Bad result from RGB mod

CMS

Previously A04_msh.
Joined
Aug 15, 2006
Posts
224
Hi,
I just got a RGB modification on me AES and the result was not as good as I hoped for. It's actually worse than before. When there's a lot of yellow on screen it gets overexposed and other colours like blue laser that used to be bright is now just under saturated.

Was this modification done bad or anyone else have a similar experience?
 

skate323k137

Professional College Dropout
10 Year Member
Joined
Jan 7, 2013
Posts
4,220
What revision AES? Who did the mod? What kind of display are you using? What kind of cable? All of these will help us help you.
 

CMS

Previously A04_msh.
Joined
Aug 15, 2006
Posts
224
What revision AES? Who did the mod? What kind of display are you using? What kind of cable? All of these will help us help you.

Oh, of course.

Rev 3-3.
Mod done by Console Passion UK.
Sony LCD.
Scart.

Thanks
 

skinny503

NAM-75 Vet
Joined
Sep 22, 2013
Posts
1,003
The cable you have, can you open up both ends and take pictures of the connections?

How are the color other then yellow going bonkers?
 

ggallegos1

Cholecystectomy Required.,
10 Year Member
Joined
Oct 1, 2013
Posts
5,157
Pics and video would be good too. Any chance the modders let you in on exactly what they did?
 

skinny503

NAM-75 Vet
Joined
Sep 22, 2013
Posts
1,003
Doesnt the AES require very little modification to output RGB? I would say open up the console and get us some pictures!
 

CMS

Previously A04_msh.
Joined
Aug 15, 2006
Posts
224
Thanks for all ideas and possible solutions.

Pics and video would be good too. Any chance the modders let you in on exactly what they did?

Here's a comparison, right image is from google and that's how my screen looked before the mod. Left is my new screen.

View attachment 13135

LCD

There is your problem.

It shouldn't be the problem, it worked great before the mod two weeks ago on that same TV. I have had this console for more than 15 yrs and used older TVs with almost same video quality output.
 
Last edited:

skinny503

NAM-75 Vet
Joined
Sep 22, 2013
Posts
1,003
Oh wow that is a big difference! the image from google, do you know if its a picture of a cab or is it a screen shot from a emulator? To me it looks like the RGB levels are to high, but lets wait for someone else to come in and confirm
 

CMS

Previously A04_msh.
Joined
Aug 15, 2006
Posts
224
Oh wow that is a big difference! the image from google, do you know if its a picture of a cab or is it a screen shot from a emulator? To me it looks like the RGB levels are to high, but lets wait for someone else to come in and confirm

Probably an emulator... but that's how the game looked on my AES before the mod.
 

CMS

Previously A04_msh.
Joined
Aug 15, 2006
Posts
224
Doesnt the AES require very little modification to output RGB? I would say open up the console and get us some pictures!

Here's a picture: View attachment 13200

Thought these "cylinders" where supposed to be removed and replaced with connecting wires on the underside?

Are there different ways of how to do a bypass mod?
 

Kyuusaku

B. Jenet's Firstmate
Joined
Mar 26, 2003
Posts
419
I'll try to explain the idiocity of most RGB mods with basic-intermediate tech talk:

First of all RGB signals transmitted to TVs are standardized, signals transmitted to arcade monitors are not. When you do a Neo Geo "bypass" mod (or many other popular mods that were not conceived by someone remotely familiar with video signals) you're typically wiring the video DAC's high-impedance (resistance) output directly to the TV input (low-impedance). This is how the MVS connects to a high-impedance load such as an arcade monitor, but it is completely incorrect to connect the DAC directly to a low-impedance load such as a TV or monitor. When you connect a low-impedance load to a high-impedance output, you get a loss in signal amplitude because the output impedance (here around 200 ohms) and the TV's load impedance (75 ohms) form a voltage divider that attenuates the signal. This makes the signal dark, because again the signal levels are standardized (0.0V is black, 0.714 is white).

The only reason this bypass mod works at all for people is because the DACs output a signal 3-5x larger than the signal arriving at the TV is supposed to be, and many TVs have the ability to crudely compensate for bad signals. If the signal level is too high (as I suspect in the pictures above), you get brightness/contrast issues.

