[jcmccorm]

One more time into the lab with the Skyworth to take the cover off...


I had read Bosef's post about the Skyworth's Green output (via the VGA connector) having a DC offset so I wanted to check this out for myself and fix it.


I had suggested that it was caused by "sync on green" and that turns out to be the case. The bad news is that Green rides on a DC offset that's about 300mv higher than blue or red. The good news is that they aren't actually *using* the sync (that input pin to the DAC just stays high), so the Green remains at the elevated DC offset and never dips lower than the 300mv. The result of this is that if your projector/display is AC coupled on the RGB inputs, as my ECP is, and I believe everything else probably is too, this makes no difference whatsoever. Once the RGB gets past the DC blocking capacitor on your display device's input, all colors will have the same min/max amplitude.


I went ahead and disabled the sync anyway just to save a little extra current that the DAC was having to drive. To do this, you have to lift pin 12 of the DAC (ADV7123) on the Sage board and run a little wire from the lifted pin to GND. This put the Green output down with Red and Blue but unless you have the tools and/or courage to perform this mod, it's really unecessary.


Next I decided to increase the output of the DAC from the 660mv peak it was currently set at to a nice and snappy 1.0V peak. (I was checking this level using Avia's vertical gray bars pattern).


Why would I do this? Well, besides the fact that the soldering iron was already hot, it improves the signal to noise ratio of the video signal since any noise that the cable picks up is constant and so any increase in real video signal results in a better signal to noise ratio. Also, if you consider that the cathode voltage on a ECP3100's CRT swings maybe about 50V peak to peak (guess at this point but probably close), that means that my video gain in the projector has to be 71x (700mv RGB x 71). I'd much rather the gain be done by the DAC than the projector. Assuming the same contrast level, with the Skyworth doing 1Vpp on RGB, the projector's gain is only 50.


To increase the output level of the DAC, change the Rset resistor that's next to the ADV7123. Unfortunately I don't remember the reference disignator but I think it was R56. Anyway, it's the only resistor marked "511" near the DAC (510ohms). 351ohms will get you 1Vpp output so I added a 1.2Kohm through-hole resistor in parallel with the 510ohm surface mount.


Here's a picture of my green output with Avia's gray bars:

http://home.hiwaay.net/~jcmccorm/cary.jpg


It's just about 1Vpp and still nice and linear at the top (I was

worried about driving the DAC too hard).


I think the big Bang-for-the-buck mods to the Skyworth will have to come from firmware mods these days but messing with the DAC was fun and I'm happier about the output levels and tamed Green output.


Cary

 

[pham]

Thanks for posting this interesting info Cary. If Yahoo is still around, this should definitely be linked to in the Skyworth FAQ. Have you noticed much improvement in your picture after the mods?

 

[kjack]

Quote:

Originally posted by jcmccorm I went ahead and disabled the sync anyway just to save a little extra current that the DAC was having to drive.

Actually, you're probably not saving any current -- usually any DAC current that isn't output is simply diverted to ground, either internally or to a ground pin by the DAC output. Anyway, it is better to disable any offset since it simplifies the work the projector has to do.

Quote:

Next I decided to increase the output of the DAC from the 660mv peak it was currently set at to a nice and snappy 1.0V peak.

So now you just increased the DAC output current (for every DAC) by 50%... Hope the chip can handle the extra power dissipation!


The ideal should be about 0.7v, although a good input circuit can handle 2x that. There is a limit to the voltage that can be on the DAC output pin before it starts going nonlinear.


BTW, that looks like a VERY noisy DAC output for a green signal! For a few seconds I thought is was a composite signal until I saw the noise at the blanking level. Some nasty overshoots at the start of white bar also... Seems like a poor PCB layout or noisy analog power pins on the DAC.

 

[sspears]

Quote:

Next I decided to increase the output of the DAC from the 660mv peak it was currently set at to a nice and snappy 1.0V peak.

Sounds like you are trying to compete with Apex and Cinematrix! Actually the Apex is only 745 mV. Some display devices, especially the digital ones, don't like it that hot. You should check the pattern on Avia with the white bars. Make sure you still have both.


Have you measured the output of the interlaced signal? In IRE, I measured 62.6, which is VERY low. This was on both Y of YC and YPbPr.

 

[Bosef]

Great, great, great job, Cary.


Thank you so much, now i can *fix* my skyworth as well.


Let's wait for te firmware upgrade now!

 

[jcmccorm]

Pham,

Changing the sync input pin of the DAC (and therefore dropping the level of green to be equal to red and blue) didn't do anything noticable which is what I'd expected (and I'd be disturbed if it did).

I wasn't able to keep the 1Vpp output for long to check it out. I was able to reduce the contrast setting on the projector but viewed some unfamiliar material so I can't compare to anything (Fifth Element Superbit - even with my lowly ECP3101 I can see the difference between this and the old disk. Really nice.).

I put in Avia's resolution pattern and saw faint, dark horizontal streaks across the 6.75MHz patch situated to the right of the white horizontal strips in the resolution wedge just to the left. Not good. I turned down the projector's contrast and the gain on my Extron buffer (108Plus) and couldn't get rid of it. I plugged a 21" monitor into the Skyworth and couldn't see a problem there with that same image. I can only conclude that the Extron has a problem with the 1Vpp input (which it shouldn't!). Anyway, I went back to the original resistor alone (.7Vpp) and the problem went away so that's where it will have to stay.

The original resistor is 510ohms. 351ohms will get you 1Vpp so maybe something in between would work but since I'm getting my .7V I'm not going to mess with it again. Someone who believes the output of the Skyworth to be a little dim could certainly benefot from this mod though.


