VIDEO DAC BRIGHTNESS ISSUE

[exxos]

If you look at the falcon voltages, it has 0.6V offset and 2.4V max. So would probably end up at 1.2V on the monitor I guess.

[PaulJ]

Took a shot of pin 6 of the video connector. If the blanking pulse level matches the video level as shown on scope the H4 video is best it can be. If the pot is changed to increase the voltage the H4 screen blacks out when then the blanking pulse is removed. When the blanking pulse voltage is adjusted toward zero the blanking pulse grows in amplitude and the H4 video gets more washed out. Maximum video quality is achieved when the blanking pulse is the same as the video pulses.

bbb.jpg (329.77 KiB) Viewed 3661 times

[exxos]

Can someone try just having those 3 pins disconnected, and maybe just wire a 100nf cap to 0v and measure the voltage on the cap... Also check effects on the video..

[PaulJ]

Can someone try just having those 3 pins disconnected, and maybe just wire a 100nf cap to 0v and measure the voltage on the cap... Also check effects on the video..

Ok, was very interesting. With the three colors floating with no cap the video is exquisite, the white is a truer white and the contrast is better. I'm going to leave it this way.

p3c.jpg (295.9 KiB) Viewed 3624 times

p1c.jpg (261.78 KiB) Viewed 3624 times

With a 33 pf cap connected to the open 3 analog inputs the video gets unusably dark. And the video green line shown at pin 6 of the video connecter as show below now sucks, see below image. Unable to measure voltage on cap as it causes horrible oscillations in the voltage and video. Didn't like scope on x1 and on x10 there's not much left to look at. Its extremely happy with nothing connected to those inputs.

p4c.jpg (239.39 KiB) Viewed 3624 times

[PhilC]

That's an interesting finding Paul, I may try it on my second board by cutting some of the tracks.

[PaulJ]

That's an interesting finding Paul, I may try it on my second board by cutting some of the tracks.

Its noticeably a better display. Tried it on a second monitor and it good there. Using the divider the white is a little dingy and the contrast is a smidge low. When running bench6 the colored fonts now pop and the contrast is better.

[PaulJ]

I've been reading the DAC pdf and for the typical application they show those analog inputs ac coupled. I'm guessing the chip wants to manage the dc level while letting you inject an ac signal.

p5c.jpg (84.27 KiB) Viewed 3618 times

Reading the analog input verbage it looks to me that the chip wants to establish levels stating it clamps the signal to a specific level. So when you force a dc level in it pulls everything off.

p6c.png (35.21 KiB) Viewed 3611 times

[exxos]

Yeah its why I thought putting the cap there instead of resistors may work better. Also my interest in the voltage on those pins to see if it was internally biased to about 1.6V. I was thinking for a while how to bias the inputs to the minimum DC output, but that would mean fighting against the chip which isn't right. So made me think it was maybe internally biased anyway.

Though odd a cap made it works.. though 33pF might not be enough. Something like 100nf would, or should, keep the line stable. I don't like to leave things floating even if the datasheet says its internally done. Though the pins lifted was how the V2 DAC prototype board was done. I don't think the V3 DAC board got built by anyone, and thats what I copied to the H4.

Its also possible it may need a seperate cap to the 3 inputs. It could be upsetting the internal bias assuming the RGB amps are not 100% the same voltage, which they probably are not. So I think that would be the thing to try next, if we can get those 3 caps on those pins and it still looks good, then I would be happy with that solution and we can leave it that way then.

[PaulJ]

Don't know @exxos , it didn't like the cap at all. looks like any attempt to input a dc value pisses off the internals of the chip. When you put the dc in via the divider and go from monitor to monitor it may need a tweak to regain the best presentation. When leaving analog inputs open it appears to self adjust for optimal presentation from monitor to monitor. It is noticeably better picture when left open. The white level is not slightly dingy and the contrast is better. I don't think your going to be able to defeat the clamping and when you feed in a dc component it messes up the internal operation of the chip. The analog inputs are really sensitive, meter really pissed it off, caps I tried made presentation horrible. Will be hard to beat the presentation when the inputs are open.

[exxos]

@PaulJ With a cap we are not feeding a DC level. The cap would maintain the internal DC bias level and should make it more stable. Thats pretty much what the circuit shows, only it has a resistor to GND as well which is likely just loading which we dont need in this case. if x10 probe is enough to mess it up, then its just not stable IMO. If we happen to route a 32MHz wire near those floating pins, it will trash the video. So we need to get some caps on there else we are building in a potential and likely point of failure.

Though only using 1 cap on all 3 pins may well cause issues.. even just wiring the 3 pins together will likely cause issues. If each RGB was slightly different, for argument's sake, 1.6V, 1.7V,1.5V and they all wired together, I can understand why it would make the colours muddy as each RGB signals is trying to bias and conflicting with other lines.

Obviously the datasheet shows 1 cap per analog input (470nf, but I think 100nf would be fine) . So this needs to be tried out.