The NG's DAC is just a logic chip functioning as a binary weighted voltage source, so the output's full range is ~3 Vpp, the same voltage range that the logic chip's outputs swing. The correct TV level again is 0.714 Vpp... Huge difference! This is a big part of why the colors are wrong--The TV can only compensate so much. Since RGB signals don't have any sort of frame of reference a TV is pretty much working blind, it can only put some basic gain control into effect, and not every display can even do that, leading to inevitable clipping/distortion (usually oversaturated colors) and brightness issues (low contrast).

Also since the DAC is created using a logic chip which isn't very good at sourcing current (it's meant to drive other logic chips often with 100k+ ohms input impedance), and because they don't want the output's current to have any influence over the voltage (creating non-linearity), they chose resistor values that were just low enough to reasonably drive an arcade monitor. Later with the AES the DAC's load became a voltage-divider that divides the 3 Vpp signals down to the correct 0.7 Vpp and feeds them into the Sony CXA RGB encoder (this is also a high-impedance load so its presence minimally affects the DAC).

When you directly connect the TV to the DAC, the 75 ohm load (low impedance) attempts to draw more current from the chip than it can provide, which drops the voltage = more non-linearity because the DAC can't supply the correct voltage, plus this puts a strain on the chip was wasn't designed for this. All bad. The DAC is only "strong" enough (it can only source/sink so much current) to drive a high-impedance arcade monitor (1k ohms) or the RGB encoder (probably 10k ohms). Again a TV is 75 ohms, 12 times too low. 75 ohms is standardized across the video industry and is essential for both amplitude compatibility between devices and also to maximize signal integrity for long cables (minimize reflection/ghosting, ringing, maximize power transfer). Some further reading (mathy): "impedance matching", "characteristic impedance", "transmission lines".

Putting it all together:

The only way to properly drive a TV (or most computer monitors with 75 ohm termination, or broadcast monitor without the option to disable the 75 ohm terminators) is to take the 3 Vpp DAC signal, divide it to 0.7Vpp and send it through a video amplifier with a gain of 2 (the CXA does this internally) so the final output is 1.4 Vpp, then send that through a 75 ohm resistor out to the TV so that the 75 ohm R at the output and the 75 ohm R at the display divide the signal to the precise 0.7Vpp amplitude desired at the TV input. If the video amplifier has too large of a DC offset for the amplifier to comfortably handle, series ac coupling capacitors can remove the offset from being delivered to the load, and reduce the current requirement. Alternately since the DAC's direct output is very large at 3 Vpp, instead of dividing it to 0.7 Vpp you can divide it directly to 1.4 Vpp, and send that through an amplifier with a gain of 1 (so its just a voltage buffer), then again out the 75 ohm resistor to achieve the right signal amplitude. This way has the benefit of not needing voltage amplification, which relaxes bandwidth requirements and improves noise figures.

The CXA already does it the first way, and it's internal to every Neo Geo console. The video quality isn't as good as it could be however because the amplifier in the CXA has limited bandwidth (as does everything), probably around 6 MHz. To achieve sharp pixels with the Neo Geo you need ~10 MHz of bandwidth, and this can be achieved with any modern video amplifier such as the ever popular THS7314. If you have a high-end display it wouldn't hurt to use a HD video amplifier (typically 30 MHz cutoff), these can be purchased for only a couple dollars now.

Do it right and it will work every time on every display... Obviously few modders understand what they're doing, and that's why they're all doing it wrong.
 
Last edited:

Elgen

New Challenger
Joined
Jun 3, 2012
Posts
63
Wow! That's got to be the best explanation I've ever heard on this subject };-P
Do you know if this applies to ALL arcade PCBs, or just Neo-Geo?
If so, doesn't that mean, that almost every SuperGun out there is made "wrong"?
Are CRTs (for home-use) in general more tolerant than modern TVs in this regard? In my own (home built) SuperGun, I only use a decoupling cap (for eliminating any DC off-set) and a pot (to crudely level the colours amongst each other) on each colour channel; and colours are normally fine when using CRTs (don't have any of these modern "thinggies" in my house). Have I just been lucky?

Again thnx for the fantastic explanation, hope it's ok with the extra questions, and that they are not too noobish... just eager to learn on this subject };-P

Loads of kudos from me, m8! };-P
 
Last edited:

Kyuusaku

B. Jenet's Firstmate
Joined
Mar 26, 2003
Posts
419
If so, doesn't that mean, that almost every SuperGun out there is made "wrong"?
The ones which wire RGB JAMMA pins to a SCART connector without active amplification, yes.

The original Asian superguns I believe do it correct--they attenuate the RGB signals with pots which load the DAC similar to an arcade monitor, into a high-impedance RGB encoder which performs the amplification (even for RGB outputs) so that the coax may be impedance matched.

