alexanrs wrote:

In CRTs, yes, it is simple, but on LCD panels the brightness controls only affect the intensity of the backlight, so there is no way to improve it.

Thats not true for every LCD. From the TFTs I used most have additional menu options that show when the analog VGA-In is used as source.

This is slightly adding some flames, but honestly I think if you're going the effort to use vintage hardware to play old DOS games you should go the extra mile to use a CRT to see the game how it was intended.

LCDs present things a little differently and stretching usually isn't all that great. And playing it boxed in the center of the screen isn't too appealing either.

Maraakate wrote:

Has anyone tested this on a CRT?

S3 Trio should look fine on a CRT and I never had to adjust monitor brightness for that.

it's only some problem with LCD computer montiors. Have not tested on TVs.

What it looks like is kind of like a crahsed Quake3 to desktop leaving the screen bright, except it'll even affect pure dos and the BIOS setup 🤣

I actually do have a CRT stored, but not enough space to assemble it. Currently I'm sharing my main PC's monitor with my retro ones (main - display port, athlon64 - hdmi, others - VGA). Once I move to a bigger place I'll have my CRT fix.
This gamma discussion actually reminded me of something - the default gamma for DOS glide games (tomb raider comes to mind) is atrocious on LCD as well, but that can be corrected through environment variables. A voodoo+this virge would probably work well on a CRT using the default settings.

Voodoo Graphics always defaulted to a ridiculously high gamma (1.7) even on CRTs. This natrually, lead to complaints about other non-3dfx cards being "too dark" (GLQuake and Quake2 on Riva128/TNTs in particular) back in the day.

Voodoo2's default were a little more sane (1.3) but that did not apply to DOS games or MiniGL however.

Maraakate wrote:

Has anyone tested this on a CRT?

Yes. It's clearly visible on POST screen. While working with windows you don't observe it because it's a whole-screen thingy.

Maybe this problem can be fixed by replacing 3 SMD resistors next to vga connector with higher value?

adalbert wrote:

Maybe this problem can be fixed by replacing 3 SMD resistors next to vga connector with higher value?

Most likely that would be the wrong place. The DAC output filter is designed to be driven with certain impedance, and if you change impedance, the filter characteristics would change. Also it would cause impedance mismatch when driving the coax video cables. In short, that would just make picture quality worse.

And if it happens to be a current-mode DAC, less load would actually make it brighter.

If the video voltage is larger than 0.7Vpp, then next best thing would be to lower DAC reference voltage/current if possible.
It is still possible the video BIOS just configures the DAC output with too large voltage, so either way the BIOS or some resistors need to be hacked, and who knows which one is at fault, the hardware or the software.

I just tried my best working GX2 on my CRT, and it is definitely too bright. Reducing the brightness to 0 (I have it defaulted to 11 since 0 is too dark normally after some tests) didn't make it look any better. I also changed the input level from 0.7V to 1.0V and it still doesn't look right. Dimmer, yes, but it's just wrong looking.

all these three display the same symptoms as descided in the thread.

Those cards use 86c368 and I can't find a datasheet for them (but I did find 86c765 datasheet..)

I don't have these cards myself but can help you debug them.

Based on the 86c765 datasheet, the DAC circuitry looks like any other video DAC. The 86c368 based cards might be similar to debug if it is a hardware design thing.

The point is to find a pin called VREF and measure it has correct reference voltage for the DAC. Then you need to find a pin called IREF which sets the DAC reference current with a resistor, to see if it has correct resistor value. If these two result into correct DAC output current range, then it is a software issue.

So, can anyone with a too bright card give a list of what video chip it uses, preferably with a link to a datasheet, and a link to a picture of the card with such accuracy that resistor values can be seen. The vgamuseum.info has a lot of cards for example.

Interesting topic.

I have a ELSA Winner 1000 Trio/V, Trio64 2MB PCI, that's about as sharp as can be.

Will get a ELSA Winner 1000/T2D soon (Trio64V2/DX), and a no-name Trio64 PCI card. Curious if these work correctly (ELSAs are usually quality cards).

My bright card is based on a 86C375 vídeo chip.

All three of my GX2s use 86C357.

Here's pics of them. http://imgur.com/a/HekGY

It isn't chip related. I have S3 Trio 64, Virge and Trio 3D cards exhibiting the same symptoms.

leileilol wrote:

Maraakate wrote:

Has anyone tested this on a CRT?

S3 Trio should look fine on a CRT and I never had to adjust monitor brightness for that.

Well I did. The colors are washed up on my CRT compared to other cards. No amount of adjusting the monitor will compensate. Only adjusting by software (S3 Color in display properties) will make any difference. I still believe flashing another bios on the cards will fix the problem.

Where did you get that S3 Color thing? I installed a bunch of different drivers on my Virge DX in Windows 95 and none of them gave me any tabs in display properties oO Maybe they show up on Windows 98 only?

Well for the chips mentioned above, I can't find datasheets for them, so I can't tell what they use as reference voltage or how the resistor sets DAC current.

However, it did occur to me, that many cards are capable of Sync-On-Green (SOG) signaling. I am not sure if these chips support it, but if they do, what if the card just enables SOG (or even to all channels), so that the sync portion is 0.3Vpp and video 0.7Vpp, totaling to 1Vpp signal? This would also make areas meant to be black into very dark gray, while making whites too bright. Then loading drivers in Windows could remove the sync so it fixes the picture. And in this case fiddling a bit or two in BIOS could also disable it.

Can somebody with an oscilloscope take a look at the signal? Or can somebody with a SOG-capable monitor and a spare VGA cable and cut the HSYNC and VSYNC pins off to see if the card outputs with SOG?

I just fixed this exact problem with a S3 Trio 64 V2,DX 86C775 PCI card I got off Ebay last week. As others have reported it is fine in Windows but
the black level was very grey in Bios post screens and of course Dos. As Phil mentioned at one point it looks like they have used
a TV black level of about 16 instead of Zero.

Getting back to the fix. I downloaded a few bioses off the web for this s3 chipset, mine had a Winbond 27E256 eraseble Rom chip.
2 of the bioses I downloaded were 64k (27c512) so they were no good, but there was a 32k revision 2 Asus bios, mine having a 1.04 bios.

I erased the Rom in my Willem programmer and flashed the Asus bios, then to my amazement booted the computer and blacl levels now perfect.
This card is a good performer with a VESA video speed of over 22000 in Speedsys, more than double the Trio32 I had installed.

This could be the answer for all s3 cards with this issue, find a bios that isn't affected and flash that if You have the ability.

(my previous post has not been sumbitted, dunno why)
For S3 Trio64V2/DX or GX you can try this video BIOS (in attachment) that does not have brightness issue. It is also capable of VBE 2.0 support (no more TSR's like UNIVBE or S3VBE20 in memory! well, almost 😀)

but be aware of card with this bios + Intel 440LX\EX or 440BX\ZX motherboards, seems that they crash Windows.

That's great news! Well done everyone 😀