GBS-8220, 23DB Connectors and Leads

[Stedy]

@Higgy,

A SCART cable to my original design should work fine. The sync signal will be inside the absolute maximum levels.

@dooklink

The Amiga uses 240p/288p video or 710x240 or 710x288 pixels, there abouts 720 pixels per line might be better as it can output 700+ pixels with overscan.

The Amiga C-Sync is a 5V CMOS signal, which is time aligned to the RGB video. A typical composite video signal is delay from the RGB video by 5-20us, depending on the RGB to composite converter delay, using the Sony CXA1145 encoder in the Amiga.

Stick to standard VESA resolutions, 1280x1024 or less. We want 5:4 or 4:3 aspect ratio.

The line doubling sounds good, what setting did you tweak?

@amigappc

I have a GBS-8200, labelled as a V4 PCB, it works quite well, I can't comment on the GBS-8220 as I don't have one yet.

The RVA development board has already been used, I posted this on my blog, how else can I convert the Atari 7800 video signal to VGA

Additional tests

I wanted to test something other than the Amiga, to see how well the GBS-82XX coped with another retro system. The Atari 7800 was nearby. My unit has been modified to output S-Video (Y/C). I then connected this to a RVA dev board, which converted the S-Video to YPbPr. The YPbPr video was fed into the GBS-8200 board. It sort of worked. I had trouble with the RVA board outputting black and white video from the Y/C source. I tried tweaking a few settings but this is a task for another time. Feeding in composite video, using a S-Video to composite lead from the Atari 7800 worked. Atari 7800 up-scaled to VGA:

Ian

[Stedy]

Hi,

Originally Posted by RedskullDC

Hi Higgy, et al.


Apologies if this has already been posted somewhere here.

Ewan of http://microbeetechnology.com.au/ has come up with a fix for the white spots on the GBS boards.

His explanation:

"The SDRAM interface from the TV5725 scaler chip on the board has no damping resistors in the clock & control lines
and fails to calibrate the timing of the interface properly as a result. It is marginal and causes bad screen refresh data.
As such it took some time to track this down as the cause.

There is a track to be cut and the resistor (SMD 0603 size, or 1608 metric) gets fitted across the cut. Then the capacitor is soldered to the ground end
of a nearby bypass cap & the free end linked to the SDRAM side of the resistor that was fitted across the track cut. It is not very elegant, but it works & cures the video noise totally."

Some pics of the mod:
http://www.microbeetechnology.com.au...200Mod-pt1.JPG
http://www.microbeetechnology.com.au...200Mod-pt2.JPG
http://www.microbeetechnology.com.au...200Mod-pt3.JPG


Have done the mod to my GBS board, and the difference is noticeable.

Hope this helps,
Red

I measured my V4 GBS-8200 board today using a fast oscilloscope. The SDRAM clock signal did not violate the electrical limits of the Hynix SDRAM. On the GBS-8200 V4 there is no need to apply this modification.

If I can get my hands on a GBS-8220 V3 board, I'll repeat the test and advise on suitability.

Ian

[kipper2k]

I have a few of these boards kicking around somewhere, is there a marking showing revision # somewhere ?

[Higgy]

Originally Posted by kipper2k

I have a few of these boards kicking around somewhere, is there a marking showing revision # somewhere ?

We got the whole world looking at this now we must be able to kick the GBS- 's ass!

@kipper the version is printed on the PCB just above the four push buttons:

http://www.imagebam.com/image/d2b006365736073

Sorry not sure if it was a typo, but not sure about revision number of board/firmware.

[rare_j]

Originally Posted by dooklink

I prefer exact line doubled output with no aspect ratio correction as I use analogue scan line generators (SLG3000 / SLG HD=).

That screenshot looks really nice. I'd use it that way as well, straight up doubled into an SLG and let the display device handle the scaling. Are you able to run the 240p test suite on your Wii through the GBS? The Wii version can switch 240p/288p/480i/576i. The checkerboard pattern seems to be a quick and easy test for showing up any deficiencies in handling the signal. It'd be great to see how this board with your mod handles that.

[dooklink]

Originally Posted by rare_j

That screenshot looks really nice. I'd use it that way as well, straight up doubled into an SLG and let the display device handle the scaling. Are you able to run the 240p test suite on your Wii through the GBS? The Wii version can switch 240p/288p/480i/576i. The checkerboard pattern seems to be a quick and easy test for showing up any deficiencies in handling the signal. It'd be great to see how this board with your mod handles that.

That's exactly what I use for testing. The checkerboard is pretty hard on the scaler. You can see horizontal artifacts on PAL signals, but seems better on NTSC. You can also see refresh tearing when the refresh is slightly out between the device and output. This is usually not noticeable though on normal content.

[Anemos]

Hi Stedy, check and this capacitor C33, is on the line Sync to the Gonbes IC, before years..if i remember fine i had cancel him and had worked better, maybe the change volume or other capacitor type is the solution..

Hi guys!
I gonna participate in custom firmware development for GBS8200.
The main question is: where is i2c registers information in datasheet http://file.yizimg.com/420856/2013041613555642.pdf
Somebody have the information about configuration registers of Trueview5725?

May be someone have another(full) version?

[Higgy]

@_dx - have you tried looking at the link to some docs in Google Docs here on the first post. Maybe something here to help you?

http://shmups.system11.org/viewtopic.php?f=6&t=52172

@dooklink - I read on shmups about people's comments regarding colours on Megadrive. I was wondering if there are similar things that apply to the AMIGA also apply to the Megadrive. As @Stedy's investigations a while ago found that some of the signals outputed were too strong for LCD TV's (but ok on original CRT) and he recommended adding resistors to many of the signals

[dooklink]

Originally Posted by Higgy

@_dx - have you tried looking at the link to some docs in Google Docs here on the first post. Maybe something here to help you?

http://shmups.system11.org/viewtopic.php?f=6&t=52172

@dooklink - I read on shmups about people's comments regarding colours on Megadrive. I was wondering if there are similar things that apply to the AMIGA also apply to the Megadrive. As @Stedy's investigations a while ago found that some of the signals outputed were too strong for LCD TV's (but ok on original CRT) and he recommended adding resistors to many of the signals

Can you link me to any evidence for the signal strength issue? I believe I know what the problem is, but I haven't measured the voltage levels yet.

Usually the chip has dynamic range expansion enabled. This way over saturates the Luma, ie brightness. The chip does all of it processing in YUV and converts back to RGB for that output type.

My settings should have the output DRE turned off, but there are other things that still need optimising. I'm just trying to determine if the problem is circuitry, ADC/DAC or Digital Processing settings.

[Higgy]

@dooklink - this thread is moving quite fast so things can get missed. @Stedy is the 'boffin' so I can't help, although I can link to some of his findings:

http://www.ianstedman.co.uk/Amiga/am...iga_scart.html
http://ianstedman.wordpress.com/gbs-82xx-experiments/

Hopefully @Stedy will spot your post and have some direct comments.
Thanks.

[Stedy]

Originally Posted by _dx

Hi guys!
I gonna participate in custom firmware development for GBS8200.
The main question is: where is i2c registers information in datasheet http://file.yizimg.com/420856/2013041613555642.pdf
Somebody have the information about configuration registers of Trueview5725?

May be someone have another(full) version?

You need the programming manual, there's a link here:

http://shmups.system11.org/viewtopic...52172&start=30

The TVIA 5725 uses I2C address $2E. Using the Raspberry Pi I2C commands, they support a 7 bit I2C address (as bit 0 is read/write indication) so you divide the address by 2 to get $17 for the device. If, like me, you have a rev 1 RPi, use I2C bus 0 not 1 in the commands/scripts.

The TVIA5725, like other video devices requires multiple banks, you write to offset $F0 the bank address, 0-5, each bank has a set of registers.

The manual is scattered. Some related settings are on different banks so you need to bank switch.

With regard to how to configure the device, that has no easy answer. Look at the number of functional blocks in the device.

I think we need to concentrate on a few blocks:

1. ADC sampling and sync recovery
2. Scan rate conversion for 50-60 Hz
3. De-interlace (Bob/weave selection)
4. 3D noise reduction
5. Scaling
6. Video enhancement (maybe)

I'm concentrating on item 1, The falling edge of the sync is absolutely critical for timing of a video device. Any jitter screws up the whole chip. It's used for standard identification, clock generation and sampling of the video. I currently use a 680 ohm resistor to couple the sync in, it works but it slows the sync signal, I will convert the Amiga 5V CMOS signal to LVTTL soon.

The Amiga and a number of other consoles, the Megadrive was mentioned, have Dc coupled video outputs. Black level is 0.6-1V approximately. Most devices expect black level to be 0V. A 100nF capacitor at the ADC helps but you can get droop issues. Coupling the video output with 220uF capacitors can help but you can adjust black level clamping

Once the device is synchronised, and correctly samples the video, I can move on. There will be an element of 'suck it and see'.

[Stedy]

Originally Posted by dooklink

Can you link me to any evidence for the signal strength issue? I believe I know what the problem is, but I haven't measured the voltage levels yet.

Usually the chip has dynamic range expansion enabled. This way over saturates the Luma, ie brightness. The chip does all of it processing in YUV and converts back to RGB for that output type.

My settings should have the output DRE turned off, but there are other things that still need optimising. I'm just trying to determine if the problem is circuitry, ADC/DAC or Digital Processing settings.

The first problem is with the use of an LM1881/sync strike type device. Continued use will eventually destroy the GBS-8220, it'll take 18 months to two years though. The LM1881 outputs 5V CMOS levels. The GBS-8220 accepts 3.3V LVTTL levels, when you exceed the device power supply (3.3V) with an input signal, you start damaging the microchip.

Reducing the Amiga Composite sync (5V CMOS amplitude) from 4.8V to <1V cured over 95% of problems with SCART TVs. A SCART TV expects a 1V signal.

The GBS-8220 (TVIA-5725) expects a TTL signal, logic 1 is >2.0V but less than 3.6V max, logic 0 is <0.8V and no less than -0.3V.

The black level offsets are important, from there all scaling takes place. I have not seen a gain control register yet but some chips can amplify or attenuate the incoming video to maintain a 100 IRE level. If it does not, we may need to play with external components or settings. I need play time

Have you played with the decimation filters?

The RGB YUV block can affect the output video if the wrong polynomials are used. Are there any particular colours that look incorrect?

With most video issues, I need to see the issue, on a calibrated monitor, to comment. Photographing a monitor is not easy.

Sorry if some of my answers are a bit vague, I'm still studying the TVIA-5725 chip.

[dooklink]

Originally Posted by Stedy

The first problem is with the use of an LM1881/sync strike type device. Continued use will eventually destroy the GBS-8220, it'll take 18 months to two years though. The LM1881 outputs 5V CMOS levels. The GBS-8220 accepts 3.3V LVTTL levels, when you exceed the device power supply (3.3V) with an input signal, you start damaging the microchip.

Reducing the Amiga Composite sync (5V CMOS amplitude) from 4.8V to <1V cured over 95% of problems with SCART TVs. A SCART TV expects a 1V signal.

The GBS-8220 (TVIA-5725) expects a TTL signal, logic 1 is >2.0V but less than 3.6V max, logic 0 is <0.8V and no less than -0.3V.

The black level offsets are important, from there all scaling takes place. I have not seen a gain control register yet but some chips can amplify or attenuate the incoming video to maintain a 100 IRE level. If it does not, we may need to play with external components or settings. I need play time

Have you played with the decimation filters?

The RGB YUV block can affect the output video if the wrong polynomials are used. Are there any particular colours that look incorrect?

With most video issues, I need to see the issue, on a calibrated monitor, to comment. Photographing a monitor is not easy.

Sorry if some of my answers are a bit vague, I'm still studying the TVIA-5725 chip.

Yep, I know exactly what your talking about. I'm no video engineer so it's good to. Have your input.

I'm wondering what to do about the 5v sync input. I'm thinking a revision of the sync strike will be needed. Composite and SOG use a negative 0.3V sync don't they? Is that what pure Csync should be?

I feed the 5V sync into my 15Khz BVM, would that cause damage as well?

The 5725 does have gain and bias registers for the ADC.

The Input and sync recovery is probably the most complex part of the chip. Your knowledge here will help greatly. The decimation filters are on in the OFW set for 4x oversampling and then averaging the result. I haven't tried disabling it yet.

I have enabled 4x oversampling on the output with interpolation, which is good.

There is a YUV conversion matrix for input and output, but I don't know if you can change them. Their is also Dynamic Range Expansion on the output.

I've done a lot of research and found some usefull info:
1. Full 5725 registers definition(http://electronix.ru/forum/index.php...post&p=1298514)
2. 5725 programming guide http://img.docin.com/players/DocinVi...ile.vonibo.com
So we have all possible information about 5725 and can do whatever we need with this chip.

But there is a small problem with MYSON MTV230M... It can be programmed via ISP only with firmware support.
So custom firmware developement is a complicated task because one wrong move and MTV230 is locked.

Here is the datasheet http://www.keil.com/dd/docs/datashts/myson/mtv230m.pdf
On page 16

The two Flash memories (OSD Flash and Code Flash) can be programmed by a specific WRITER in parallel
mode, or by IIC Host in serial mode while the system is working

As you can see there is some parallel WRITER! And we need it!
But i can't find any info about this WRITER

May be you guis have some ideas?

We can use any popular controller(or even rasp.PI) to configure 5725 but i like MTV230M!
It can draw OSD and it is already on board! We need to develop custom firmware with extended OSD options and then anyone can update it via ISP mode and be happy with GBS8200.

The only one problem is MTV230M parallel programmer needed during firmware development!

Add:
Found two programmers. both are VERY expensive for me
http://www.keil.com/dd/chip/3272.htm

Add2:
LabTool-48UXP supports MTV230M in ~$350 with shipment in Russia
Add2:
http://www.aliexpress.com/item/Free-...984434156.html

[dooklink]

@_dx Thanks for your input. That register list is what I've been missing. I've had the programming guide for a while.

If you don't know, I've got a link to datasheets in a Google Drive folder on my original thread on the shmups forum at http://shmups.system11.org/viewtopic.php?f=6&t=52172

I've already got assembly running on the MTV230 via the ISP function which is entered by shorting port 8 on the GBS boards. The ISP is enabled with the original firmware, and is not locked. So I've already dumped the OSD flash and two different versions of the OFW. I've also found it hard to find any info on parallel writers. I've already bricked one board from erasing the flash and not having access to ISP.

I'm currently working on creating a qusi bootloader and getting serial comms working on the UART port.

[dooklink]

@Stedy

Just fitted a 10k and 22k resistor devider on my LM1881 output. Measured at just below 3V, works fine .

Originally Posted by dooklink

I'm currently working on creating a qusi bootloader and getting serial comms working on the UART port.

I don't think you need this. Entering ISP mode is very easy and its done in few steps. But quality of bricked devices will be to high anyway.. I have some experience... Definitely we need parallel writer to develop firmware from scratch.

I am ready to develop custom firmware with almost any possible 5725 features and settings available from OSD, but i need community donate to buy parallel writer.

If anybody interested in this project and ready to donate - please let me know.
I need about $350. Or LabTool 48UXP programmer(may be someone have one)

[dooklink]

Originally Posted by _dx

I don't think you need this. Entering ISP mode is very easy and its done in few steps. But quality of bricked devices will be to high anyway.. I have some experience... Definitely we need parallel writer to develop firmware from scratch.

I am ready to develop custom firmware with almost any possible 5725 features and settings available from OSD, but i need community donate to buy parallel writer.

If anybody interested in this project and ready to donate - please let me know.
I need about $350. Or LabTool 48UXP programmer(may be someone have one)

No I do not think you need the parallel programmer. I've loaded code at the end of the OFW by creating a jump branch.

There is a brick risk with the initial flash, but after that you don't need to mess with ISP code path. You can just write after the ISP check.

I've disassembled the original firmware in IDA Pro and have found the ISP initialisation and can simulate assembly in S51 simulator in the SDCC pack, then flash it to the chip no problems. You could buy 10 GBS boards for $350.

[voyager_1701e]

This thread has gotten quite technical, certainly way over my head!

I know my question is rather away from the current line of the thread....But....Ive got 3 V3 GBS8220 which are marked 5v input. Stupidly, and accidentally, I connected 12v to one of them and destroyed it. The GBS8200, however, notes on the printed board 5v-12v. Anyone run their 8200 off 12v?