It's my understanding that most people put potentiometers configured as variable resistors in series to provide the attenuation (generally without understanding that the 75 ohm load creates that attenuation so they choose pot sizes too large like 10k), or perhaps take output from the wiper not realizing that the TV puts a 75 ohm resistor in parallel which makes the attenuation curve non-linear. Both ways are bad because adjusting the pot changes the impedance load to the DAC, ideally you want to keep it at a constant 1k for consistency.

Also they will only give a reasonable picture if the DAC is capable of driving the load in the first place. The 74LS chips used for a DAC in most arcade games really are horrible at sourcing current so you really want to keep the load at 1k or above since it's what the games are designed for. The problem is that when the series pot is kept that large there is a huge amount of attenuation at the voltage divider between the huge source impedance and the small 75 ohm TV terminator. For this reason supergun adjustment is often a laughable balancing act between loading the DAC too much (providing the largest signal) or attenuating the signal at the TV too much.

With the "correct" way you should be able to set the pots once for typical 3 -> 0.7V attenuation and be able to reasonably view 99% of 80s-mid 90s boards without further adjustment. To get the attenuation perfect should only require slight turns to compensate for different DAC voltages, it wouldn't be dependent at all on the DAC's ability to supply current which varies chip to chip, output to output.

Lastly this direct connection method makes no attempt to impedance match the 75 ohm coax cable going to the TV. This isn't very critical due to the low bandwidth of 240p video, some people even dismiss it who should know better, but IMO it's stupid not to get the best performance out of your equipment. The high impedance output of the DAC (+ additional from the attenuator) is very bad at driving a coax cable which has capacitance it must fight with to get the signal across, the impedance + capacitance reduces bandwidth and blurs pixel edges. You also always want a low impedance connection between devices to get the best noise immunity. Ignoring these simple things makes a lot of projects touting excellent picture quality an engineering joke (especially since most engineers could care less about this stuff).

Are CRTs (for home-use) in general more tolerant than modern TVs in this regard?
Not really, modern TVs employ digital signal processing which allows them to compensate in ways that aren't feasible with analog circuits. Even conceptually basic components such as a voltage controlled amplifier (which are used to control the contrast setting from a menu) are really difficult to build with analog circuits so CRTs use them very sparingly. Modern TVs have no such limitations since a few hundred million multiplications/second to do the same thing costs practically nothing.

Have I just been lucky?
Yup, along with many many others. Doing it wrong is popular since the circuits often have few parts and since the consequences are often minimal they're often the ones that spread. There's only a problem when the bad circuit which "works" for one person, doesn't work for the next. Then you get TV compatibility lists instead of blaming the circuit lol
 
Last edited:

Xian Xi

JammaNationX,
15 Year Member
Joined
Dec 1, 2005
Posts
27,750
This is some great info. Should I be putting anything between the amp and the RGB socket?
 

Green Beret

Fio's Quartermaster
Joined
Feb 27, 2013
Posts
498
The ones which wire RGB JAMMA pins to a SCART connector without active amplification, yes.

The original Asian superguns I believe do it correct--they attenuate the RGB signals with pots which load the DAC similar to an arcade monitor, into a high-impedance RGB encoder which performs the amplification (even for RGB outputs) so that the coax may be impedance matched.

It's my understanding that most people put potentiometers configured as variable resistors in series to provide the attenuation (generally without understanding that the 75 ohm load creates that attenuation so they choose pot sizes too large like 10k), or perhaps take output from the wiper not realizing that the TV puts a 75 ohm resistor in parallel which makes the attenuation curve non-linear. Both ways are bad because adjusting the pot changes the impedance load to the DAC, ideally you want to keep it at a constant 1k for consistency.

Also they will only give a reasonable picture if the DAC is capable of driving the load in the first place. The 74LS chips used for a DAC in most arcade games really are horrible at sourcing current so you really want to keep the load at 1k or above since it's what the games are designed for. The problem is that when the series pot is kept that large there is a huge amount of attenuation at the voltage divider between the huge source impedance and the small 75 ohm TV terminator. For this reason supergun adjustment is often a laughable balancing act between loading the DAC too much (providing the largest signal) or attenuating the signal at the TV too much.

With the "correct" way you should be able to set the pots once for typical 3 -> 0.7V attenuation and be able to reasonably view 99% of 80s-mid 90s boards without further adjustment. To get the attenuation perfect should only require slight turns to compensate for different DAC voltages, it wouldn't be dependent at all on the DAC's ability to supply current which varies chip to chip, output to output.

Lastly this direct connection method makes no attempt to impedance match the 75 ohm coax cable going to the TV. This isn't very critical due to the low bandwidth of 240p video, some people even dismiss it who should know better, but IMO it's stupid not to get the best performance out of your equipment. The high impedance output of the DAC (+ additional from the attenuator) is very bad at driving a coax cable which has capacitance it must fight with to get the signal across, the impedance + capacitance reduces bandwidth and blurs pixel edges. You also always want a low impedance connection between devices to get the best noise immunity. Ignoring these simple things makes a lot of projects touting excellent picture quality an engineering joke (especially since most engineers could care less about this stuff).


Not really, modern TVs employ digital signal processing which allows them to compensate in ways that aren't feasible with analog circuits. Even conceptually basic components such as a voltage controlled amplifier (which are used to control the contrast setting from a menu) are really difficult to build with analog circuits so CRTs use them very sparingly. Modern TVs have no such limitations since a few hundred million multiplications/second to do the same thing costs practically nothing.


Yup, along with many many others. Doing it wrong is popular since the circuits often have few parts and since the consequences are often minimal they're often the ones that spread. There's only a problem when the bad circuit which "works" for one person, doesn't work for the next. Then you get TV compatibility lists instead of blaming the circuit lol
I made the mod and removed 5 caps, will be any damage to tbe board?
 

Kyuusaku

B. Jenet's Firstmate
Joined
Mar 26, 2003
Posts
419
This is some great info. Should I be putting anything between the amp and the RGB socket?
RGB in or out?

JAMMA edge RGB -> flying lead wires -> pot attenuator -> ac coupled video amplifier -> 75 ohm R -> 220 uF C -> coax to TV

I made the mod and removed 5 caps, will be any damage to tbe board?
I need a little more context. Which mod? Which 5 caps?

I'm sure it's not a good idea to remove random caps, they were there for a purpose.
 

Green Beret

Fio's Quartermaster
Joined
Feb 27, 2013
Posts
498
RGB in or out?

JAMMA edge RGB -> flying lead wires -> pot attenuator -> ac coupled video amplifier -> 75 ohm R -> 220 uF C -> coax to TV


I need a little more context. Which mod? Which 5 caps?

I'm sure it's not a good idea to remove random caps, they were there for a purpose.
I made the RGB mode.
On 3.3 revision there are 5 caps for video output. 3 for RGB 1 for sync and 1 extra arround these 4, not sure for what. All these 5 were changed in the past from 100uf to 470uf. Now I removed them and wired on the back according to MKL's photo except I didn't put any resistors on the 3 RGB wires.
Picture became great buf sound has noise.
 

Kyuusaku

B. Jenet's Firstmate
Joined
Mar 26, 2003
Posts
419
Those are ac coupling capacitors and they are necessary to lower the current requirements of the CXA chip's outputs. They have almost no effect on the image* (zero effect on sharpness) so there is no point in removing them. Removing them will only put stress on the CXA chip which may cause distortion or possibly though not likely, chip damage.
If you want to improve picture sharpness you can use a different RGB amp such as the THS chips with higher bandwidth. And if you dc couple amplifier's input, then it's not necessary to ac couple the output so the caps can be left out.


*without using gigantic capacitors (2200 uF or so) or an amplifier circuit with something called sag correction, the brightness in ac coupled video changes slightly from the left side of the picture to the right side. This isn't very noticeable so AFAIK no video games even take it into consideration.

Composite video's cap is really the only one that must be upgraded; the stock 100 uF is too small and is the reason most TVs will lose sync during black and white flashing effects in Samurai Shodown, KOF etc. 100 uF just isn't enough to keep the signal steady (same principle as sag effect) without wobble from the low frequency / high amplitude animation, and the black level dips past the sync threshold.
 
Last edited:

Xian Xi

JammaNationX,
15 Year Member
Joined
Dec 1, 2005
Posts
27,750
RGB in or out?

JAMMA edge RGB -> flying lead wires -> pot attenuator -> ac coupled video amplifier -> 75 ohm R -> 220 uF C -> coax to TV

Pot attenuator, should I be using a static attenuation or adjustable? I don't know if there was ever a JAMMA standard when it came to video output. How do I standardize the impedance load on the DAC across multiple boards?
 
Top