GBS 8200/8220 CFW Project

The place for all discussion on gaming hardware
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

True that. Also there's no VGA input.
The i2c bus will be available on the control chip and the EEPROM, soldering required.
Ah well, maybe I'll get one of those and see what can be done :p
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: GBS 8200/8220 CFW Project

Post by Syntax »

Would be sweet if you could intercept the dac and get RGBHV from the HDMI board.
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: GBS 8200/8220 CFW Project

Post by maxtherabbit »

I think I remember reading that scaler chip does output digital RGB...
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Syntax wrote:Would be sweet if you could intercept the dac and get RGBHV from the HDMI board.
I don't get it, sorry :p
The current chain is analog RGB + (digital) HV leave out of the 5725. Instead of the normal 75Ohm termination and out to the VGA connector, these signals are now double 75Ohm terminated and enter the HDMI transcoder.
This acts as if the receiving display was right next to the 5725, instead of behind 2 meters of VGA cable by the way.
maxtherabbit wrote:I think I remember reading that scaler chip does output digital RGB...
The 5725 can do digital output:
- 16bit YPbPr digital output with syncs
- 24bit YPbPr/RGB digital output with syncs

The pins for that are all on the left side on GBS boards. Soldering would be an expert job, so this isn't particularly useful here.
If they ever make a board with those I/O connected, this would be a pretty big improvement.
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

Odd, I can no longer compile the latest version of gbscontrol. I get this:

Code: Select all

Arduino: 1.8.5 (Windows 10), Board: "NodeMCU 1.0 (ESP-12E Module), 80 MHz, 4M (1M SPIFFS), v2 Lower Memory, Disabled, None, Only Sketch, 115200"

Build options changed, rebuilding all
C:\Users\AndehX\Desktop\gbs-control\gbs-control.ino:26:30: fatal error: WebSocketsServer.h: No such file or directory

 #include <WebSocketsServer.h>

                              ^

compilation terminated.

exit status 1
Error compiling for board NodeMCU 1.0 (ESP-12E Module).

This report would have more information with
"Show verbose output during compilation"
option enabled in File -> Preferences.
Ok nevermind, I didn't realise I needed to install another library. It's been a while since I updated.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

AndehX wrote: Ok nevermind, I didn't realise I needed to install another library. It's been a while since I updated.
Did you find out what to do by the comment above that line? I hope this is how others may help themselves as well :p
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: GBS 8200/8220 CFW Project

Post by Syntax »

The comment saved my ass last month, Thanks Rama.
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

rama wrote:
AndehX wrote: Ok nevermind, I didn't realise I needed to install another library. It's been a while since I updated.
Did you find out what to do by the comment above that line? I hope this is how others may help themselves as well :p
Yeah I saw it and realised I must be missing something, lol

Couple of pics from my projector of the latest firmware

Image

Image
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Neat ;)

I'm working on Input Formatter based horizontal image shifting right now (vertical is already implemented).
Progress is slow since I don't have any of the original formulas.
But when it works, we can shift the picture in all 4 directions without creating any glitches.
This allows for some tighter timings and a nicer picture from more oversampling as well :)
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

Is there any way to have a 480p component bypass mode where it just converts the signal to VGA?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

I haven't found one last time I looked.
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

rama wrote:I haven't found one last time I looked.
Could the feedback presets work with 480p component then?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Only the 1280x960 presets support ED and HD resolutions for now.
The reason is that I use the base presets and tweak some parameters to make it work.
Trying to support the other presets would blow up this code with conditionals that I won't be able to manage reliably.

There's several ways to tackle this problem.
- try and find a solution that directly uses the source sync (this works great if the source is 480p or higher, as seen in RGBHV bypass)
- create more presets for manual selection via web ui (this would require dropping ATmega328 Arduino support)
- possibly reduce the number of factors that need to be adjusted for any preset to be adapted to a ED / HD source

Is there a particular reason you want another preset / output resolution?
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

rama wrote:Is there a particular reason you want another preset / output resolution?
The reason is that the feedback presets look better on my monitor compared to the 1280x960 presets
so It would just be nice if my 480p component sources looked as good as it could be just like my 240p/480i sources.
rama wrote:There's several ways to tackle this problem.
- try and find a solution that directly uses the source sync (this works great if the source is 480p or higher, as seen in RGBHV bypass)
- create more presets for manual selection via web ui (this would require dropping ATmega328 Arduino support)
- possibly reduce the number of factors that need to be adjusted for any preset to be adapted to a ED / HD source
Why not just create another preset and let users who have an Atmega328 Arduino just delete presets they
Don't need? I'm certain most people don't need all of the ntsc and pal presets included. :wink:
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

I'd prefer to find a way that works for every MCU and on any preset.
Everything else is really just workarounds layered on top of each other. I know this is a dead end from earlier in the project.
I'll continue looking for something more general for ED/HD sources.
The other day I got further by getting disabled line doubling working.

In the meantime, there's many options to fine tune the working preset. It depends on what looks better in the lower resolution presets.
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

rama wrote:I'd prefer to find a way that works for every MCU and on any preset.
Everything else is really just workarounds layered on top of each other. I know this is a dead end from earlier in the project.
I'll continue looking for something more general for ED/HD sources.
The other day I got further by getting disabled line doubling working.

In the meantime, there's many options to fine tune the working preset. It depends on what looks better in the lower resolution presets.
That's fine. I'll just wait for when you find something that works for everyone. :)
rama wrote:The other day I got further by getting disabled line doubling working.
Could you please elaborate on this?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Line doubling is a function of the Input Formatter.
The unit makes sense of the incoming data from the ADC, defining how many lines there should be and how long they last.
All of that has to be programmed and there are tons of rules to follow, otherwise we only get garbage.
One of the rules is the line double feature on/off, which changes all the other parameters.
A feature like that usually means doubling or halving most timings, but here it seems to bypass some registers instead.
I wish I had a proper manual :p

Anyway, I got line double = off working and that in turn helped correcting all the other parameters (not running into green borders or corruption as much).

Attached is a "ntsc_feedbackclock.h" that may work with your source. Good luck!
Spoiler

Code: Select all

const uint8_t ntsc_feedbackclock[] PROGMEM = {
44, // s0_40
183, // s0_41
0, // s0_42
20, // s0_43
37, // s0_44
17, // s0_45
127, // s0_46
23, // s0_47
235, // s0_48
11, // s0_49
5, // s0_4A
0, // s0_4B
0, // s0_4C
42, // s0_4D
0, // s0_4E
44, // s0_4F
0, // s0_50
0, // s0_51
0, // s0_52
0, // s0_53
0, // s0_54
0, // s0_55
0, // s0_56
0, // s0_57
0, // s0_58
0, // s0_59
0, // s0_5A
0, // s0_5B
0, // s0_5C
0, // s0_5D
0, // s0_5E
0, // s0_5F
36, // s0_90
0, // s0_91
71, // s0_92
84, // s0_93
0, // s0_94
62, // s0_95
50, // s0_96
0, // s0_97
0, // s0_98
0, // s0_99
0, // s0_9A
0, // s0_9B
0, // s0_9C
0, // s0_9D
0, // s0_9E
0, // s0_9F
162, // s1_0
128, // s1_1
95, // s1_2
0, // s1_3
0, // s1_4
0, // s1_5
0, // s1_6
0, // s1_7
0, // s1_8
0, // s1_9
0, // s1_A
64, // s1_B
199, // s1_C
103, // s1_D
64, // s1_E
3, // s1_F
0, // s1_10
0, // s1_11
2, // s1_12
0, // s1_13
0, // s1_14
0, // s1_15
0, // s1_16
0, // s1_17
39, // s1_18
3, // s1_19
116, // s1_1A
0, // s1_1B
0, // s1_1C
0, // s1_1D
5, // s1_1E
0, // s1_1F
0, // s1_20
3, // s1_21
65, // s1_22
3, // s1_23
0, // s1_24
0, // s1_25
16, // s1_26
0, // s1_27
4, // s1_28
1, // s1_29
0, // s1_2A
0, // s1_2B
0, // s1_2C
0, // s1_2D
0, // s1_2E
0, // s1_2F
202, // s1_30
0, // s1_31
128, // s1_32
0, // s1_33
63, // s1_34
0, // s1_35
128, // s1_36
44, // s1_37
204, // s1_38
0, // s1_39
0, // s1_3A
0, // s1_3B
0, // s1_3C
1, // s1_3D
192, // s1_3E
0, // s1_3F
0, // s1_40
1, // s1_41
192, // s1_42
0, // s1_43
0, // s1_44
1, // s1_45
192, // s1_46
0, // s1_47
0, // s1_48
1, // s1_49
192, // s1_4A
0, // s1_4B
0, // s1_4C
1, // s1_4D
192, // s1_4E
0, // s1_4F
0, // s1_50
1, // s1_51
192, // s1_52
0, // s1_53
0, // s1_54
0, // s1_55
0, // s1_56
0, // s1_57
0, // s1_58
0, // s1_59
0, // s1_5A
0, // s1_5B
0, // s1_5C
0, // s1_5D
0, // s1_5E
0, // s1_5F
98, // s1_60
98, // s1_61
32, // s1_62
38, // s1_63
65, // s1_64
62, // s1_65
178, // s1_66
154, // s1_67
78, // s1_68
214, // s1_69
177, // s1_6A
142, // s1_6B
124, // s1_6C
99, // s1_6D
139, // s1_6E
118, // s1_6F
112, // s1_70
98, // s1_71
133, // s1_72
105, // s1_73
83, // s1_74
72, // s1_75
93, // s1_76
148, // s1_77
178, // s1_78
70, // s1_79
198, // s1_7A
238, // s1_7B
140, // s1_7C
98, // s1_7D
118, // s1_7E
156, // s1_7F
169, // s1_80
46, // s1_81
5, // s1_82
4, // s1_83
0, // s1_84
12, // s1_85
202, // s1_86
0, // s1_87
0, // s1_88
0, // s1_89
0, // s1_8A
0, // s1_8B
0, // s1_8C
0, // s1_8D
0, // s1_8E
0, // s1_8F
255, // s2_0
3, // s2_1
204, // s2_2
0, // s2_3
0, // s2_4
0, // s2_5
0, // s2_6
0, // s2_7
7, // s2_8
0, // s2_9
140, // s2_A
4, // s2_B
220, // s2_C
0, // s2_D
0, // s2_E
72, // s2_F
32, // s2_10
128, // s2_11
142, // s2_12
0, // s2_13
0, // s2_14
0, // s2_15
140, // s2_16
0, // s2_17
11, // s2_18
4, // s2_19
0, // s2_1A
16, // s2_1B
0, // s2_1C
0, // s2_1D
0, // s2_1E
16, // s2_1F
80, // s2_20
2, // s2_21
4, // s2_22
15, // s2_23
0, // s2_24
0, // s2_25
76, // s2_26
12, // s2_27
0, // s2_28
0, // s2_29
0, // s2_2A
0, // s2_2B
0, // s2_2C
0, // s2_2D
0, // s2_2E
0, // s2_2F
0, // s2_30
0, // s2_31
52, // s2_32
127, // s2_33
153, // s2_34
56, // s2_35
3, // s2_36
11, // s2_37
4, // s2_38
100, // s2_39
128, // s2_3A
4, // s2_3B
143, // s2_3C
0, // s2_3D
0, // s2_3E
0, // s2_3F
50, // s3_0
87, // s3_1
211, // s3_2
32, // s3_3
48, // s3_4
131, // s3_5
5, // s3_6
6, // s3_7
112, // s3_8
0, // s3_9
117, // s3_A
224, // s3_B
1, // s3_C
249, // s3_D
113, // s3_E
31, // s3_F
86, // s3_10
67, // s3_11
11, // s3_12
236, // s3_13
169, // s3_14
0, // s3_15
255, // s3_16
179, // s3_17
51, // s3_18
1, // s3_19
0, // s3_1A
9, // s3_1B
0, // s3_1C
0, // s3_1D
16, // s3_1E
1, // s3_1F
14, // s3_20
2, // s3_21
12, // s3_22
2, // s3_23
48, // s3_24
3, // s3_25
0, // s3_26
207, // s3_27
38, // s3_28
7, // s3_29
1, // s3_2A
147, // s3_2B
224, // s3_2C
47, // s3_2D
32, // s3_2E
240, // s3_2F
64, // s3_30
58, // s3_31
127, // s3_32
0, // s3_33
0, // s3_34
128, // s3_35
30, // s3_36
41, // s3_37
0, // s3_38
0, // s3_39
254, // s3_3A
1, // s3_3B
1, // s3_3C
0, // s3_3D
128, // s3_3E
0, // s3_3F
7, // s3_40
3, // s3_41
0, // s3_42
0, // s3_43
248, // s3_44
255, // s3_45
249, // s3_46
24, // s3_47
249, // s3_48
16, // s3_49
249, // s3_4A
32, // s3_4B
249, // s3_4C
10, // s3_4D
27, // s3_4E
30, // s3_4F
48, // s3_50
0, // s3_51
112, // s3_52
8, // s3_53
36, // s3_54
10, // s3_55
139, // s3_56
0, // s3_57
26, // s3_58
0, // s3_59
0, // s3_5A
26, // s3_5B
0, // s3_5C
196, // s3_5D
63, // s3_5E
4, // s3_5F
4, // s3_60
155, // s3_61
128, // s3_62
9, // s3_63
233, // s3_64
255, // s3_65
127, // s3_66
64, // s3_67
210, // s3_68
13, // s3_69
216, // s3_6A
255, // s3_6B
63, // s3_6C
0, // s3_6D
0, // s3_6E
0, // s3_6F
0, // s3_70
8, // s3_71
0, // s3_72
188, // s3_73
5, // s3_74
0, // s3_75
0, // s3_76
0, // s3_77
0, // s3_78
0, // s3_79
0, // s3_7A
0, // s3_7B
0, // s3_7C
0, // s3_7D
0, // s3_7E
0, // s3_7F
128, // s4_0
48, // s4_1
0, // s4_2
0, // s4_3
48, // s4_4
119, // s4_5
2, // s4_6
51, // s4_7
1, // s4_8
153, // s4_9
17, // s4_A
127, // s4_B
0, // s4_C
116, // s4_D
0, // s4_E
6, // s4_F
0, // s4_10
144, // s4_11
7, // s4_12
1, // s4_13
150, // s4_14
5, // s4_15
0, // s4_16
0, // s4_17
6, // s4_18
0, // s4_19
80, // s4_1A
68, // s4_1B
0, // s4_1C
0, // s4_1D
0, // s4_1E
0, // s4_1F
0, // s4_20
65, // s4_21
1, // s4_22
31, // s4_23
255, // s4_24
255, // s4_25
207, // s4_26
255, // s4_27
255, // s4_28
31, // s4_29
0, // s4_2A
132, // s4_2B
30, // s4_2C
0, // s4_2D
0, // s4_2E
0, // s4_2F
0, // s4_30
0, // s4_31
0, // s4_32
0, // s4_33
0, // s4_34
0, // s4_35
0, // s4_36
180, // s4_37
0, // s4_38
180, // s4_39
0, // s4_3A
0, // s4_3B
0, // s4_3C
0, // s4_3D
3, // s4_3E
0, // s4_3F
0, // s4_40
0, // s4_41
104, // s4_42
0, // s4_43
255, // s4_44
255, // s4_45
7, // s4_46
255, // s4_47
255, // s4_48
7, // s4_49
0, // s4_4A
68, // s4_4B
0, // s4_4C
96, // s4_4D
40, // s4_4E
62, // s4_4F
192, // s4_50
0, // s4_51
0, // s4_52
0, // s4_53
0, // s4_54
0, // s4_55
0, // s4_56
0, // s4_57
204, // s4_58
0, // s4_59
204, // s4_5A
76, // s4_5B
0, // s4_5C
0, // s4_5D
0, // s4_5E
0, // s4_5F
248, // s5_0
0, // s5_1
21, // s5_2
59, // s5_3
0, // s5_4
0, // s5_5
63, // s5_6
63, // s5_7
63, // s5_8
90, // s5_9
90, // s5_A
90, // s5_B
0, // s5_C
0, // s5_D
1, // s5_E
0, // s5_F
0, // s5_10
146, // s5_11
88, // s5_12
3, // s5_13
0, // s5_14
0, // s5_15
47, // s5_16
6, // s5_17
129, // s5_18
145, // s5_19
0, // s5_1A
0, // s5_1B
0, // s5_1C
0, // s5_1D
128, // s5_1E
140, // s5_1F
10, // s5_20
2, // s5_21
16, // s5_22
0, // s5_23
64, // s5_24
0, // s5_25
8, // s5_26
0, // s5_27
0, // s5_28
0, // s5_29
6, // s5_2A
0, // s5_2B
0, // s5_2C
2, // s5_2D
0, // s5_2E
2, // s5_2F
0, // s5_30
47, // s5_31
0, // s5_32
16, // s5_33
5, // s5_34
21, // s5_35
0, // s5_36
4, // s5_37
4, // s5_38
7, // s5_39
4, // s5_3A
0, // s5_3B
0, // s5_3C
0, // s5_3D
16, // s5_3E
3, // s5_3F
12, // s5_40
16, // s5_41
0, // s5_42
39, // s5_43
0, // s5_44
64, // s5_45
0, // s5_46
80, // s5_47
0, // s5_48
4, // s5_49
0, // s5_4A
68, // s5_4B
0, // s5_4C
206, // s5_4D
0, // s5_4E
53, // s5_4F
3, // s5_50
2, // s5_51
0, // s5_52
6, // s5_53
0, // s5_54
80, // s5_55
1, // s5_56
192, // s5_57
0, // s5_58
16, // s5_59
0, // s5_5A
2, // s5_5B
0, // s5_5C
2, // s5_5D
0, // s5_5E
0, // s5_5F
0, // s5_60
0, // s5_61
0, // s5_62
15, // s5_63
0, // s5_64
0, // s5_65
0, // s5_66
0, // s5_67
0, // s5_68
0, // s5_69
0, // s5_6A
0, // s5_6B
0, // s5_6C
0, // s5_6D
0, // s5_6E
0, // s5_6F
};
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

rama wrote:Line doubling is a function of the Input Formatter.
The unit makes sense of the incoming data from the ADC, defining how many lines there should be and how long they last.
All of that has to be programmed and there are tons of rules to follow, otherwise we only get garbage.
One of the rules is the line double feature on/off, which changes all the other parameters.
A feature like that usually means doubling or halving most timings, but here it seems to bypass some registers instead.
I wish I had a proper manual :p

Anyway, I got line double = off working and that in turn helped correcting all the other parameters (not running into green borders or corruption as much).
Sounds cool! Hopefully someday you'll be able to get line doubling fully working.
rama wrote:Attached is a "ntsc_feedbackclock.h" that may work with your source. Good luck!
Thanks man. 8) I tested it out and besides the fact that it is shifted a little too much to the left causing the colors to go wrong(Easily fixed using the buttons on the GBS.) it works perfectly!
Higgy
Posts: 54
Joined: Thu Mar 15, 2018 1:29 pm

Re: GBS 8200/8220 CFW Project

Post by Higgy »

LackofTrack wrote: Thanks man. 8) I tested it out and besides the fact that it is shifted a little too much to the left causing the colors to go wrong(Easily fixed using the buttons on the GBS.) it works perfectly!
I thought the GBS buttons did not work with CFW running?
LackofTrack
Posts: 7
Joined: Thu Dec 07, 2017 5:07 am

Re: GBS 8200/8220 CFW Project

Post by LackofTrack »

Higgy wrote: I thought the GBS buttons did not work with CFW running?
In the CFW there is a line of code that you have to uncomment in order for the buttons to work. You also have to connect the buttons to the Arduino(4 wires).

Once you do that the buttons allow you to shift and stretch the picture both horizontally and vertically.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Yeah, this is useful for Arduino Uno and similar.
ESP8266 based boards have these functions in the web ui.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Hi Rama,

I did build it, I was waiting for the D1 board.
Image

I use a 5V power supply (in the carton box) for the moment until I have it working properly and pack all in a plastic box.
You can see I added the 100ohm resistor between S and GND pins.

This GBS is connected to an accelerated Amiga 1200 using this connection:
Image

Your program works very well, I could configure this PAL system to use the full screen width, not the entire hight to make proportions fine with 4/3 picture on a Dell 5/4 screen.
Image is rock stable, but I have a lot of interferences in form of vertical S like lines.

Image
The background is simply gray.

Do you know what causes this?

As far as I remember, I did have them too with the stock GBS but maybe less visible.

Edit: i had to make the RAM run at 109Mhz if I remember or I get green or purple horizontal line in 180 and 200+ frequencies.
Current sync:

Code: Select all

----
htotal: 2156
HS ST: 77
HS SP: 270
HB ST (display): 2154
HB SP (display): 480
HB ST (memory): 2064
HB SP (memory): 308
----
vtotal: 1002
VS ST: 990
VS SP: 993
VB ST (display): 960
VB SP (display): 8
VB ST (memory): 5
VB SP (memory): 6
Many thanks
Last edited by Rodrik on Mon Jul 30, 2018 1:31 pm, edited 1 time in total.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Yep, the problem is 2 things:
First, the C11 SMD capacitor causes the LDO to easily oscillate.
Feed the GBS 5V for power to mostly avoid this.
Better yet, remove the capacitor.

2nd, the Intersil video amplifier causes noise as well. Only a complicated hardware mod can fix this.
(See this thread, a few posts back :p )
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Ok, then it's what you explained me some pages ago.
Power is 5.02V from my multimeter.

I did remove C11 now and it's far better but still interferences.
So I think I will have to do this tiny surgery on R58 and R59.

Note: I also did this some months ago while I did use the the GBS without any modification:
Image

I hope it does not generate problems now. It did improve the image a little in term of stability on the stock GBS.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Weeellll, the shielding has a drastic effect on the memory bus.
The problem is that the effect can be good, nothing or disastrous :p

Your installation definitely covers the memory bus traces + a lot of the analog section (the ground and signals near the video inputs).
If you're unlucky, the noise enters that area and the effect is very bad indeed.

I would just remove the shielding to be safe.
No board with my software requires it, as my memory timings are much more relaxed.
(I tuned the timings with a scope on the Vcc of the memory controller and settled for values that cause the least noise.
Noise levels are 10x lower than on stock firmware.)

Since removing C11 helped, there's probably still some power supply noise issue.
You could order a few 10uF or 20uf 6.3V SMD capacitors and place them on the 3.3V line across the board.
A good place for them is side by side of the installed bypass caps:
Image
Image
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Ok many thanks for all these advices.

Shielding removed! Maybe a little more interferences but always good to remove maybe problematic things.

Then I will order some SMD 6.3V capacitors. Would you recommend 10uF or 22uF?
I will let the tiny surgery around R58/59 for later if required but will collect parts anyway.

Another question regarding input format. Can the GBS make a progressive image from the PAL Hires laced used by Amiga workbench and some desktop application (640*512 - 50Hz PAL)?
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

Rama, could you make a quick list of all the capacitors that would benefit from an added SMD bypass cap?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Rodrik:
If the remaining noise is faint and it appears to be very similar in all presets, then it's most likely the Intersil charge pump.
Honestly, you'll have an easier time just getting a $15 new GBS of any of the later designs :p

AndehX:
I'd recommend adding 5 x 22uF. 1 somewhere on the SDRAM, the other 4 around the Tvia chip.
You can use the underside of the board, or place them in parallel to the upperside existing caps.
Very useful as well is an extra X7R (high quality) 220nF / 10V or more SMD cap on the main buck converter input pin.
But this is a super tricky installation! Maybe better scrape off a littler solder mask on the trace and place it there.
Image

The other parts of the board don't need it, except for the first revision with the Intersil. That chip has noisy Vcc and can use more bypassing.

By the way, there should be next to none visible noise even on a stock board.
The C11 issue may have to be fixed, but other than that I mean.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Rodrik wrote: Edit: i had to make the RAM run at 109Mhz if I remember or I get green or purple horizontal line in 180 and 200+ frequencies.
108 Mhz is a good frequency but the best one is the "feedback clock".
Anything > 160Mhz is out of spec with my timings. Those frequencies are just there for completeness / because the chip offers them ;p

SMD caps on Ebay:
https://www.ebay.de/itm/50x-SMD-0805-Ko ... 2174614915

These are worth gold for bypassing Vcc on retro consoles!
Rodrik wrote: Another question regarding input format. Can the GBS make a progressive image from the PAL Hires laced used by Amiga workbench and some desktop application (640*512 - 50Hz PAL)?
This is supported, but it will have a slight interlace jitter, comparable to a TV set. Just try it out, no special settings required.
You may want to switch to the hires mode first, then reset the MCU, so it can pick up the slightly different horizontal timings.
This will avoid some jerks. (There's a control for this retiming in the web ui as well.)

Proper 2 field deinterlacing is possible with the chip but the result would have a more lag and may not look very nice.
See the original firmware artefacts for an example.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

rama wrote:108 Mhz is a good frequency but the best one is the "feedback clock".
Ok, thank you. "feedback clock" works very fine.
rama wrote:See the original firmware artefacts for an example.
Yes I remember, the mouse pointer was surrounded by pixels moving along with the pointer.
I will try to reset the D1 once the Amiga has booted in laced resolution.

Back on interferences, I will try to fix this V3 GBS board following your inputs whatever time/skill it requires, it's more about the journey than about the result and/or cost :D I like modding things.

I need a second GBS anyway and I have a second ESP8266 (a Lolin V3) already. What GBS revision would you recommend knowing the target screen is a Dell 1908FP (19" 5/4 1280*1024) with only VGA and DVI input. There are so many GBS in ebay and most without information on the revision except if you trust the pictures.
Post Reply