd 50msps10fsctest.r8
^C3133+0 records in
3132+0 records out
1642070016 bytes (1.6 GB, 1.5 GiB) copied, 32.8782 s, 49.9 MB/s

root@:/laserdisc# ld-decode 50msps10fsctest.r8 out/50msps -f 50
file frame 0 CAV frame 23684
file frame 1 CAV frame 23685
file frame 2 CAV frame 23686
file frame 3 CAV frame 23687
file frame 4 CAV frame 23688
file frame 5 CAV frame 23689
file frame 6 CAV frame 23690
file frame 7 CAV frame 23691
file frame 8 CAV frame 23692
file frame 9 CAV frame 23693
file frame 10 CAV frame 23694
file frame 11 CAV frame 23695
file frame 12 CAV frame 23696
^CTerminated, saving JSON and exiting

this is the crystal i used: https://www.mouser.com/datasheet/2/115/ ... 129572.pdf
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 04:26 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
Freaking amazing, you actually managed to overclock your card?? At 40 or 50, does the quantization noise of 8-bit samples no longer really matter? I wonder if that allows you to recover more horizontal resolution. A higher sample rate may help ld-decode discern rapid changes in frequency needed for sharpness. Your previous test images seem to show about 450 lines on the scale before they blur together. Have you noticed an increase in SNR with one of your prior samples? Using the same portion of the same disc (as SNR increases as a CAV disc is played)?

In other news, I plan on directly tapping my chip this weekend using nothing more than a 1uf ceramic DC blocking capacitor in series and 13MHz low pass filter from Mini Circuits. Tony and I determined the 50 ohm termination is likely not needed unless the signal is strong enough to cause image based distortion and aliasing (due to reflections). I can always terminate later of course if needed.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 04:33 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
The main motivation for this, was because at any other rate besides 40, the samples are resample via ffmpeg, which lengthens the time to decode.

I am decoding two full side ggv captures now. However comparisons with previous captures won't be 100% indicative because it is a different card than previous ones.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 04:33 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
One full side of ggv at 40msps and one at 50msps
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 10 Nov 2020, 20:44 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
titan91 wrote:
Freaking amazing, you actually managed to overclock your card?? At 40 or 50, does the quantization noise of 8-bit samples no longer really matter? I wonder if that allows you to recover more horizontal resolution. A higher sample rate may help ld-decode discern rapid changes in frequency needed for sharpness. Your previous test images seem to show about 450 lines on the scale before they blur together. Have you noticed an increase in SNR with one of your prior samples? Using the same portion of the same disc (as SNR increases as a CAV disc is played)?

In other news, I plan on directly tapping my chip this weekend using nothing more than a 1uf ceramic DC blocking capacitor in series and 13MHz low pass filter from Mini Circuits. Tony and I determined the 50 ohm termination is likely not needed unless the signal is strong enough to cause image based distortion and aliasing (due to reflections). I can always terminate later of course if needed.


Ld-Decode resamples all input that isn't 40msps, to 40msps, before processing, using ffmpeg, if i understand correctly, so i don't expect to see any increase in resolution. I decoded the 2 captures, and interestingly, they have slight variations in S/N, but not consistent ones. At some points, the 50 is better, and at others, the 40 is better. The differences are small, .2 to .4 db. These differences could well be attributed to other minor fluctuating factors. More testing will be needed.

I am still waiting for my (hopefully) sufficient amplifier. When that comes, reintroduction of termination may be needed. Right now, capturing with just the signal of the player RF gain, the terminator caused more detriment than having no terminator at all.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 11 Nov 2020, 21:33 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
I also noticed, the tuner has a few pins soldered to the board. I decided to remove the tuner and slot my BNC jack in its place.

Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 14 Nov 2020, 04:02 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
The mod is done. And it went well! I got some help and we removed the tuner with a desoldering gun. Once that was out of the way I searched for a top side trace for the video output of the tuner, but had no luck. So I stuck with the vmux2 line.

I had access to this awesome surface mount component remover. Taking the 50 ohm resistor off was a 2 second process.

Image
Image

I then tacked on just enough copper to ensure no bridges or shorts.

Image
Image

I then installed the BNC header. The nut fits perfectly. I used a 104 100nf ceramic capacitor in series for DC isolation.

Image
Image
Image

Finally, some low temperature hot glue secures the mod.

Image
Image

Will test tomorrow.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 14 Nov 2020, 04:22 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
Nice!
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 14 Nov 2020, 22:14 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
Results are not good. Although the actual SNR appears to be better, the blatantly annoying interference remains, with and without my 13MHz low pass filter. I'm still having to max out the gain at 31. The interference manifests itself as waves. Zooming in on the decode, I don't see any noise behind the waves, so it's just that interference that's left evidently. The filter helps with this, but it's still there and even messes up the TBC. I'll try to adjusting the RF gain in my player, but I'm not having high hopes for the design of these ATI cards.

Without 13MHz filter:

Image

With 13MHz filter:

Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 01:15 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
Well that problem didn't last long! I cut the long trace where my ground point for the cable was soldered in. Once I did this the noise cleared up 100% and the TBC now correctly lines up the image.

Image

Without 13MHz filter:

Image

Image

With 13MHz filter:

Image

Image

No difference that I can visually see with or without the filter. But the filter really seems to attenuate my FM audio signals for some reason? Although my video signal is now up, I'll still need to boost it some.


Last edited by titan91 on 15 Nov 2020, 01:25, edited 2 times in total.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 01:18 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
I think you may still have the onboard LPF in the mix.

Good find on the interference though!
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 01:29 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
9954tony wrote:
I think you may still have the onboard LPF in the mix.

Good find on the interference though!


You're definitely right, those tan colored components aren't 75 ohm resistors, they're capacitors. It's possible one of them is still acting as a 5MHz low pass filter. My next step it to remove them both.
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 19:08 
Serious fan
Serious fan
User avatar

Joined: 12 Sep 2015, 05:57
Posts: 210
Location: United States
Has thanked: 5 times
Been thanked: 54 times
@titan91

Your frequency plot with no external LPF indicates that there is most likely an onboard LPF still in play. With no onboard LPF, you should see signal reflections in the upper spectrum. Here is an example of no LPF:

Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 21:15 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
9954tony wrote:
@titan91

Your frequency plot with no external LPF indicates that there is most likely an onboard LPF still in play. With no onboard LPF, you should see signal reflections in the upper spectrum.


Yes sir, and here it is connected to my signal pin which I'm pointing at with a meter probe. The cap in question is on the left. There's continuity to ground on the left side of the capacitor. The right side is connected to the input pin. It's next on my list to flick off the board.

Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 15 Nov 2020, 21:52 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
Cap removed, will test again tonight.

Image
Offline
 Profile  
 
 Post subject: Re: (WIP) Laserdisc software image decoder from raw signal
PostPosted: 16 Nov 2020, 14:09 
Honest fan
Honest fan
User avatar

Joined: 14 Jul 2019, 02:53
Posts: 81
Location: United States
Has thanked: 10 times
Been thanked: 4 times
Before testing again, I learned the pin I need is in fact pin 143, and tapping pin 146 was a red herring caused by another path. I found another set of capacitors on the back of the board lined up with these pins and removed them. Unfortunately the pad circled in red going to pin 143 came off the board when I soldered to it. So the plan now is to solder to pin 143 directly from the top of the chip.

Image
Offline
 Profile  
 
Display posts from previous:  Sort by  
 Page 27 of 31 [ 606 posts ]  Go to page Previous  1 ... 24, 25, 26, 27, 28, 29, 30, 31  Next


Who is online

Users browsing this forum: No registered users and 12 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to: