Sync problem with GBS-8200

[sv1mne]

Hi, I have a GBS-8200 v4.0 (20091204) which is an excellent video converter!

I have one problem when I use it with a composite video signal with an uncommon sync 19.2 kHz from an old HP 8924c test set

I can get picture only from YPbPr , but the left part of the picture is cropped, as in the attached image.

Please advise

I thank you in advance

73

Yiannis SV1MNE

[ramapcsx2]

I think I call this something like "mid-res" in the logs. Can you confirm? Do any of the picture position controls help?

[GXTX]

I have a similar issue with a arcade console I have (Chihiro). If I directly attach it to my capture card I can shift it manually, but when scaled via GBS the part of the signal is just gone. I've attempted shifting in GBS as well as shrinking and it's still not there.

I don't know the exact sync but it is slightly not standard, older VGA monitors aren't able to sync but most newer ones are.

This isn't so much of a big deal as the games would expect that part of the screen to be overscanned in most cases.

[skiphansen]

@ramapcsx2 Hi Robert, my application is the same as the original poster and I had the same results more or less.

Here's a scope capture of the video timing.

Just as a back ground the hp8924c was designed primarily for CDMA cell phones which are now long obsolete. As a result these monsters (59 lbs) have appeared on the surplus market. Ham radio operators have picked them up over the years for use as general purpose RF test equipment. They originally sold for > $30k 25 years ago and still command $1K+ used today.

Since they are now over 25 years old they are starting to die and replacement CRT monitors are very expensive if you can find one at all.

So ... this project might allow us to replace continue to use these excellent pieces of test equipment with "modern" VGA monitors.

Thanks for any help you can offer !

[ramapcsx2]

Well, all I can say for pretty sure is that the GBS will be able to handle this, given the right config. Figuring this out though requires someone to develop it, using exactly this test gear as the source signal. Some points:

• This is CSync, with the amplitude roughly acceptable, but do measure Vpp in circuit with all devices powered up. I think 1Vpp in circuit was still acceptable.
• Watch the debug logs and try every option available, see if anything produces a better result.
• If logs say this is a mid res or "24khz" resolution, it's trying to do the right thing already, but might need some more tweaks. Check source code then around the parts that set up this mode.

[skiphansen]

The scope shot is in circuit, the levels are about 2x without termination so the source impedance looks to be spot on at 75 ohms.

At the moment my gbs8200 is hacked up. I tried adding an LM1881 sync separator chip as suggest by several people and it works somewhat but the results are poor. The blocking capacitors charge after a few minutes and the image goes to solid white. Your code doesn't seem have that problem.

I'm an engineer and programmer so I may try to restore my board back to stock and try to debug the issue. If nothing else I find this project and the TIVA scaler chip fascinating.

[ramapcsx2]

That sounds like a solid foundation :) The ASIC sync input channel for CSync sources seems to be part of the ADC block, only specialized for sync extraction. As such, it reacts to some ADC hidden bits in the programmers manual. None of those are properly documented, but I found that some bits seem to act as clamps and limiters. I picked a combination of options there that had a good effect on the sync processor performance across multiple input sources. This may explain why some things work better in GBSC than stock, but there are so many factors.. In any case, I would expect the problem to be in the ADC to Input Formatter stages, but logs will tell more.