CurtPalme.com Home Theater Forum

[Nashou66]

Mike I think we need a High BW SPDT switch to put the signal from the RGB op amp to either one of two outputs. One to the SD5401 for OSD and the other to bypass the SD5401. The select high would put the signal into the SD5401 for the OSD and select Low would switch it to bypass the sd5401. I can not find any single High BW switches in a single in and two out but there are plenty of quad chips.

There is one in a QFN16L case it is very small and has pads vs pin we are used to. But I think we can use it either for all three channels on the single chip or use three chips and only one section for the switching. This might eliminate any cross talk that is most likely the case in the SD 5401( between internal and external video)

http://www.st.com/st-web-ui/static/active/en/resource/technical/document/datasheet/CD00148730.pdf


This has a 800mhz band width, some Ti chips i looked at were only 300mhz.

We can use D4 S4 and 4SEL and 4D to 4S1 for OSD and 4D to 4S2 for ext video. 4S1 is select High and 4S2 is select Low.

or use SEL1,2,3 for all colors on one chip instead of three.


we can use the SEL signals that are used by the SD5401 to activate the switches in the new chip unless they turn on and off quickly. I think the one to add INT video does . Or we can find two new select signals to accomplish this. Not sure what voltage is needed for the low state and if it need to be turned off for the OSD to over ride it.

Not well versed in this stuff but you or maybe Jorge would be able to get it to work.

Athanasios

[Nashou66]

This may also work. Use D as input and SA and SB as outs with a voltage to EN constant to keep chip on and the the SEL for
IN to select the Switch to be used.

Its a 520mhz chip.

http://www.analog.com/static/imported-files/data_sheets/ADG1219.pdf

Athanasios

[barclay66]

Hi,

I think I will be looking into this soon. The SD5401 seem to have at least two main disadvantages: They cut off some of the bandwith of a signal transmitted to them and they reduce the signal's amplitude too (no wonder with an ON-resistance of about 60 Ohms). On the other hand they offer two advantages: Very high switching velocity and no need for a power supply (because there are only four matched MOSFETs in that chip).

What we need is a new method/part which is able to combine the advantages while avoiding the problems. I think that the main focus should be placed on minimizing the interference with the transmitted signal.
Using relays would be ideal with this regard, but they are much too slow. The device needs to be fast enough for switching between external video (the picture signal) and internal video (the menus) during an active scan line. We're talking about micro- and nanoseconds here which is far beyond any mechanical system.

That won't be an easy one. I'm currently thinking about a replacement circuit which will work differently: Instead of either closing or opening a signal path it should be possible to selectively shorting the signals to GND. Imagine something like this: We add a series resistor (something between 10-50 ohm) to the arriving signal and after the resistor we have a small MOSFET which -when turned on- will tie the signal to GND. The MOSFET would need to have a very low ON-resistance (0.1 ohm or less), it would need to be very fast (OFF/ON and ON/OFF switch time below 5ns) and the Drain-Source capacitance should be as low as possible (otherwise it would cut down high frequencies). If there's no MOSFET suitable for this application I'll have to look for other alternatives.

Regards,
barclay66

[Nashou66]

I don't know Jorge, i think we need to keep the sd5401 there just by pass it when not needing the OSD. A relay would also work as well
or a mosfet. But I am not sure of the VSEL and VSEL* signals if they are also turning on and off durring the blanking portion of the scan or just when a menu item is selected . reason is i'd want to use them for the sel of the new CMOS device.

Athanasios

[mp20748]

My "mini board" mod replaces the video section (two of the four switches) of each of the SD5401. The switch for the restoration sections stay with the SD5401.

I'm using three switches very similar to the ADG1219. They are mounted on a little PC board, that I call my "mini board mod". And it works just like the the SD5401, but without bandwidth or gain limitations.



see in picture below:

[CIR Engineering]

[CIR Engineering]

Also Mike, if you look at the low resolution part of the pattern in you red, the red is ringing like crazy. Maybe the peaking is wrong, you have an impedance mismatch, or other issue with the video chain. It's possible that the ringing is covering all the detail in the 1:1 pattern making the resolving power invisible.

Aside from making halos, ringing covers any edge detail. And at 1:1 you need edge detail.

craigr

[mp20748]

The 1:1 is perfect on the screen, the added intensity in the red causes it to look like that. Sometimes it's not easy to capture the 1:1 with a camera. And simply because the two top line groups are matching, with the one on the left not being dimmer, it shows 234mhz being resolved.

I'll do it again later, but this time I would have to decrease the contrast of the RED.

Look at the larger lines in the bottom line groups and you'll be able to see the bleeding even in the larger lines.

[Nashou66]

Mike where is the syn signal from the internal patterns? I was thinking we could possibly divert the internal patterns to the input we are not using and switching to that input for the menu. But it not affect the video input we are watching. forget it

Nashou

[mp20748]

what do you mean by "syn" ?

[Nashou66]

oops sync.

Nashou

[mp20748]

I've not looked at where they switch the sync. So I'm not sure, but glad you brought this up, because i might want to keep the jumpers in place and impose the "internal" on the signal rather than switch. That would make it a bit cumbersome, but I would not have a problem switching away from the source to use the internal.

[Nashou66]

[mp20748]

Here you go Craig. The next shot is a better shot as promised. The RED is hard for the camera to capture.



1920X1440 /60hz - 234mhz -- Chroma Generator into RGBHV on 02 VIM





.
Now we're going to go a lot further:


On this second shot, I'm showing a very slight roll-off on the top left pattern group, which proves that I was right when I said my neck boards would do about 250mhz.. I know you're going to find this hard to believe, but that's near twice the actual bandwidth performance (not specs) of both the G90 and the Barco 909.

And you're wondering why this bandwidth is even achieved. This is a special version of the mods that's targeted for another market I have in mind for it, not Home Theater related. And even the regular version of the latest mods will do up to 200mhz. And the question is why would there be a need for the 200hz, when full 1080P is around 150mhz. Because by making a higher bandwidth video chain, it will produce a flatter bandwidth rating. So at 200mhz, there will still be 150mhz of bandwidth headroom, meaning also a truer and flatter bandwidth rating (not -3db). And when ever you can provide a truer than the -3db bandwidth requirement at full 1080P, you then get into experiencing what 2D really looks like, because you're also producing a very sharp image at the higher resolution.

It takes a lot of power to produce a true 2D (depth) High Def image..
.

2048X1536 /60hz - 265.25mhz -- Chroma Generator into RGBHV on 02 VIM

[jbmeyer13]

Darth to Luke: "Impressive...most impressive"

From experience, it's even sharper in person. Resolve the CLM line issue and it's hard to imagine pushing the envelope any further.

[Nashou66]

Mike is that with the jumper?

Nashou

[stridsvognen]

WOW.. i cant test anything over 200 mhz, the 03VIM i have here looks almost perfect when i go max on the Radiance,

I think its hard to understand the impact the higher bandwidth have on the image for those that never seen it.

But it makes all standard CRT machines look flat soft and boring.

And its even very visible on a machine where the magnetic, and focus is not 100% perfect.

If the ringing and the CLM noise isue gets solved, the Marquee will be perfect, and make everything else look like a joke..

Keep it coming Mike..!!

[mp20748]

[mp20748]

[stridsvognen]

Mike.. You just made 1080P 96hz posible on the marquee. Full blown 16:9 with no cropping .. . If there is a VP that will do it.. Maybe the new Radiance 4K machines.

And close to make full 4K 60hz blend possible