Kjack,

Noisy indeed!! I couldn't get rid of it! I had my scope ground attached to the GND blade of the component output connector on the inside of the case (nice big GND there). It was probably too far away from my probe tip and I should've made myself a good GND like I usually do by soldering a bit of solder-braid to the GND side of a nearby decoupling cap. Anyway, I looked at the same signal with an analog 100MHz scope and it looked *really* smooth. Whatever it is it's high frequency stuff since only the 1GHz scope could see it. This still bugs me though and I might go play with it some more. Maybe more/better supply decoupling at the DAC?

Also, the DAC datasheet says that with the SYNC pin high, another current source gets switched onto the green output (in addition to the current from the DAC) and adds about 8ma. Since the green output is tied to GND through 37.5ohms (75 per cable end) that 8ma is definitely getting out of the DAC and into GND.


Stacey,

Apex and Cinematrix bump up their output levels? Hmm, is that like the Circuit City guy cranking up the contrast on the show-room TV's to make them look better? Your mentioning that some devices don't like an input over .7V is advice well taken. (As all the advice you give on this forum is!!) I know my ECP3101 can take it, but apparently my Extron cannot. Maybe this should fall under the "mod that has good intentions but shouldn't really be tried" category.

Also, I looked at Avia's white bars and they're fine. I was a little suprised that the DAC held in there so well! I haven't looked at the interlaced output (62mv is almost just noise!) but I may take the lid off again to look at this noisy stuff and I could look then.


Bosef, thanks! Thank *you* for being the first one to report this level shift on green. Since it's a constant shift, it should be ok, but is it possible that digital projectors have a problem with this because they don't AC couple their inputs? Digital, baaad....


Cary

 

[Bosef]

Quote:

Originally posted by jcmccorm Bosef, thanks! Thank *you* for being the first one to report this level shift on green. Since it's a constant shift, it should be ok, but is it possible that digital projectors have a problem with this because they don't AC couple their inputs? Digital, baaad....

You're welcome. I'm glad i can be usefull, i had such good advise on this forum, that i really need to give something in change.


This is right the point.

My friend who first reported the thing, said there wasn't any problem with CRT projectors, but this green offeset would be ugly for the DC coupled sets, digital sets indeed!

I still don't know if we will change the resistor... i don't have splitter or video buffer before my Pj. I also hadn't the contrast control set to high so i think maybe i don't need to perform this mod. What would you do?


I *forwarded* the translation of your post on it.hobby.home-cinema, there are a lot of Skyworth's owner(some of them are using it with DLP)... they asked me about pictures. Do you have any, just to show them the mod?


Thank you again Cary!


P.S.: I see the rainbow effect on DLP's with 6 color wheels... i let you guess what's my opinion about digital proiectors...

 

[jcmccorm]

Hi Bosef! I could've kicked myself for not having the digital camera with me when I did those mods. I'd like to poke around the Skyworth a little more (to look into all that noise) and I'll take pictures this time.


In the mean time, the mod for Green is pretty easy to describe. You have to lift pin 12 of the DAC (marked ADV7123 on the Sage board). Luckily, pin 12 is on the corner. Pin 1 is marked on the PC board with a "1" and pin 12 is the last pin on that same side of the IC as pin 1. I just heated it up and lifted it with solder-scribe (any really sharp object that you can get underneath the pin to lift it up when the solder melts will do it). Don't lift it too high because you don't want to break it off.


Next I took my ohmeter and connected one end to a mounting lug on the VGA connector (GND) and searched for the nearest capacitir nearest the lifted pin. I found one end of a capacitor that was connected to GND (ohmeter showed 0ohms) and ran (soldered) a wire from the lifted pin to this end of the cap.


A picture is worth a thousand words though and I'll definitely get a couple.


As for the resistor change, if you're happy with your contrast and your not trying to match the Skyworth's output to some other source that is brighter, I'd leave it alone. As Stacey pointed out, some stuff can't take it.


Cary

 

[budiman]

jcmccorm:

As I understand it, per NTSC spec, a 100 IRE bar needs to be output at 710 mV. Most of the time, a lot of components are not very up to par, as such a 100 IRE pattern is output at less than 710 mV.


Boosting the Skyworth outputs by itself (and gaining better S/N ratio) is not wrong, however it will screw up your display settings compared to other sources. Most people will have the Skyworth hooked up and calibrate their display to the Skyworth's output. Since the white will be output at 1 V, where all other sources are outputting at 700-ish mV, all other sources will appear darker.


Just a thought.

 

[jcmccorm]

Hi Budiman, yes, I *do* have another source - an iScan Plus I use for VCR and satellite sources and it definitely appeared darker. I just figured that I'd turn the contrast back up when I switched to those. Since I backed off on the 1Vpp and went back to 0.7V, I'm ok.


Cary

 

[Bosef]

Quote:

Originally posted by jcmccorm In the mean time, the mod for Green is pretty easy to describe... A picture is worth a thousand words though and I'll definitely get a couple.

Hi Cary. All right it doesn't seem too difficult and i think that my friend will perform this mod without particular problems. (i'm not very skilled with soldering stuff).

Whenever you got pictures of it it could be nice to have the link for them anyway.

Quote:

As for the resistor change, if you're happy with your contrast and your not trying to match the Skyworth's output to some other source that is brighter, I'd leave it alone. As Stacey pointed out, some stuff can't take it.

You're right and that's exactly what i tought.

I won't change any resistor.